Microsoft at 50: Pivoting to AI and Cloud While Reflecting on Its Legacy

How Microsoft made it through 50 years

Tom Warren is a senior editor and author of Notepad, who has been covering all things Microsoft, PC, and tech for over 20 years. In 2005, Microsoft's leaders were starting to get worried. Windows and Office were doing well, but the company's lead software architect, Ray Ozzie, warned them that an emerging trend threatened Microsoft's very existence. "A new business model has emerged in the form of advertising-supported services and software," warned Ozzie in his famous memo. "This model has the potential to fundamentally impact how we and other developers build, deliver, and monetize innovations." Ozzie wanted Microsoft to get ready for the web-based world and cloud computing, fearing the company would otherwise be left behind. In the years after Ozzie's memo, Google started to build an online competitor to Office with its Google Docs web service that it was selling to businesses and offering free to consumers. But the idea of moving Office to the cloud remained controversial internally. Doing so would upend Microsoft's traditional method of selling software -- and potentially eat into short-term profits in a big way. "It was a gutsy call, it wasn't a popular call at Microsoft at that time," says Rajesh Jha, executive vice president of Microsoft's experiences and devices group. Steve Ballmer, then Microsoft's CEO, decided to go ahead with it, moving Office to the cloud and launching Azure fully in 2010. Office 365 launched initially in beta, and Microsoft faced challenges transforming Office for the web. "The early days of Office 365 were rough," admits Jha. Microsoft had to perform the painstaking transition of moving apps that were built for a different architecture and era to the cloud and make that scale properly. "I remember Steve [Ballmer] talking to me and saying 'I've got your back, go fix this stuff." It was the type of change that Microsoft has navigated time and time again over its 50-year history, an epoch marked by success, pivots, and failures. It's easy to think Microsoft would be embarrassed by the launch of Windows 8, the performance of the Zune, and its failure in smartphones -- surely, the company wishes those things had done better. But earnest attempts at something new are the whole point. "We're still around and leading things because we accept change," says Steven Bathiche, head of Microsoft's applied sciences group, in an interview with The Verge. "That's the pearl about Microsoft that's kind of unique with a company as old as ours, in an industry that changes as rapidly as ours." Microsoft turns 50 years old today. The company was founded in 1975 by Bill Gates and Paul Allen as Micro-Soft. A lot has changed since then, but in many ways, the company's core business model and ethos remain the same: make software that everyone needs and get it installed everywhere. Gates and Allen began that strategy with Altair BASIC, a BASIC interpreter for the Altair 8800 -- a computer made by MITS, rather than one they made themselves. After making an interpreter for one computer, they started bringing it to every other PC maker that would take them. Following the success of Microsoft BASIC, the company would repeat the strategy with MS-DOS, and eventually Windows and Office. But building those must-use platforms has taken a lot of experimentation, and Microsoft has often had the right idea at the wrong time -- and missed out on major markets as a result. Microsoft had SPOT smartwatches decades before the Apple Watch and Windows XP-powered tablet PCs years before the iPad. One of the company's biggest misses was mobile. Microsoft has experimented with so much hardware and software over the years that it should have been in the ideal position to be the primary alternative to the iPhone. Instead, its platform burned out, and the company wasted billions on the acquisition of Nokia in a fruitless attempt to catch up. Still, Microsoft insiders say it's these experiments that have teed up many of the company's eventual successes. "We used to do these invention sessions with Bill [Gates] which was a lot of fun," says Bathiche. "I was pretty young when Bill was around so I'd come in as this mad scientist and show him all this cool stuff we were doing in hardware and software, and I think he really appreciated that." Some of those experiments would unlock touch computing and cloud advances, helping pave the way for Microsoft's own laptops and tablets. Other hardware and software experiments led to the Xbox, Microsoft's best example of combining software and hardware into a successful consumer product. At its heart, Xbox is about enabling developers to build compelling games with software, and it was originally dubbed a "Windows entertainment project" and developed by members of Microsoft's DirectX team to fend off the threat of Sony's PS2 console in the living room. Other Microsoft projects have taken a more roundabout route to becoming successful products. One major case that Bathiche mentions is Cortana, the voice assistant first designed for its phone platform. The assistant didn't catch on as an alternative to Siri and Alexa and was eventually shut down. "The Cortana days, that was really difficult," admits Bathiche. The project required Microsoft to develop technologies around local compute and to optimize for Arm chips. "So that project, even though I kind of failed on enabling that sort of solution, I took all those technologies that we developed as a team and applied them to Windows and Surface." Those technologies, from optimizing Cortana to getting Windows Phone to run on Arm chips, helped with Microsoft's work with the Surface Pro X and the early AI-powered experiences in Windows. "That's going to set us up for the next decade," says Bathiche. That history speaks to why Microsoft CEO Satya Nadella has been so aggressive when it comes to AI. One of Microsoft's most significant deals under Nadella has been its partnership with OpenAI. Microsoft originally invested $1 billion into OpenAI in 2019 to become the exclusive provider of cloud computing services to the AI research lab. It then extended that partnership in 2023 with an additional $10 billion investment, just months after ChatGPT launched and became the fastest-growing app in history. This investment has fueled Microsoft's ongoing AI transformation, amid fears that Google was years ahead of it in scaling up AI efforts. For Bathiche, this new AI era illustrates patterns that allow companies like Microsoft to take advantage of the constant change in compute ability and build new experiences or new device form factors. "When we had the first computers they would load a very specific program and that's the only thing they could do," says Bathiche. Microsoft saw the opportunity early on to build Windows as a way for developers to talk to PC hardware easily and create apps with a GUI on top. Then the internet was a layer on top of that, which changed how software could be delivered. "Now we're on the next layer, which is the agent layer," says Bathiche, describing intelligent software that acts autonomously on behalf of a user. "That's the exciting one we're trying to figure out." Microsoft has always developed software by compiling an app like Word, and then everyone gets that same piece of software, whether it was delivered via a CD-ROM or over the internet. Microsoft is now trying to build software that adapts on the fly. "The race is on to compile that piece of code uniquely, individually, for every person on the planet. The reason is that these AI agent experiences we're talking about are really about deep personalization and understanding and customization," says Bathiche. This push toward unique and personalized software through AI has driven compute needs like never before, and Microsoft's Azure AI teams are constantly having to adapt to new models or new hardware that unlocks even more AI capabilities. We saw that recently with Microsoft's quick adoption of DeepSeek's R1 model into its Azure AI Foundry. "All of the systems that we've built for 50 years need to apply to AI agents," says Asha Sharma, corporate vice president of Microsoft's AI platform. "For Azure AI Foundry we're thinking about how we evolve to become the operating system on the backend of every single agent." Microsoft envisions a world where AI agents will work alongside you as digital colleagues and collaborators. "I think a lot of the systems we've built for 50 years will be critical infrastructure for the digital workforce," says Sharma. "We need to think about having not just human understandable interfaces, but machine understandable interfaces." Part of winning this AI battle will be down to Microsoft's own models, and using competitor models wisely. Microsoft unveiled its own Muse AI model recently, which has been trained on the Xbox game Bleeding Edge to generate gameplay. Nadella stressed the importance of Microsoft's own AI work during a recent employee town hall meeting, saying, "When I think about what we're trying to do with Muse, that's the bar. You go do fundamental research, you bring out a model, and then translate it into a real breakthrough feature in Copilot." Beyond digital colleagues and Xbox changes, Microsoft is also hinting at fundamental ways AI could live in the physical world. "I believe that some of our AI agents in the future will be embodied agents that are able to make plans and take actions on behalf of users in physical environments," says Ashley Llorens, corporate vice president and managing director of Microsoft Research. "Not just the same very carefully engineered environments that we have many robots in today, but in many of the semi-structured environments that people live and work in. One of the challenges there is teaching AI how to reason over the relationships between objects and environments in time and space." AI-powered robots might sound like science fiction right now, but Microsoft is increasingly focused on robotics, healthcare, and even quantum computing, so anything seems possible for its next 50 years. Since Microsoft's mobile miss, I've always gotten the feeling that the company is wary about missing the next big thing, so I'm fully expecting it to pivot if the AI era slows down. Microsoft's resilience and its ability to predict what's next -- like Gates in the early days of Microsoft or Ozzie in 2005 -- have defined the company's existence over the past 50 years. Moments of potential demise have forced Microsoft to make big bets that weren't always successful but did lead to change. Now, Microsoft has gone all in on AI. Its efforts might even contribute to the "singularity," a point in time when artificial intelligence exceeds the abilities of humans and, perhaps, even our control. What that will look like is anyone's guess. But if the AI hype bubble bursts and it's more a Windows Mobile flop than an Azure hit, I'm sure of one thing: Microsoft will adapt to meet the challenge.

Think DeepSeek has cut AI spending? Think again

The stock market collapse in January, prompted by the sudden fervor for the Chinese artificial intelligence (AI) breakthrough DeepSeek AI, and its apparently much cheaper computing approach, might make you think that companies are dramatically reducing their spending on chips and systems for AI. A generative AI conference on Wednesday in New York, hosted by the Bloomberg Intelligence unit of Bloomberg, left me with the opposite impression: the hunger to expand the use of generative AI is so great that spending is set to continue to drive enormous investment. Also: What is DeepSeek AI? Is it safe? Here's everything you need to know Titled "Generative AI: Scaling Laws Post DeepSeek," the daylong event featured constant references to how demand will drive greater spending. "We had ten panels today, and not a single person on those panels said we have more capacity than we need," said Mandeep Singh, a senior technology analyst with Bloomberg Intelligence, and an organizer of the event. "And no one was talking about a bubble" in infrastructure, added Singh. "The most important question right now in front of everybody is the AI infrastructure build. Yeah. Where are we in that cycle?" Anurag Rana, Singh's colleague at Bloomberg Intelligence, and the senior IT services and software analyst, told me. "Nobody knows" for certain, said Rana. However, the hope raised by DeepSeek AI is that a lot of work can be done with less expense. "DeepSeek shook a lot of people," he said. "If you are not needing that many GPUs to run models, then why do we need $500 billion for the Stargate project," he observed, referring to a planned US AI project run by Japan's SoftBank Group, OpenAI, and database giant Oracle. Everyone in the industry, said Rana, hopes that AI costs will come down rapidly just like the cost of cloud computing dropped. Also: Is DeepSeek's new image model another win for cheaper AI? "That drop in the cost curve, which probably took six, seven, eight years to store one terabyte of data in Amazon AWS, when it started versus today, the economics were good," he said. "And that's what everybody's hoping, that on the inference side" of AI, "if the curve falls to that level, oh my god, the adoption rate on AI on the end-user side of it, or, the enterprise side of it, is going to be spectacular." Singh agreed that DeepSeek AI's moment has "changed everyone's mindset about achieving efficiency." That was certainly my impression of the day's proceedings. Numerous panels focused on enterprise AI projects, from the conceptual stage to finally putting them into production. However, panelists also talked about the need to bring the costs of serving AI down dramatically in order to expand access. "I don't think DeepSeek was a surprise," said Shawn Edwards, Bloomberg's chief technologist, in an interview with David Dwyer, the head of Bloomberg Intelligence. "What it made me think is that it would be great if you could wave a wand and have these models run incredibly efficiently," he said, meaning, all AI models. One of the reasons many panelists expect more AI infrastructure investment, not less, is because of the proliferation of AI models. A theme throughout the day was that there won't be one AI model for all tasks. "We use a family of models," said Edwards. "There is no such thing as a best model." Panelists agreed that while "foundation" or "frontier" large language models will continue to be developed, individual enterprises may use hundreds or even thousands of AI models. Also: The rise of AI PCs: How businesses are reshaping their tech to keep up These models might be trained on a company's proprietary data via fine-tuning, the act of re-training a neural network after its initial "pre-training" on generic data. "Agents in enterprise require optionality among the models," said Jed Dougherty, the head of platform strategy for venture-backed data science firm Dataiku. "They need the ability to control and create, and to have auditability" of AI models. "We want to put the tools to build these things in the hands of people," he said. "We don't want ten PhDs building all the agents." In a similar spirit, design tools maker Adobe is betting that custom models will be a key use case in the new age of creatives. "We can train custom model extensions for your brand that can be a help for a new ad campaign," said Adobe's head of new business ventures, Hannah Elsakr, in a chat with Bloomberg TV anchor Romaine Bostick. As with AI models, the proliferation of AI agents inside companies is increasing the processing demand, many speakers suggested. "You won't cram a whole process into one agent, you'll break it up into parts," said Ray Smith, Microsoft's head of Copilot Studio agents and automation. Through a single interface, predicted Smith, such as Copilot, "we will interact with hundreds of agents -- they are just apps in the new world" of programming. "We will give the agent the business process, tell it what we want to accomplish," and the agent will carry out tasks. "Agentic apps are just a new way of workflow," he said. Also: Nvidia dominates in gen AI benchmarks, clobbering 2 rival AI chips Such everyday scenarios are "all technologically possible," said Smith, "it's just the pace at which we build it out." The trend to get AI "agents" to more people in an organization is further demanding a cost reduction, said James McNiven, the head of product management for microprocessor maker ARM Holdings, in a chat with Bloomberg's Hyde. "How do we provide access on more and more devices," he posed. "We are seeing models at a PhD-level" of task capability, he said. Such agents need to be an assistant to humans, he suggested, making the comparison to when payment systems were brought to developing countries a decade ago via mobile phones: "How do we get that to people who can use that ability?" Even the generic foundation models are proliferating at an amazing rate. Amazon AWS has 1,800 different AI models available, Dave Brown, the head of AWS computing and networking, told Bloomberg TV anchor Caroline Hyde. The company is "doing a lot to bring down the cost" of running the models, he said, including developing custom AI chips, such as Trainium. AWS is "using more of our own processors than other companies' processors," said Brown, alluding to Nvidia, AMD, Intel, and other general-purpose chip suppliers. Also: ChatGPT's new image generator shattered my expectations - and now it's free to try "Customers would do more if the cost was lower," said Brown. AWS works daily with Anthropic, makers of the Claude language model family, noted Brown. Anthropic head of application programming interfaces Michael Gerstenhaber, in the same chat with Hyde, noted that "Thinking models cause a lot of capacity to be used," referring to the trend of so-called reasoning models, such as DeepSeek R1 and GPT-o1, to output verbose statements about the arguments for their final answers. Anthropic is working closely with AWS on ways to trim the compute budget, such as "prompt caching," storing the computations from prior answers. Despite that trend, he said, "Anthropic needs hundreds of thousands of accelerators," meaning, AI-focused silicon chips, "across many data centers" to run its models. In addition, the spiraling energy cost of powering AI shows no sign of slowing, said Brown. Current data centers are consuming hundreds of megawatts, he noted, and will eventually require gigawatts. "The power it consumes," meaning AI, "is large, and the footprint is large in many data centers." Also: Global AI computing will use 'multiple NYCs' worth of power by 2026, says founder Despite the ambitious scenarios, one condition could upend all the use cases and investment plans: the economy. As the conference was drawing to a close Wednesday evening, panelists and guests were checking the after-hours plunge in the stock market. US President Donald Trump, in a Rose Garden ceremony, had just unveiled a global package of tariffs that were larger and far more sweeping than most on Wall Street had anticipated. Traditional areas of tech investment, such as servers and storage, and not AI, could be the initial victims of any economic contraction, said Bloomberg's Rana. "The other big thing we are focused on is the non-AI tech spending," he said regarding the tariffs. "When we look at the likes of IBM, Accenture, Microsoft, and all the others, when we just put aside AI for a second, that is something that is going to be a struggle going into this earnings season." CFOs of major companies could prioritize AI, and shift funds, even if they have to trim their budgets amidst economic uncertainty and recession, should it come to pass, he told me. That rosy view is by no means guaranteed, however. "The thing I'm most interested in finding out, is, if all these large companies are going to keep their cap-ex [capital spending] targets intact," said Rana, including AI data centers, "or are they going to say, You know what? It's too uncertain."

The rise of AI PCs: How businesses are reshaping their tech to keep up

The AI race has led to a hardware revolution, with businesses investing in new types of PCs. The artificial intelligence (AI) boom is transforming industries and reshaping work. Now, it's setting its sights on a long-overlooked tool: your work laptop. The reach and capabilities of AI models go far beyond what people have seen from tinkering around with ChatGPT, which, although a great tool for proofreading or debugging code, only gives a brief glimpse into what large language models (LLMs), the technology powering tools like the chatbot, can do. Also: What is an AI PC exactly? And should you buy one in 2025? For instance, HCLTech, a consulting firm, worked with one of the largest end-to-end healthcare providers in the US to help implement a user-friendly, compliant AI clinical advisor. The clinical advisor, trained using one of the world's largest clinical libraries, allows medical professionals to conversationally ask for the information they need to consult without wasting time digging for it. According to Alan Flower, EVP and head of AI & Cloud Native Labs at HCLTech, running the models locally on the device was key to effectively implementing this model. Also: Nvidia dominates in gen AI benchmarks, clobbering 2 rival AI chips "By using the clinician's laptop, it frees up clinician time whilst also ensuring the privacy and security of the patient's information," said Flower. "It has that secondary advantage -- it reduces the cost to run the solution since we're not consuming expensive cloud tokens when we can use the neural processing unit on an AI PC, effectively, for free." Flower said that since implementing the AI clinical advisor, within a typical 20-minute consultation with a doctor, the AI clinical advisor has freed up three minutes from the appointment, time the doctor can spend researching and focusing on the patient's care. Also: With AI models clobbering every benchmark, it's time for human evaluation Companies are discovering that if they want to take full advantage of AI and run models locally, they need to upgrade their employees' laptops. This realization has introduced a hardware revolution, with the desire to update tech shifting from an afterthought to a priority and attracting significant investment from companies. This trend is reflected in the numbers. Researcher IDC predicts that AI PCs designed to run generative AI tasks locally will grow from nearly 50 million units in 2024 to over 167 million by 2027, eventually accounting for almost 60% of all PC shipments worldwide. The latest data from analyst firm Gartner suggests 63% of all PCs will be AI-enabled by the end of 2026. This acceleration in PC investment aligns with a decade-long trend, where factors, such as improved design, exemplified by the shift from bulky laptops to ultraslim notebooks, have steadily driven up the average selling price of PCs, an indicator of shifting consumer preferences and market trends, according to Linn Huang, research vice president of devices and displays at IDC. The industry has seen selling prices "steadily climb over the past decade," said Huang. "AI PCs will undoubtedly accelerate this trend in the coming years as new AI-enabled workloads, use cases, and features enter the market." This new category of AI PCs is differentiated from legacy PCs by having an NPU, or neural processing unit, which is designed to turbocharge AI tasks and allow models to run right on the device. The benefits of running locally include cost, privacy, and security. Also: ChatGPT's new image generator shattered my expectations - and now it's free to try Most AI products, like ChatGPT, send information to the cloud to be processed. This information is then sent back to the device, which allows for the potential for the data to be hacked, leaked, or mishandled in transit. Instead, running models locally gives organizations more control over their information and reduces reliance on third-party services. That setup is crucial for companies in financial services, healthcare, and other industries where privacy is a big concern or a regulatory requirement. "For them, on-device AI computer, it's not a nice to have; it's a need to have for fiduciary and HIPAA reasons, respectively," said Mike Bechtel, managing director and the chief futurist at Deloitte Consulting LLP. Another advantage is that local running reduces lag and creates a smoother user experience, which is especially valuable for optimizing business applications. That speed is key because the appeal of AI solutions, like chatbots, is that they can respond at the touch of a button. Device manufacturers are already seeing this growing interest in AI PCs. Microsoft, a major player in the AI market, a key partner of model specialist OpenAI, and the creator of AI tools like Copilot, has introduced a new hardware category called Copilot+PCs, developed with partners such as HP, Dell, and Lenovo. Also: Anthropic launches Claude for Education, an AI to help students think critically Mark Linton, vice president of device partner sales at Microsoft, said business leaders have shown strong interest in purchasing and trialing these new devices. "Leaders are particularly focused on integrating on-device AI capabilities into their broader AI business and technology strategies," said Linton. "This interest is reflected in adoption rates -- 20 times faster than touch PC adoption -- demonstrating the urgency with which organizations are approaching this transformation." The benefits of AI PCs go beyond running AI models on devices. For example, a PC can overheat when it uses a power-intensive application, and the fan may start blowing and the device might begin running slowly. That situation can occur partially because the excessive workload is offloaded to the CPU or GPU, other key components of PCs. AI PCs use NPUs to handle AI tasks more efficiently, reducing the load on the CPU and improving overall system performance. This capability covers non-AI tasks that are compute-intensive, such as editing video or even having a video call. Also: What is Enterprise General Intelligence? How the next stage of AI affects you "AI PCs deliver significant performance improvements for the apps users already use, leveraging new CPU, GPU, and NPU architectures," said Alex Thatcher, senior director of AI experiences, business personal systems, at HP Inc. "AI PCs also deliver significant improvements in battery life due to the inclusion of small cores and highly power efficient NPUs." As the refresh cycle for many businesses' hardware approaches, and its advance is accelerated by other external factors, such as Microsoft's end of support for Windows 10, many organizations are forced to purchase new devices. This leaves one question: should businesses go for legacy or AI-capable devices? Because of the productivity advantages, most industry experts, including Flower or Bechtel, who are in positions to advise clients on the next steps, agree that upgrading to AI-capable PCs is the obvious choice. "It's a big brainer: make sure that your next wave of capital expenditure has the set of bells or whistles, else you're going to be considering an extra refresh in the next couple of years," said Bechtel. Also: Microsoft's new AI agents aim to help security pros combat the latest threats As more companies get in on the action and AI-capable computers become ubiquitous, the premium price of AI PCs will continue to drop. Furthermore, Flower said the potential gains in performance offset any price differences. "In those high-value professions, the productivity gain is so significant that whatever small premium you're paying for that AI-enhanced device, the payback will be nearly immediate," said Flower. Upgrading technology keeps companies at the forefront of innovation, giving them the flexibility to invest in AI applications, whether they're ready to dive in now or not. As more companies opt for AI-capable PCs, this shift paves the way for deeper AI integration into business processes. The result of this investment is an ecosystem where the promise of AI applications drives hardware upgrades, and the upgraded hardware, in turn, accelerates AI adoption in the workforce, rapidly transforming industries and reshaping how professionals work.

Reflections on Microsoft's 50th: What Will It Take to Make It to 2075?

As I sat at Microsoft's 50th anniversary event on Friday, watching all three of the company's CEOs -- Bill Gates, Steve Ballmer, and Satya Nadella -- on stage, I couldn't help but think about the company's past and its future. Nadella has focused the company on his AI Copilot strategy with a number of new features announced for the consumer versions. These changing aims can also be seen in the current roadmaps. But this is far from the first time the company has changed, or at least broadened its focus. Always Pushing the Limit At the event, Gates reflected on starting the company in 1975 in Albuquerque, New Mexico, when there was really no software industry. (He has often spoken of the impetus for starting the company: Paul Allen picked up a copy of Ziff-Davis' Popular Electronics with an Altair 8800 on the cover and brought it to his dorm room.) At that point, Gates said, he and Allen were surprised that no one else understood a key fact about personal computers. The exponential improvement in chips would mean that the hardware and the computation would be basically free, but you would need software to use the computer in a helpful way. So they started a company to make that lucrative software. From the beginning, Gates said, he knew Microsoft would not be a single-product business, as most of his competitors were. Instead, he said, "We were building a software factory, hiring the best, creating better tools, always pushing the limit." Microsoft's first product was a BASIC language interpreter for the Altair, and soon the company began making BASIC software for nearly every computer in the nascent personal computing market. I first encountered Microsoft through its BASIC for the TRS-80 Model 1. By the time I first met Gates and visited Microsoft, the company had moved to Bellevue, Washington, (where it was still small enough to fit in a single building), and was selling a number of different programming languages. Most notable among them was DOS, originally produced for the IBM PC and later running on lots of machines. Despite having a big vision of "a PC on every desk, and in every home, running Microsoft software," Gates told the crowd at the anniversary celebration that in the early days of the company, he and Ballmer were always the ones who were a little bit conservative about the forecast. Even though they wanted to create lots of different kinds of software, they took "a 'one step at a time' approach to it." 'There Was Chaos' As Ballmer, who was second-in-command to Gates throughout the period, remembered it, "In 1980, there was no industry, there was chaos. Everybody was small until the entry of IBM in 1981, and then things got big and more professional, and it really took the better part of the next 20 years for software and personal microprocessor-powered devices to really take off. It was pretty cool." A lot of momentous developments were in the works. I remember visiting Microsoft around that time and seeing an early version of software they then called Microsoft Interface Manager, later to be renamed Windows. (As far as I know, I'm the only person who wrote about it under the Interface Manager name.) The first version of Windows shipped in 1985 as an interface that ran on top of DOS, but it took until Windows 95 to really become the industry standard. Microsoft moved to Redmond in 1986, and at the time the campus seemed huge for a software company. (It's many times bigger now.) The next 10 years are probably best remembered for Windows and the complicated relationship -- and later competition -- with IBM. I had numerous conversations with Microsoft executives about both the products and the relationship with the hardware behemoth. Nadella teased Gates and Ballmer at the event about having to manage Jim Allchin (one of the lead developers of what would become Windows NT) and Brad Silverberg (who was in charge of Windows 95). I recall great product-focused conversations with both of them as well. Fierce Competition But I also remember it as the era when Microsoft's applications started to take off, with Microsoft Word competing with WordPerfect and Lotus's AmiPro; Excel competing with Lotus 1-2-3 and later Borland's Quattro; and PowerPoint competing with Harvard Graphics and later Lotus and WordPerfect programs. The competition at the time was fierce, and I fondly remember visiting each of these companies multiple times and seeing new versions. Microsoft's entries, of course, were later combined as Microsoft Office, which remains hugely popular to this day in its online version, Microsoft 365. At the anniversary event, Ballmer called out the employees for not only being "super smart" and innovative, but also competitive. "You obviously can't miss out on innovation, but it's a competitive game," he said. "It's like sports -- at the end of the day, somebody wins and somebody loses. At Microsoft, there are no losers here, baby!" That competitive edge was on full display in the 1990s. When the Web began to take off in the mid-1990s, Microsoft was originally seen as lagging behind Netscape Navigator. After it responded with new versions of Internet Explorer, Microsoft became the target of an antitrust lawsuit that seemed to consume the company. For a time, it seemed like Microsoft might get broken up, but it eventually settled. In 2000, Ballmer replaced Gates as CEO. A lot happened in that period, including the introduction of the Xbox and, later, Microsoft's failures with a portable music player (the Zune) and Phones. But for me, the 2000s were the decade when Microsoft's enterprise products -- such as Windows Server, SQL Server, Exchange, and the business versions of Office -- became standards. Nadella replaced Ballmer in 2014 after running Microsoft's Cloud and Enterprise division. I think of his era, at least until recently, as being characterized by a strong move to the cloud, notably with the Office 365 (now Microsoft 365) applications and the emergence of Azure as one of the leading cloud platforms. Along the way, Microsoft has released dozens of Azure platform services, data and analytics tools, and security tools, among others. 'On the Verge of Something Even More Profound' Now, of course, Copilot is front and center, and it was the focus of the event. (More on that in a later post.) Gates said, "Now the frontier is intelligence, and...we're on the verge of something even more profound than what came for those first 50 years." For me, I still enjoy hearing about, and trying out, the latest in technology from Microsoft and its competitors. I'm excited by the possibilities of the new AI software, though I'm a bit skeptical about a lot of the claims and a bit worried about the potential downsides. I like much of the thinking behind Copilot and its competitors, but to me, most of the tools haven't quite lived up the hype yet. Still, it's early days. Looking forward, the three CEOs were asked what they hoped the company would achieve by its 100th anniversary. Gates joked, "I hope Copilot's the CEO." Ballmer pointed out that it's very unusual for companies to stay important and relevant for a century. If Microsoft is still alive and well then, Ballmer said, it would be a testament not only to him, Bill, and Satya, but to the next three CEOs as well. Nadella said that after 50 years, the company is still talking about the same issues as when it was first started, except with different technologies. "Tech has come and gone, but we're talking about the same thing, which is, how do you put software, how do you put intelligence, into everybody's hand so that they can do more?"

At 50, Microsoft Makes a Midlife Pivot

As Microsoft turns 50, you might think the megacorporation would unveil some major upgrades to its core software. You would be wrong. The venerable giant has published roadmaps of what it plans to do for the rest of the year, and very little of it has to do with the mainstays that have defined the company for years: Word, Excel, PowerPoint, and Outlook. At least not with their core functionality. Instead, the focus is on generative AI assistants: Copilots and agents, with Copilots designed to answer questions and help you perform a task and agents doing the tasks for you. Microsoft is developing a range of these helpers, aimed at many of the individual applications within 365. Expect specific Copilots for most of the applications within the Office suite and more. The company plans to beef up its AI support for enterprise applications, too, along with a bundle of different upgrades. Here are some of the changes Microsoft has in store for 2025. Microsoft 365: A Spread of AI Options The Microsoft 365 roadmap (as I write this) lists 682 features in development, with another 141 rolling out. That sounds like a lot, though a bunch of these features are pretty minor. And few new features are planned for the core productivity apps in the 365 suite, other than those related to Copilot and agents. The big add-on to the central products -- and the one that has gotten the most attention -- is Python support in Excel. It's already rolling out in some versions, and due for others this month. What is coming for 365 is a spread of AI options. The Microsoft 365 Copilot alone accounts for 151 new features, plus another 17 rolling out. Beyond Copilot, in fact, Microsoft hasn't talked much about new features for Windows itself. Right now, the Copilot has tools for things including creating presentations from PDF files in PowerPoint, taking emails and meetings into account when drafting content in Word, and creating banners for documents with Copilot and Designer. Later this year, Microsoft plans important changes, such as better ways of creating a presentation from a Word file and using Copilot to prioritize your inbox. Microsoft is also concentrating on newer applications. The Purview data-management and control system includes 191 new features, the largest of any category, while the Viva employee-experience software is getting 97. Microsoft is also continuing its push with Teams. The software has 79 new features in development, with another 59 rolling out. In the next few months, we should see better support for Windows devices in Teams rooms, the ability to share notes to Outlook, and live transcription, among other things. Microsoft Azure: Slow But Steady In many respects, the cloud-computing service Azure has replaced Windows as Microsoft's key infrastructure platform, but it only has 123 features in development, almost all of which have been listed for more than a year. Another 2,600 are still in preview (which means not completely finished but rolling out to some but not all user); some have been sitting there for many years. Azure, simply put, is getting mature, so there aren't a ton of improvements ready to roll. True, Microsoft does regularly announce new compute instances -- virtual machines -- and it continues to update others. But the basic functionality of things like computing and storage doesn't seem to be changing much. The features that are listed for Azure center on infrastructure-as-a-service (IaaS) and platform-as-a-service (PaaS). But they don't include newer cloud offerings like Fabric, Power BI, or Power Platform, which are covered in separate roadmaps. Almost 500 features for databases are listed as being in preview, including a number of enhancements to the Cosmos DB. And, of course, there's AI. Microsoft lists 140 of these upgrades on the way, but all are in preview. They cover everything from Copilot for Azure itself to Microsoft Copilot for Security. Microsoft Fabric and Power BI: Where Copilots Shine Copilots are a much stronger presence in the roadmap for Fabric and PowerBI, Microsoft's data and analytics platform and business intelligence tool. For instance, Fabric is getting a bunch of AI-driven features, such as a Copilot for Data Warehouse that allows users to query data using natural language. Copilots are also being developed for applications like Data Science and Real-Time Intelligence. In addition to Copilots, Fabric is due to receive other tools that make it easier to manage and analyze real-time data. The Copilot for Power BI, meanwhile, is due for new features such as the ability to automatically generate summaries of report subscriptions delivered via email. Also on the way: improvements to visualization tools and standardizing on some open-data formats such as Parquet. Upgrades for the data services will make it easier for Fabric Data Warehouse to ingest data into Data Warehouse, while OneLake will be able to more easily access data stored in other locations and formats. Dynamics 365 and Power Platform: Agents Take Center Stage For Microsoft's Dynamics 365, which covers enterprise resource planning (ERP) and customer relationship management (CRM), again AI Copilots and Agents take center stage. Earlier this month, Microsoft released two Copilots for Sales. First, there's Sales Agent, which gathers information from a swath of places -- CRM, company price sheets, the web, and even Microsoft 365 data like emails -- to research customers, set up meetings, and draft messages. Sales Chat gives sales representatives practical takeaways from CRM data, pitch decks, meetings, emails, and the web. Other AI assistants on the way will help out in many specialized areas like customer service and finance. Watch for features like improved capabilities for handling self-service customer calls and features that speed up purchase-order updates based on incoming emails. AI offerings are also in the offing for Power Platform (Copilots for Power Automate, and Power Apps) and Copilot Studio (the ability to use prebuilt agents as a starting point for Copilot creation, the ability to use Teams chats as knowledge sources, and new features for testing and debugging actions). Microsoft Windows: Are Copilot+ PCs Ready to Take Off? Last month, Microsoft released a roadmap for Windows itself, largely focused on features that are part of its Copilot+ PCs initiative. This includes features such as the long-promised Recall, which stores snapshots of everything you do on your hard drive so you can search it, and Click To-Do, which identifies text and images on the screen and suggests actions you can take. Both of these are now in Preview for some members of the Windows Insider program. Other things coming for Copilot + PCs include improved Windows Search. Most of the new features for older machines are relatively minor, although I'm looking forward to seeing simpler "top cards" designed to give a quick view of your PC's key specifications on the About page. The 50-Something Pivot As you can see, Microsoft's emphasis -- and indeed the emphasis of the whole software industry -- has changed over the years. In its earliest years, the company focused on software languages, then moved to focus on Windows and core office functions such as word processing and spreadsheets. More recently, the emphasis moved to developing and building out enterprise applications and its Azure cloud platform. On its 50th anniversary, it now seems to have shifted to AI and Copilots, which seem to be the future -- at least for now.

Microsoft's AI division head wants to create a lasting relationship between chatbots and their users

Fifty years after the founding of Microsoft, the CEO of its artificial intelligence division has a big task: develop a new product line as integral to daily life as the software giant's past innovations. "We're really trying to land this idea that everybody is going to have their own personalized AI companion," said Mustafa Suleyman in an interview with The Associated Press. "It will, over time, have its own name, its own style. It will adapt to you. It may also have its own visual appearance and expressions." Suleyman is laying out that vision on Microsoft's 50th anniversary Friday. The celebration at Microsoft's headquarters in Redmond, Washington will feature the first public gathering in more than a decade of co-founder and former CEO Bill Gates with his two successors: Steve Ballmer, who led the company from 2000 to 2014, and the current chief executive, Satya Nadella. Giving the mic to Suleyman, who joined the company and Nadella's senior leadership team just over a year ago to head a newly formed Microsoft AI division, signals how important getting its AI right is to the company's future -- in the next five years if not the next 50. The company's flagship product of this AI era, Copilot, already combines a chatbot with Microsoft's suite of workaday tools, from Excel spreadsheets and PowerPoint presentations to the Windows operating system that defines how most computers work. But Suleyman is striving for something that sounds a little more like science fiction -- a technology that can form a "lasting, meaningful relationship" with its users. "One that knows your name, gets to know you, has a memory of everything that you've shared with it and talked about and really comes to kind of live life alongside you," he said. "It's far more than just a piece of software or a tool. It is unlike anything we've really ever created." Some of those updates -- such as new "visual memory" capabilities that keep track of a chatbot user's digital activity, if they want that -- roll out on mobile apps Friday. Other features are still in development. Suleyman, 40, came to Microsoft last year with plenty of credentials in the AI business. In his 20s, the British entrepreneur co-founded the legendary DeepMind AI research lab in London, which Google later purchased in 2014. He worked there until 2022, when he left to set up the new company Inflection AI with LinkedIn co-founder Reid Hoffman. Microsoft then scooped up Suleyman and other Inflection leaders without formally acquiring the startup, attracting months of antitrust scrutiny. He also co-wrote a 2023 book, "The Coming Wave," focused on AI's promise and the need to limit its potential perils. But unlike DeepMind, or another famous AI research lab, San Francisco-based OpenAI, both of which have explicitly set out to build better-than-human AI known as artificial general intelligence, Suleyman is not too worried about such abstractions in his new job running Microsoft's consumer AI business. "My goal is really to create a true personal AI companion," he said. "And the definition of AGI sort of feels very far out to me and sort of not what I'm focused on in the next few years." The race to build the best AI personal assistant that sticks with consumers is a competitive one. In recent weeks, rivals including Google and Meta Platforms have shaken up the teams in charge of AI research or products. ChatGPT maker OpenAI, which is both Microsoft's most important AI partner and a growing competitor, has also reorganized its leadership. Amazon is also looking to catch up by imbuing its already-ubiquitous digital assistant Alexa with more advanced AI capabilities. "It's a super competitive market but this is absolutely foundational to us," Suleyman said. "Copilot in the workplace, Copilot at home is the future of the company. On the consumer side, we are going to be committed to this for many decades to come. We really think it's the major platform shift that we have to win." Even as competition ramps up, so does wariness from Wall Street and big business customers about whether these AI products are worth their huge costs in computing power and energy. Suleyman's own daily interactions with Copilot demonstrate the shortcomings of generative AI technology that is prone to confidently spouting falsehoods -- known as hallucinations -- and still struggles to match some of the common-sense reasoning skills that come naturally to humans. Asked about how he's using the technology in his daily life, Suleyman scrolled through his phone and candidly revealed his most recent chatbot conversations -- including, in preparation for Friday's event, his request for it to calculate Microsoft's all-time cumulative revenue over 50 years. "It didn't correctly add up all of the revenue," he said. "That's a pretty hard problem, although it got it really close." He also recounted more playful encounters, such as a long chat about "whether to put minced beef or chunks of beef" in a chili recipe and uploading a photo of his açai bowl so that Copilot could visually analyze its contents. "We talked about how many calories were in it -- it was 465 calories -- and does it vary based on how much sweetener is in there and how much honey," Suleyman said. "It couldn't see the honey, but it could see all the other pieces. And I kind of got into a long back and forth about what is açai and where does it come from." Microsoft's own researchers, in collaboration with Carnegie Mellon University, recently published a study that found that generative AI can inhibit critical thinking skills of human workers and lead to overreliance on the technology -- a conclusion that Suleyman says he disagrees with. Making chatbots fun, useful and personable might be key in winning over workers wary of the technology -- especially when Gates, Suleyman and other tech leaders from Microsoft and elsewhere have loudly sounded the warning about their coming effects on employment. "The truth is that the nature of work is going to change," Suleyman said during a video interview on Microsoft's Teams platform, a way of communicating he said would have been unimaginable 30 to 50 years ago. "Now, it's kind of unimaginable to think that you're going to have a bunch of agents working for you at the office," he said. "There will be much less of the administration, much less of the drudgery ... which I think is going to free us up as knowledge workers to be a lot more creative and focus on the bigger picture." In our home life, too, a personal AI assistant is "going to take care of a tremendous amount of boring, tiring, exhausting administration in the back end. Booking, planning, organizing, buying, sourcing, researching, comparing. We've never had anything like that before that can automate all of those things. It's going to be pretty amazing."

Microsoft is turning 50. Is it ready for the next technological frontier?

A version of this article originally appeared in Quartz's members-only Weekend Brief newsletter. Quartz members get access to exclusive newsletters and more. Sign up here. In late 2022, Satya Nadella made what might be the defining bet of his career: pouring billions into OpenAI while restructuring Microsoft around artificial intelligence. At first, the move looked like strategic genius. While Google, Apple, and Meta scrambled to catch up in the AI race, Microsoft appeared to have secured the winning horse, giving it a commanding early lead. Wall Street rewarded this vision, adding over a trillion dollars to Microsoft's market cap. By early 2025, however, that vision has collided with a harsher reality. Microsoft now faces a perfect storm of challenges in its AI gambit: massive data center pullbacks, lagging stock performance relative to tech peers, growing tensions with OpenAI, and a market that may be showing some slow down in its interest in all things AI. For a company that has masterfully navigated every technological revolution of the past five decades, these warning signs suggest its biggest bet yet may be faltering. The stock market's verdict has been particularly harsh. Microsoft shares narrowly avoided their first eight-week losing streak since the 2008 financial crisis only through a last-minute rally on March 21, according to CNBC. Since reaching a closing high of $467.56 in July 2024, Microsoft is down about 16%. Even more telling, Microsoft is down 7% for the year while other AI-focused tech giants have surged, making it the only Magnificent 7 tech stock without a gain over the past year. Behind the scenes, Microsoft has been quietly backing away from its ambitious AI infrastructure plans. The company has canceled planned data centers that would have powered the next generation of AI systems. According to TD Cowen, Microsoft has abandoned facilities that would have boosted its computing power by roughly 14% - equivalent to the entire computing capacity of a major global tech hub like Tokyo. Microsoft has also decided to skip out on a $12 billion option for more data center capacity from cloud computing company Coreweave, according to Semafor. These retreats seem especially telling for a company that repeatedly claimed AI demand was "exceeding supply." Nadella himself hinted at the reality in a recent podcast, admitting "there will be overbuild" in data centers related to AI. Adding to Microsoft's challenges is a deteriorating relationship with OpenAI. According to The Information, Microsoft's AI chief Mustafa Suleyman has a dual mandate that reveals the company's growing uncertainty: maintain the OpenAI partnership while simultaneously putting Microsoft "on a path to self-sufficiency in AI so it won't have to rely indefinitely on OpenAI's technology." Microsoft is already developing its own family of AI models (internally called MAI) and testing models from competitors including Anthropic, xAI, DeepSeek and Meta as potential OpenAI replacements for Copilot. It has its work cutout for it with Copilot, which has struggled to gain meaningful traction in the market. A Gartner survey found few companies moving Copilot initiatives beyond initial testing, with "tangible business impact proving elusive" and implementation requiring "more effort than anticipated." One government IT executive described Copilot as "so far behind that it is frustrating to use." But Microsoft may not be alone in facing these challenges. There are growing signs that all AI tools might be struggling to find more users. According to the Fall 2024 Slack Workforce Index, AI adoption growth rates among U.S. workers have slowed to a mere percentage point gain over the last three months, compared to nearly double-digit gains in the same period a year earlier. A paper by Gartner said that CIOs were increasingly nervous about AI spending, with one Gartner analyst telling the UK tech publication The Register that 2025 could be the "year of the slide." As Microsoft celebrates its 50th anniversary, the company finds itself at a crossroads. Its AI strategy -- raising prices while forcing AI integration into products, developing competing models while maintaining its OpenAI partnership, and scaling back infrastructure while claiming unprecedented demand -- suggests a company trying to hedge its bets rather than confidently doubling down on the next technological revolution. For a tech giant that has thrived by repeatedly reinventing itself, Microsoft's contradictory approach to AI may be the clearest signal yet that even industry leaders aren't sure if the AI revolution will live up to its trillion-dollar hype.

Microsoft and the next 'technological frontier'

In late 2022, Satya Nadella made what might be the defining bet of his career: pouring billions into OpenAI while restructuring Microsoft around artificial intelligence. At first, the move looked like strategic genius. While Google, Apple, and Meta scrambled to catch up in the AI race, Microsoft appeared to have secured the winning horse, giving it a commanding early lead. Wall Street rewarded this vision, adding over a trillion dollars to Microsoft's market cap. By early 2025, however, that vision has collided with a harsher reality. Microsoft now faces a perfect storm of challenges in its AI gambit: massive data center pullbacks, lagging stock performance relative to tech peers, growing tensions with OpenAI, and a market that may be showing some slow down in its interest in all things AI. For a company that has masterfully navigated every technological revolution of the past five decades, these warning signs suggest its biggest bet yet may be faltering. The stock market's verdict has been particularly harsh. Microsoft shares narrowly avoided their first eight-week losing streak since the 2008 financial crisis only through a last-minute rally on March 21, according to CNBC. Since reaching a closing high of $467.56 in July 2024, Microsoft is down about 16%. Even more telling, Microsoft is down 7% for the year while other AI-focused tech giants have surged, making it the only Magnificent 7 tech stock without a gain over the past year. Behind the scenes, Microsoft has been quietly backing away from its ambitious AI infrastructure plans. The company has canceled planned data centers that would have powered the next generation of AI systems. According to TD Cowen, Microsoft has abandoned facilities that would have boosted its computing power by roughly 14% - equivalent to the entire computing capacity of a major global tech hub like Tokyo. Microsoft has also decided to skip out on a $12 billion option for more data center capacity from cloud computing company Coreweave, according to Semafor. These retreats seem especially telling for a company that repeatedly claimed AI demand was "exceeding supply." Nadella himself hinted at the reality in a recent podcast, admitting "there will be overbuild" in data centers related to AI. Adding to Microsoft's challenges is a deteriorating relationship with OpenAI. According to The Information, Microsoft's AI chief Mustafa Suleyman has a dual mandate that reveals the company's growing uncertainty: maintain the OpenAI partnership while simultaneously putting Microsoft "on a path to self-sufficiency in AI so it won't have to rely indefinitely on OpenAI's technology." Microsoft is already developing its own family of AI models (internally called MAI) and testing models from competitors including Anthropic, xAI, DeepSeek and Meta as potential OpenAI replacements for Copilot. It has its work cutout for it with Copilot, which has struggled to gain meaningful traction in the market. A Gartner survey found few companies moving Copilot initiatives beyond initial testing, with "tangible business impact proving elusive" and implementation requiring "more effort than anticipated." One government IT executive described Copilot as "so far behind that it is frustrating to use." But Microsoft may not be alone in facing these challenges. There are growing signs that all AI tools might be struggling to find more users. According to the Fall 2024 Slack Workforce Index, AI adoption growth rates among U.S. workers have slowed to a mere percentage point gain over the last three months, compared to nearly double-digit gains in the same period a year earlier. A paper by Gartner said that CIOs were increasingly nervous about AI spending, with one Gartner analyst telling the UK tech publication The Register that 2025 could be the "year of the slide." As Microsoft celebrates its 50th anniversary, the company finds itself at a crossroads. Its AI strategy -- raising prices while forcing AI integration into products, developing competing models while maintaining its OpenAI partnership, and scaling back infrastructure while claiming unprecedented demand -- suggests a company trying to hedge its bets rather than confidently doubling down on the next technological revolution. For a tech giant that has thrived by repeatedly reinventing itself, Microsoft's contradictory approach to AI may be the clearest signal yet that even industry leaders aren't sure if the AI revolution will live up to its trillion-dollar hype.

Microsoft rode the cloud to new heights: Can it do the same with AI?

When Satya Nadella entered Microsoft's C-suite in 2014, the company wasn't one of the tech darlings dazzling Wall Street. After founder Bill Gates turned the CEO role over to Steve Ballmer in 2000 and until Ballmer stepped away in 2014, Microsoft's stock price basically followed the market. The company was still printing money thanks to its dominance in personal computing software. But it couldn't draw futurist headlines like those chasing companies that represented FAANG, a catchall acronym for Facebook, Apple, Amazon, Netflix and Google, the sexy tech stocks of the early 2010s. While Apple and Google seized the smartphone market, Microsoft fumbled its Windows-powered cellphone. As it revolutionized retail, Amazon also beat Microsoft to the punch on cloud computing with its debut of Amazon Web Services in 2002. Microsoft was resting on its laurels, even if those behind the scenes were feverishly trying to break into new markets with ultimately doomed products like the Zune. Enter Nadella, the engineer. Climbing through the ranks at Microsoft in various product divisions, Nadella had found a way at each stop to bring the company into a new digital age: take it to the cloud. As a concept, cloud computing is abstract. It took years to drag customers, both commercial and consumer, into Microsoft's cloud umbrella, Azure. But the bet paid off. And with the winnings, Nadella and others are pushing more chips each year into a technology better known than cloud computing but perhaps even less understood: artificial intelligence. From July 2024 through June 2025, the company projects it will have spent $80 billion on AI infrastructure, mostly by building data centers. That represents almost 33% of Microsoft's total revenue for 2024, which the company reported as $245 billion. Microsoft isn't all in, but it's still a big bet for a company worth $2.9 trillion. To the cloud You'd be hard-pressed to find someone who says "send that to me in a WordPerfect document." Nobody tracks their company budget in a Lotus 1-2-3 spreadsheet. Two applications that were ubiquitous for computer users in the 1980s are now completely lost to history thanks to Microsoft Office. Those Office products -- chiefly Word, PowerPoint and Excel -- once drove the Redmond-based tech giant. Under Nadella, cloud computing outstripped them all. When Nadella was named CEO, Office and the cloud were essentially tied in terms of revenue. Office brought in $24.3 billion in revenue. Microsoft's cloud computing division reported $21.7 billion in sales that year. Cloud computing is booming for Microsoft. Last year, Intelligent Cloud reported more than $100 billion in annual sales. Office's revenue was about half of that. Though tens of thousands of employees are underneath Nadella developing Microsoft's successful cloud products, it's Microsoft's chief who's been credited with turning around a company that continually failed to stay on the cutting edge of technology for more than a decade. Windows didn't make a dent in the smartphone market, now the domain of Apple's iOS and Google's Android operating systems. The ill-fated Zune didn't take down the iPod. And though MSN looked to be a contender as a search platform early on, the company lost interest in capturing that market before Google did. Though it's a success now, Microsoft was late in launching its cloud platform, Azure. Amazon Web Services entered the arena in 2002, six years before Azure did. Amazon spent that time gobbling up market share. While Microsoft and others were growing cloud services slowly, Amazon Web Services was humming. Amazon still has a healthy lead in the cloud market, with about 30% of it, according to the Synergy Research Group. But Microsoft's share has climbed to 21%. After more than a decade of dragging customers to the cloud, Nadella and Microsoft are pulling them toward AI, a technology that's costly, unreliable and greedy, using vast amounts of energy and water, but potentially extraordinarily lucrative. General-purpose technology Tech companies have developed large language models and machine learning products for the past decade that fall under AI. It's technology that allows a robot vacuum to clean a house more efficiently and produces better predictive ads while you're browsing the internet. But the flashier products are in the zeitgeist now. Chatbots that can craft essays and image generators that produce uncanny pictures are all available to the public. The tech industry is scrambling to include AI -- or at least the term AI -- in new products, annual earnings calls and marketing efforts. Facebook parent company Meta injected AI chatbots into its social media platforms, while Expedia has an AI feature for travel planning and Google's search has been handed over to the technology. For about $1,000, you can even own the world's first AI-enabled grill from a company called Brisk It. Microsoft's biggest consumer-facing AI product has been Copilot, a complement to the company's Office products that can automate tasks. Behind the scenes, Microsoft plans to pour money into AI. In a regulatory filing from October, Microsoft said it has committed $13 billion in investments toward OpenAI, an AI research organization that develops chatbots and image generators. Microsoft has such a large investment in the ChatGPT maker that it reported a $683 million expense due to its equity stake. As Wall Street rides the roller coaster of AI hype, Microsoft claims it's chasing the technology almost altruistically. The company believes that AI is a general-purpose technology. "A single-purpose technology is a great advancement and helps a lot of people," said Mary Snapp, vice president of strategic initiatives for Microsoft, in an interview with The Seattle Times. "Like my electric sewing machine when I was growing up. Boy, I sewed a lot of clothes and it got me to the prom, but it didn't do anything else." But an example of general-purpose technology, she said, is electricity. It spawns innovation and boosts productivity. An ambassador for the company's policy and tech efforts, Snapp travels around the world trying to demystify AI and advocate for retraining workforces. She likens it to the government trying to drive social acceptance of electricity in rural areas in the 1930s to build the electric grid. "We needed people to know how to run electrical appliances at scale and in industrial settings," Snapp said. "We're going to need the same thing with artificial intelligence." AI in action In early March, Microsoft President Brad Smith was in Tanzania to see Microsoft's AI for Good Research Lab in the field. The technology's selling point is improved productivity and efficiency in any setting. So, how could it help conservation? The lab had been partnering with the Wild Nature Institute, a conservation group, and was working with the organization to track the East African country's giraffe population. "Turns out that AI is critical," Smith said in an interview with The Seattle Times. "Every giraffe is unique, so for decades, these scientists would go out into the wild and they would have these diagrams of the giraffes that were drawn. Now you take a photograph, AI crops it, and then it instantaneously matches that giraffe to the giraffe in the dataset." Outside of Microsoft's research efforts, the company has already introduced its customers to AI. Deb Cupp, president of Microsoft's business operations across North America and South America, is in charge of an $80 billion products business that's now tasked with delivering AI to customers and educating them on the technology. "We spent time explaining what AI was and where the areas of opportunity actually were," she says. "We started to help customers think through whether they should try and create use cases, finding ways to fit in the technology. But the answer is actually the complete opposite." Renton-based Providence Health & Services said it's been working with the tech giant to identify use cases for AI. Health care has been moving to the digital world for the past 10 to 15 years, a move that Providence said has set it up for AI. Susan Huang, chief physician executive for health care giant Providence, said AI already has more buy-in than the organization's move to the cloud years ago. "AI is part of the colloquial dialogue whereas cloud-based services were somewhat niche," Huang said. "Whether you're a caregiver, doctor or nurse, everyone has an idea about how AI can be applied. And they're already using it on the outside and in their personal lives." It uses a lot of energy. The International Energy Agency said in a report last year that from 2022 to 2026, global electricity demand from data centers could double in part due to AI. The demand from data centers would be roughly equal to the entire electricity consumption of Japan. The average amount of electricity used in a traditional Google search is 0.3 watt-hours. For a ChatGPT request, it's about 2.9 watt-hours, the equivalent of running an LED light bulb for roughly 17 minutes. Essentially, the company is looking in the margins for where it can cut back on energy and water consumption while tightening AI's belt. If the models and code can be more efficient, they'll use less energy with each request. AI's other issue is its tendency to hallucinate. Google's Gemini feature regularly spits back incorrect information, as it's drawing from unreliable sources across the internet. OpenAI's ChatGPT has made up facts and gets tricked by word problems. Grok, the AI platform for X, routinely aggregates trendy jokes and reports them back as news. A recent study from the Columbia Journalism Review found that AI search tools were "generally bad at declining to answer questions they couldn't answer accurately, offering incorrect or speculative answers instead." Former Microsoft Chief Technology Officer Nathan Myrhvold believes it will take a miracle or maybe five for AI to reach human-level understanding. At Geekwire's Microsoft@50 anniversary event on March 20, Myrhvold expressed optimism for AI. But the feature it lacks is the ability to take abstract concepts and find meaning, as humans can. It can't reason in the abstract as humans do until a miracle happens. "And that could happen tomorrow. Or maybe it already happened tonight and they just haven't told us," Myrhvold said. "Or it could take another 10 years." For now it's about getting a return on investment and satisfying Wall Street. How Microsoft makes money through AI is a bit murky because it's injecting the technology throughout its revenue segments. But Nadella gave analysts a good sales pitch in January, during the company's latest earnings call. The company said its AI business has a revenue run rate, effectively a projection of revenue for the next year, of $13 billion. On the same day, Microsoft revealed it spent $22.6 billion in a single quarter on AI infrastructure. Nadella and the company deem those rising expenditures necessary. "As AI becomes more efficient and accessible," Nadella said, "we will see exponentially more demand." 2025 The Seattle Times. Distributed by Tribune Content Agency, LLC.

Microsoft's AI division head wants to create a lasting relationship between chatbots and their users

Fifty years after the founding of Microsoft, the CEO of its artificial intelligence division has a big task: develop a new product line as integral to daily life as the software giant's past innovations. "We're really trying to land this idea that everybody is going to have their own personalized AI companion," said Mustafa Suleyman in an interview with The Associated Press. "It will, over time, have its own name, its own style. It will adapt to you. It may also have its own visual appearance and expressions." Suleyman is laying out that vision on Microsoft's 50th anniversary Friday. The celebration at Microsoft's headquarters in Redmond, Washington will feature the first public gathering in more than a decade of co-founder and former CEO Bill Gates with his two successors: Steve Ballmer, who led the company from 2000 to 2014, and the current chief executive, Satya Nadella. Giving the mic to Suleyman, who joined the company and Nadella's senior leadership team just over a year ago to head a newly formed Microsoft AI division, signals how important getting its AI right is to the company's future -- in the next five years if not the next 50. The company's flagship product of this AI era, Copilot, already combines a chatbot with Microsoft's suite of workaday tools, from Excel spreadsheets and PowerPoint presentations to the Windows operating system that defines how most computers work. But Suleyman is striving for something that sounds a little more like science fiction -- a technology that can form a "lasting, meaningful relationship" with its users. "One that knows your name, gets to know you, has a memory of everything that you've shared with it and talked about and really comes to kind of live life alongside you," he said. "It's far more than just a piece of software or a tool. It is unlike anything we've really ever created." Some of those updates -- such as new "visual memory" capabilities that keep track of a chatbot user's digital activity, if they want that -- roll out on mobile apps Friday. Other features are still in development. Suleyman, 40, came to Microsoft last year with plenty of credentials in the AI business. In his 20s, the British entrepreneur co-founded the legendary DeepMind AI research lab in London, which Google later purchased in 2014. He worked there until 2022, when he left to set up the new company Inflection AI with LinkedIn co-founder Reid Hoffman. Microsoft then scooped up Suleyman and other Inflection leaders without formally acquiring the startup, attracting months of antitrust scrutiny. He also co-wrote a 2023 book, "The Coming Wave," focused on AI's promise and the need to limit its potential perils. But unlike DeepMind, or another famous AI research lab, San Francisco-based OpenAI, both of which have explicitly set out to build better-than-human AI known as artificial general intelligence, Suleyman is not too worried about such abstractions in his new job running Microsoft's consumer AI business. "My goal is really to create a true personal AI companion," he said. "And the definition of AGI sort of feels very far out to me and sort of not what I'm focused on in the next few years." The race to build the best AI personal assistant that sticks with consumers is a competitive one. In recent weeks, rivals including Google and Meta Platforms have shaken up the teams in charge of AI research or products. ChatGPT maker OpenAI, which is both Microsoft's most important AI partner and a growing competitor, has also reorganized its leadership. Amazon is also looking to catch up by imbuing its already-ubiquitous digital assistant Alexa with more advanced AI capabilities. "It's a super competitive market but this is absolutely foundational to us," Suleyman said. "Copilot in the workplace, Copilot at home is the future of the company. On the consumer side, we are going to be committed to this for many decades to come. We really think it's the major platform shift that we have to win." Even as competition ramps up, so does wariness from Wall Street and big business customers about whether these AI products are worth their huge costs in computing power and energy. Suleyman's own daily interactions with Copilot demonstrate the shortcomings of generative AI technology that is prone to confidently spouting falsehoods -- known as hallucinations -- and still struggles to match some of the common-sense reasoning skills that come naturally to humans. Asked about how he's using the technology in his daily life, Suleyman scrolled through his phone and candidly revealed his most recent chatbot conversations -- including, in preparation for Friday's event, his request for it to calculate Microsoft's all-time cumulative revenue over 50 years. "It didn't correctly add up all of the revenue," he said. "That's a pretty hard problem, although it got it really close." He also recounted more playful encounters, such as a long chat about "whether to put minced beef or chunks of beef" in a chili recipe and uploading a photo of his açai bowl so that Copilot could visually analyze its contents. "We talked about how many calories were in it -- it was 465 calories -- and does it vary based on how much sweetener is in there and how much honey," Suleyman said. "It couldn't see the honey, but it could see all the other pieces. And I kind of got into a long back and forth about what is açai and where does it come from." Microsoft's own researchers, in collaboration with Carnegie Mellon University, recently published a study that found that generative AI can inhibit critical thinking skills of human workers and lead to overreliance on the technology -- a conclusion that Suleyman says he disagrees with. Making chatbots fun, useful and personable might be key in winning over workers wary of the technology -- especially when Gates, Suleyman and other tech leaders from Microsoft and elsewhere have loudly sounded the warning about their coming effects on employment. "The truth is that the nature of work is going to change," Suleyman said during a video interview on Microsoft's Teams platform, a way of communicating he said would have been unimaginable 30 to 50 years ago. "Now, it's kind of unimaginable to think that you're going to have a bunch of agents working for you at the office," he said. "There will be much less of the administration, much less of the drudgery ... which I think is going to free us up as knowledge workers to be a lot more creative and focus on the bigger picture." In our home life, too, a personal AI assistant is "going to take care of a tremendous amount of boring, tiring, exhausting administration in the back end. Booking, planning, organizing, buying, sourcing, researching, comparing. We've never had anything like that before that can automate all of those things. It's going to be pretty amazing." © 2025 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed without permission.

Will Microsoft -- maker of Zune and Bing -- win tech's new big thing?

When Satya Nadella entered Microsoft's C-suite in 2014, the company wasn't one of the tech darlings dazzling Wall Street. After founder Bill Gates turned the CEO role over to Steve Ballmer in 2000 and until Ballmer stepped away in 2014, Microsoft's stock price basically followed the market. The company was still printing money thanks to its dominance in personal computing software. But it couldn't draw futurist headlines like those chasing companies that represented FAANG, a catchall acronym for Facebook, Apple, Amazon, Netflix and Google, the sexy tech stocks of the early 2010s. While Apple and Google seized the smartphone market, Microsoft fumbled its Windows-powered cellphone. As it revolutionized retail, Amazon also beat Microsoft to the punch on cloud computing with its debut of Amazon Web Services in 2002. Microsoft was resting on its laurels, even if those behind the scenes were feverishly trying to break into new markets with ultimately doomed products like the Zune. Enter Nadella, the engineer. Climbing through the ranks at Microsoft in various product divisions, Nadella had found a way at each stop to bring the company into a new digital age: take it to the cloud. As a concept, cloud computing is abstract. It took years to drag customers, both commercial and consumer, into Microsoft's cloud umbrella, Azure. But the bet paid off. And with the winnings, Nadella and others are pushing more chips each year into a technology better known than cloud computing but perhaps even less understood: artificial intelligence. From July 2024 through June 2025, the company projects it will have spent $80 billion on AI infrastructure, mostly by building data centers. That represents almost 33% of Microsoft's total revenue for 2024, which the company reported as $245 billion. Microsoft isn't all in, but it's still a big bet for a company worth $2.9 trillion. To the cloud You'd be hard-pressed to find someone who says "send that to me in a WordPerfect document." Nobody tracks their company budget in a Lotus 1-2-3 spreadsheet. Two applications that were ubiquitous for computer users in the 1980s are now completely lost to history thanks to Microsoft Office. Those Office products -- chiefly Word, PowerPoint and Excel -- once drove the Redmond-based tech giant. Under Nadella, cloud computing outstripped them all. When Nadella was named CEO, Office and the cloud were essentially tied in terms of revenue. Office brought in $24.3 billion in revenue. Microsoft's cloud computing division reported $21.7 billion in sales that year. Cloud computing is booming for Microsoft. Last year, Intelligent Cloud reported more than $100 billion in annual sales. Office's revenue was about half of that. Though tens of thousands of employees are underneath Nadella developing Microsoft's successful cloud products, it's Microsoft's chief who's been credited with turning around a company that continually failed to stay on the cutting edge of technology for more than a decade. Windows didn't make a dent in the smartphone market, now the domain of Apple's iOS and Google's Android operating systems. The ill-fated Zune didn't take down the iPod. And though MSN looked to be a contender as a search platform early on, the company lost interest in capturing that market before Google did. Though it's a success now, Microsoft was late in launching its cloud platform, Azure. Amazon Web Services had entered the arena in 2002, six years before Azure did. Amazon spent that time gobbling up market share. While Microsoft and others were growing cloud services slowly, Amazon Web Services was humming. Amazon still has a healthy lead in the cloud market, with about 30% of it, according to the Synergy Research Group. But Microsoft's share has climbed to 21%. After more than a decade of dragging customers to the cloud, Nadella and Microsoft are pulling them toward AI, a technology that's costly, unreliable and greedy, using vast amounts of energy and water, but potentially extraordinarily lucrative. General-purpose technology Tech companies have developed large language models and machine learning products for the past decade that fall under AI. It's technology that allows a robot vacuum to clean a house more efficiently and produces better predictive ads while you're browsing the internet. But the flashier products are in the zeitgeist now. Chatbots that can craft essays and image generators that produce uncanny pictures are all available to the public. The tech industry is scrambling to include AI -- or at least the term AI -- in new products, annual earnings calls and marketing efforts. Facebook parent company Meta injected AI chatbots into its social media platforms, while Expedia has an AI feature for travel planning and Google's search has been handed over to the technology. For about $1,000, you can even own the world's first AI-enabled grill from a company called Brisk It. Microsoft's biggest consumer-facing AI product has been Copilot, a complement to the company's Office products that can automate tasks. Behind the scenes, Microsoft plans to pour money into AI. In a regulatory filing from October, Microsoft said it has committed $13 billion in investments toward OpenAI, an AI research organization that develops chatbots and image generators. Microsoft has such a large investment in the ChatGPT maker that it reported a $683 million expense due to its equity stake. As Wall Street rides the roller coaster of AI hype, Microsoft claims it's chasing the technology almost altruistically. The company believes that AI is a general-purpose technology. "A single-purpose technology is a great advancement and helps a lot of people," said Mary Snapp, vice president of strategic initiatives for Microsoft, in an interview with The Seattle Times. "Like my electric sewing machine when I was growing up. Boy, I sewed a lot of clothes and it got me to the prom, but it didn't do anything else." But an example of general-purpose technology, she said, is electricity. It spawns innovation and boosts productivity. An ambassador for the company's policy and tech efforts, Snapp travels around the world trying to demystify AI and advocate for retraining workforces. She likens it to the government trying to drive social acceptance of electricity in rural areas in the 1930s to build the electric grid. "We needed people to know how to run electrical appliances at scale and in industrial settings," Snapp said. "We're going to need the same thing with artificial intelligence." AI in action In early March, Microsoft President Brad Smith was in Tanzania to see Microsoft's AI for Good Research Lab in the field. The technology's selling point is improved productivity and efficiency in any setting. So, how could it help conservation? The lab had been partnering with the Wild Nature Institute, a conservation group, and was working with the organization to track the East African country's giraffe population. "Turns out that AI is critical," Smith said in an interview with The Seattle Times. "Every giraffe is unique, so for decades, these scientists would go out into the wild and they would have these diagrams of the giraffes that were drawn. Now you take a photograph, AI crops it, and then it instantaneously matches that giraffe to the giraffe in the data set." Outside of Microsoft's research efforts, the company has already introduced its customers to AI. Deb Cupp, president of Microsoft's business operations across North America and South America, is in charge of an $80 billion products business that's now tasked with delivering AI to customers and educating them on the technology. "We spent time explaining what AI was and where the areas of opportunity actually were," she says. "We started to help customers think through whether they should try and create use cases, finding ways to fit in the technology. But the answer is actually the complete opposite." Renton-based Providence Health & Services said it's been working with the tech giant to identify use cases for AI. Health care has been moving to the digital world for the past 10 to 15 years, a move that Providence said has set it up for AI. Susan Huang, chief physician executive for health care giant Providence, said AI already has more buy-in than the organization's move to the cloud years ago. "AI is part of the colloquial dialogue whereas cloud-based services were somewhat niche," Huang said. "Whether you're a caregiver, doctor or nurse, everyone has an idea for how AI can be applied. And they're already using it on the outside and in their personal lives." It uses a lot of energy. The International Energy Agency said in a report last year that from 2022 to 2026, global electricity demand from data centers could double in part due to AI. The demand from data centers would be roughly equal to the entire electricity consumption of Japan. The average amount of electricity used in a traditional Google search is 0.3 watt-hours. For a ChatGPT request, it's about 2.9 watt-hours, the equivalent of running an LED light bulb for roughly 17 minutes. Essentially the company is looking in the margins for where it can cut back on energy and water consumption while tightening AI's belt. If the models and code can be more efficient, they'll use less energy with each request. AI's other issue is its tendency to hallucinate. Google's Gemini feature regularly spits back incorrect information, as it's drawing from unreliable sources across the internet. OpenAI's ChatGPT has made up facts and gets tricked by word problems. Grok, the AI platform for X, routinely aggregates trendy jokes and reports them back as news. A recent study from the Columbia Journalism Review found that AI search tools were "generally bad at declining to answer questions they couldn't answer accurately, offering incorrect or speculative answers instead." Former Microsoft Chief Technology Officer Nathan Myrhvold believes it will take a miracle or maybe five for AI to reach human-level understanding. At Geekwire's Microsoft@50 anniversary event March 20, Myrhvold expressed optimism for AI. But the feature it lacks is the ability to take abstract concepts and find meaning, as humans can. It can't reason in the abstract as humans do until a miracle happens. "And that could happen tomorrow. Or maybe it already happened tonight and they just haven't told us," Myrhvold said. "Or it could take another 10 years." For now it's about getting a return on investment and satisfying Wall Street. How Microsoft makes money through AI is a bit murky because it's injecting the technology throughout its revenue segments. But Nadella gave analysts a good sales pitch in January, during the company's latest earnings call. The company said its AI business has a revenue run rate, effectively a projection of revenue for the next year, of $13 billion. On the same day, Microsoft revealed it spent $22.6 billion in a single quarter on AI infrastructure. Nadella and the company deem those rising expenditures necessary. "As AI becomes more efficient and accessible," Nadella said, "we will see exponentially more demand."

Microsoft's AI division head wants to create a lasting relationship between chatbots and their users

The company's flagship product of this AI era, Copilot, already combines a chatbot with Microsoft's suite of workaday tools, from Excel spreadsheets and PowerPoint presentations to the Windows operating system that defines how most computers work. But Suleyman is striving for something that sounds a little more like science fiction - a technology that can form a "lasting, meaningful relationship" with its users. Fifty years after the founding of Microsoft, the CEO of its artificial intelligence division has a big task: develop a new product line as integral to daily life as the software giant's past innovations. "We're really trying to land this idea that everybody is going to have their own personalised AI companion," said Mustafa Suleyman in an interview with The Associated Press. "It will, over time, have its own name, its own style. It will adapt to you. It may also have its own visual appearance and expressions." Suleyman laid out that vision on Microsoft's 50th anniversary Friday. The celebration at Microsoft's headquarters in Redmond, Washington featured the first public gathering in more than a decade of co-founder and former CEO Bill Gates with his two successors: Steve Ballmer, who led the company from 2000 to 2014, and the current chief executive, Satya Nadella. Giving the mic to Suleyman, who joined the company and Nadella's senior leadership team just over a year ago to head a newly formed Microsoft AI division, signals how important getting its AI right is to the company's future - in the next five years if not the next 50. The company's flagship product of this AI era, Copilot, already combines a chatbot with Microsoft's suite of workaday tools, from Excel spreadsheets and PowerPoint presentations to the Windows operating system that defines how most computers work. But Suleyman is striving for something that sounds a little more like science fiction - a technology that can form a "lasting, meaningful relationship" with its users. "One that knows your name, gets to know you, has a memory of everything that you've shared with it and talked about and really comes to kind of live life alongside you," he said. "It's far more than just a piece of software or a tool. It is unlike anything we've really ever created." Some of those updates - such as new "visual memory" capabilities that keep track of a user's digital activity, if they want that - roll out on mobile apps Friday. Other features are still in development, such as an animated avatar - a talking peacock in Suleyman's demo Friday - that would embody a person's AI companion. Suleyman, 40, came to Microsoft last year with plenty of credentials in the AI business. In his 20s, the British entrepreneur cofounded the legendary DeepMind AI research lab in London, which Google later purchased in 2014. He worked there until 2022, when he left to set up the new company Inflection AI with LinkedIn co-founder Reid Hoffman. Microsoft then scooped up Suleyman and other Inflection leaders without formally acquiring the startup, attracting months of antitrust scrutiny. He also cowrote a 2023 book, "The Coming Wave," focused on AI's promise and the need to limit its potential perils. But unlike DeepMind, or another famous AI research lab, San Francisco-based OpenAI, both of which have explicitly set out to build better-than-human AI known as artificial general intelligence, Suleyman is not too worried about such abstractions in his new job running Microsoft's consumer AI business. "My goal is really to create a true personal AI companion," he said. "And the definition of AGI sort of feels very far out to me and sort of not what I'm focused on in the next few years." The race to build the best AI personal assistant that sticks with consumers is a competitive one. In recent weeks, rivals including Google and Meta Platforms have shaken up the teams in charge of AI research or products. ChatGPT maker OpenAI, which is both Microsoft's most important AI partner and a growing competitor, has also reorganized its leadership. Amazon is also looking to catch up by imbuing its already-ubiquitous digital assistant Alexa with more advanced AI capabilities. "It's a super competitive market but this is absolutely foundational to us," Suleyman said. "Copilot in the workplace, Copilot at home is the future of the company. On the consumer side, we are going to be committed to this for many decades to come. We really think it's the major platform shift that we have to win." Even as competition ramps up, so does wariness from Wall Street and big business customers about whether these AI products are worth their huge costs in computing power and energy. Suleyman's own daily interactions with Copilot demonstrate the shortcomings of generative AI technology that is prone to confidently spouting falsehoods - known as hallucinations - and still struggles to match some of the commonsense reasoning skills that come naturally to humans. Asked about how he's using the technology in his daily life, Suleyman scrolled through his phone and candidly revealed his most recent chatbot conversations - including, in preparation for Friday's event, his request for it to calculate Microsoft's all-time cumulative revenue over 50 years. "It didn't correctly add up all of the revenue," he said. "That's a pretty hard problem, although it got it really close." He also recounted more playful encounters, such as a long chat about "whether to put minced beef or chunks of beef" in a chili recipe and uploading a photo of his acai bowl so that Copilot could visually analyze its contents. "We talked about how many calories were in it -- it was 465 calories -- and does it vary based on how much sweetener is in there and how much honey," Suleyman said. "It couldn't see the honey, but it could see all the other pieces. And I kind of got into a long back and forth about what is acai and where does it come from." Microsoft's own researchers, in collaboration with Carnegie Mellon University, recently published a study that found that generative AI can inhibit critical thinking skills of human workers and lead to overreliance on the technology - a conclusion that Suleyman says he disagrees with. Making chatbots fun, useful and personable might be key in winning over workers wary of the technology - especially when Gates, Suleyman and other tech leaders from Microsoft and elsewhere have loudly sounded the warning about their coming effects on employment. "The truth is that the nature of work is going to change," Suleyman said during a video interview on Microsoft's Teams platform, a way of communicating he said would have been unimaginable 30 to 50 years ago. "Now, it's kind of unimaginable to think that you're going to have a bunch of agents working for you at the office," he said. "There will be much less of the administration, much less of the drudgery ... which I think is going to free us up as knowledge workers to be a lot more creative and focus on the bigger picture." In our home life, too, a personal AI assistant is "going to take care of a tremendous amount of boring, tiring, exhausting administration in the back end. Booking, planning, organizing, buying, sourcing, researching, comparing. We've never had anything like that before that can automate all of those things. It's going to be pretty amazing." As he outlined those ideas in his keynote speech Friday, a shouting protester interrupted Suleyman to protest Microsoft's contracts to provide AI and cloud computing services to the Israeli military. An investigation revealed earlier this year that AI models from Microsoft and OpenAI had been used as part of an Israeli military program to select bombing targets during the recent wars in Gaza and Lebanon. "Thank you, I hear your protest," Suleyman said repeatedly, before the shouting stopped and he returned to talking about AI companions.