Generative AI in 2025: Solution Providers Focus on ROI and Industry-Specific Applications

GenAI's Year Three: Solution Providers Zero In On ROI, Industry Use Cases

Even when an AI solution reaches the production stage, that doesn't mean the work is done. Here is how six key players are stepping up and standing out. In the third year since OpenAI's ChatGPT ushered in this era of generative artificial intelligence, users no longer see the technology as a shiny new toy. They instead seek, above all, returns on investment, solutions tailored to their industries and ways to make specific job roles or "personas" more productive and less mundane, executives with some of the largest solution providers in the world told CRN. Recent innovations, including small language models and AI agents, are helping speed up GenAI's adoption and bring down its scaling cost. Solution providers are working on new ways to package AI tools and services to better market and implement the technology with customers and stand out from the noise. While much of the business world is early in its AI adoption journey, some of the more mature AI users are looking for ways to optimize usage, showing that for solution providers, the AI engagement doesn't end once a solution is in production. [RELATED: How The CEOs Of AMD, HP, Lenovo And Acer Plan To Seize On AI In 2025] "Now people are a lot more pragmatic about what can be done -- what are the cost implications," Nidhi Srivastava, vice president and global head of AI Cloud offerings at India-based Tata Consultancy Services (TCS), No. 2 on CRN's 2024 Solution Provider 500, told CRN in an interview. "There is this whole shift in stance from the world of use cases to the world of business cases," she said. In January, Gartner forecast $5.61 trillion in worldwide IT spending in 2025, up almost 10 percent year over year. Data center systems, devices and software should each see double-digit growth in 2025, with the research firm attributing that largely to GenAI hardware upgrades. Srivastava told CRN that an example of the solution provider's AI wares taking off in manufacturing is an agent assistant for plant operators, which offers machine troubleshooting in local languages. For one large chemical plant customer, the agent reduced machine downtime by 5 percent, translating into millions of dollars in savings over the year, Srivastava said. TCS customers have largely moved past the AI experimentation phase and into a need to see ROI, she said. Out-of-the-box AI solutions and small language models have helped bring down costs and the time to implement AI solutions. Srivastava said that now about 30 percent to 40 percent of an AI solution purpose-built for niche industry use cases -- such as automating part of a job role -- could need customization. Banking, financial services, life sciences and media are some of the other industries where TCS is seeing demand for AI, she said. "You want to use AI for the right problem," she said. "You have to find the right nail for the hammer called AI." On TCS' latest quarterly earnings call, held Jan. 9, CEO and Managing Director K Krithivasan said that "GenAI, AI and cloud services continue to see significant growth for us this quarter" and told analysts "we are looking to infuse AI in every program." At HCLTech, health care and life sciences are some of the industries where the India-based solution provider of 220,000-plus employees worldwide sees early traction for its AI wares. Shrikanth Shetty, HCLTech chief growth officer and global head of life sciences and health-care industry, told CRN his customers want AI tailored to their industries instead of general-purpose AI, in part to meet strict regulations. Health-care customers, for example, have to account for HIPAA rules when applying GenAI to patient data, and "data is the oil that drives the AI engine," Shetty said. Solution providers are essential to landing AI with health-care customers who need help sifting through the hype to find out how to apply the technology and see the best ROI, he said. "Clients look at us for how we can bring [AI] to life within their industry, within their context and in an efficient fashion" he said. One of the more repeatable AI solutions HCLTech offers that doesn't require extensive customization is aimed at pharmaceutical customers. The solution ingests and summarizes information for federal audits and internal reviews to help speed up clinical trials and regulatory approvals. With health-care customers, HCLTech has leveraged GenAI-powered document summarization to more quickly collect patient information in doctors' offices and speed up the intake process. Shetty believes that these incremental savings and whatever advancements come with GenAI in the future can result in cost savings for health-care providers and better health outcomes for patients. "AI is a force multiplier here," he said. "This is not as far-thinking as discovering a molecule using GenAI, but these are far more real-world examples where we are seeing initial traction because that's where a lot of our enterprises are seeing [the ROI]." During HCLTech's latest quarterly earnings conference call on Jan. 13, CEO and Managing Director C Vijayakumar told analysts the solution provider saw "a sizable number of deals influenced by AI and Gen AI" and predicted "companies [will] increase their IT investments" in 2025. Valorem Reply -- No. 367 on CRN's 2024 Solution Provider 500 -- has an AI computer vision solution in production for a nonprofit customer, which other customers are also testing, Ryan McCamy, lead for tech for social impact at the Kansas City, Mo.- based provider, told CRN. The nonprofit leverages AI to analyze items as they come into its network of thrift stores and determine if the item will sell better online and how to best describe it online to encourage a purchase. It also could open more jobs at the retailer to people with visual impairments. "If you think about any kind of thrift store operating at scale around the country, millions of items are coming in the back door," he said. "If they can speed up their processing of those items by even 20 seconds per item, that represents a huge efficiency gain for them." For nonprofits, the relatively quick experimentation and adoption of AI is not just because of excitement around the technology, as plenty of nonprofits are still in the early journeys of cloud computing, he said. Instead, AI is seen as a new essential for efficiency, making existing workforces more productive, which is key for continually resource-strapped nonprofits. "We're trying to help them get way more done, knowing that they can never hire the people they need," he said. "It's the ability to use AI to help accelerate your grant-writing process. It's the ability to use AI to more focus and tailor the messaging to your donors. It's really those operational back-office productivity gains that you can get fairly quickly out of some real basic AI capabilities that you can get right out of the box." On the latest earnings call for Valorem Reply parent Reply, held in November, CEO Tatiana Rizzante said the Italy-based firm expects "a strong demand for artificial intelligence" over the next few years. A repeatable AI solution cutting across industries for France-based Capgemini -- No. 4 on CRN's 2024 Solution Provider 500 -- shortens the time taken responding to RFPs and requests for information with AI agents, Doug Ross, CTO of Capgemini-owned professional services and technology provider Sogeti, told CRN. Capgemini has also leveraged the tool internally for efficiency and markets the solution for customers as an accelerator -- prepackaged, plug-and-play software tools that leverage microservices architecture so customers aren't locked in to any one model or hyperscaler. Meanwhile, a brand voice AI accelerator the solution provider offers can be combined with a search optimization tool for writing better product descriptions and increasing online sales. "The ROI is so high on that," Ross said. "We're doing gangbusters business," he added. "[Accelerators are] a very useful tool to be able to go into a client and showcase not just a generic solution, but one that's really focused on their documents." Throughout 2025, Ross sees customers applying the RFP AI solution not only to larger proposals asking for hundreds of pages of information, but also smaller ones. He foresees customers using the technology in one department scaling to use the AI organization-wide as they grow more comfortable. Capgemini also iterates on the solution after it goes into production to make use of better models and technology on the market. "It's not a value-add for people just to be reading through those very lengthy documents and putting in and pasting in boilerplate," he said. Although a minority of customers still include contract language prohibiting AI from production environments, Ross predicts that AI angst will completely go away due to vendors' indemnification policies and the security controls. "As we get more attorneys familiar with how these partnerships work, with how the architectures work, they're going to be more and more comfortable," he said. On Capgemini's latest quarterly earnings call, held in October, CEO Aiman Ezzat said the data and AI pipeline has "almost doubled year on year" and that the "dynamism of the market in AI and GenAI" aligned with the firm's 6.5 percent growth year over year in strategy and transformation. Justin Mescher, vice president of strategy for AI, cloud and data center at Herndon, Va.-based ePlus -- No. 26 on CRN's 2024 Solution Provider 500 -- told CRN that even among the minority of customers he would call AI-mature, the solution provider is finding repeat work optimizing those customers' spend on GPUs and power consumption as they scale AI solutions and amass more data. In the case of a research firm that was racking up AI costs running workloads 24 hours a day related to product development and marketing, ePlus built a new managed service for the firm's machine learning operations. EPlus continues to manage that service and has implemented a similar service with other customers caught off guard by AI's fast proliferation that don't have the time and resources to hire AI knowledge workers. "We created a line of communication between their IT team and their data scientists and their AI developers," he said. "It gave us access to new buyers, and it gave us access to new revenue streams as part of that." Still, Mescher said a majority of ePlus' customers are curious about AI, looking to the solution provider to help with proofs of concept, putting in data policies and applying AI frameworks. EPlus has also differentiated itself in the market through its status as one of 10 Nvidia Deep GPU Xceleration-ready MSPs worldwide. During ePlus' latest quarterly earnings call, held in November, President and CEO Mark Marron said, "We are encouraged by the progress we have made against our artificial intelligence initiatives." He added that the solution provider has been training all of its sales teams to counter "elongated" sales cycles from AI. "AI right now is a little bit of a headwind in that space where a lot of customers are evaluating what to do but haven't made decisions," Marron said. "It's real early innings there. And over time, we'd expect to see some upside there." Oil and gas companies have looked to AI from India-based Wipro to reduce accidents and carbon emissions, Nagendra Bandaru, Wipro president, managing partner and global head of enterprise futuring, told CRN. At least four of these customers have moved from the beta stage to piloting with the InspectAI solution from Wipro, No. 15 on CRN's 2024 Solution Provider 500. The Wipro solution leverages drones to predict and snuff out fires, with drones pouring water on sites when they detect sparks, according to Bandaru. "The [human element] is completely removed," he said. "It is very sustainable, and we are saving lives." His message to AI vendors is to focus on repeatable solutions for partners and understand the importance of interoperability to ease AI adoption in existing IT environments. While content generation and contact center AI solutions have received a great deal of hype, Bandaru said he believes that applying AI to problems with a smaller amount of variables is the way to find early success as the technology matures. "The problem with content is that it is not revenue-generating," he said. "What we are trying to focus on is to look at problems that can actually generate revenue for our customers and that are a pain for customers." On Wipro's latest quarterly earnings call, held Jan. 17, CEO and Managing Director Srini Pallia said that he expects "significant growth in AI spending" and described "good progress in our consulting-led, AI-powered, industry and cross-industry solutions." He also said that 50,000 Wipro employees now hold advanced AI certifications. "We are committed to driving innovation for our clients by leveraging the transformative power of AI," he said.

NTT Data CDAO: C-Suite Leaders Have 'Angst' About GenAI, Still Go Full Throttle | PYMNTS.com

Enterprises are accelerating their adoption of generative artificial intelligence (GenAI) at scale to gain a competitive advantage, especially in light of productivity and efficiency benefits across industries. But C-suite leaders also face a thorny dilemma: How to innovate fast and yet do it responsibly to ensure privacy and avoid regulatory scrutiny. The C-suite is saying, "I need to run fast. We need to drive business results, but I've got to do it in a secure, responsible way. I can't have my data leaked to the outside world. I can't drive an action that harms someone," said Andrew Wells, chief data and AI officer, North America, of NTT Data, the digital transformation unit of Japanese telecom giant NTT, in an interview with PYMNTS. "It was that dichotomy ... that was causing a lot of angst in a lot of the CEOs (and CIOs) that I talked to," Wells said. He should know. NTT Data serves 75% of the global Fortune 100 companies. It is "probably the biggest company no one's ever heard of," Wells said. NTT Data is the fifth largest IT service provider globally, with $30 billion in annual revenue across 50 countries. It is also the third largest data center provider. But this dilemma is not slowing down business leaders. Wells said he expects to book $2 billion in revenue for his Smart AI Agent tool alone by 2027. AI agents are a step up from AI chatbots like ChatGPT in that they not only provide information but also perform multistep tasks for users autonomously. For example, they can plan a vacation. AI agency is "the next big wave that's hitting us right now," Wells said, adding that today's CEOs would be the last to manage all-human teams, in a nod to Salesforce CEO Marc Benioff's comment at the World Economic Forum in January. Wells said the GenAI demand is there: Companies have "backlogs" of use cases that are just waiting to be transformed by GenAI apps -- but these apps have not yet been created or made available yet. Wells cited an NTT Data survey of 2,300 executives last November showing that 97% of CEOs see GenAI having a "material impact" on their operations, and 70% of CEOs expect it to lead to "significant" transformation in 2025. Those numbers dovetail with PYMNTS' own research. A PYMNTS Intelligence report shows that most chief financial officers see GenAI as having a crucial and growing role in financial reporting. Contrary to the public perception of AI hype reaching a plateau, Wells noted that executive enthusiasm remains strong. "At the time we did the [survey] in November, there was this feeling in the zeitgeist that [GenAI] was overhyped and going to be on the decline, and that's not what we were seeing at all in the data -- everyone was full on." Executives also were "excited" and "amazed" by GenAI, the survey said. But at the fast pace of AI development, GenAI is already starting to be commoditized even though it is a fairly new technology. "I look at GenAI as the phase that we're in now and starting to get commoditized," Wells said. "I think DeepSeek definitely shot the arrow across our bow to say that we can do this for a lot less and make models that have high efficacy." DeepSeek is a Chinese AI startup that developed foundation AI models that performed at par with top models from OpenAI, Google and Anthropic, but at a fraction of the cost. GenAI's rapid commoditization is reshaping how enterprises approach AI implementation. According to Wells, the focus is on building applications and agents on top of existing AI models rather than developing new foundation models from scratch. "Right now, we have the GenAI layer created, and you're going to start to see applications built on top of that, and those are more than likely going to take the shape of agents," he predicted. Wells said NTT Data is developing small language models for clients customized for their business processes. This makes the business processes smarter. They then layer AI agents on top to automate the tasks. Wells used the example of routing a customer call or email to the right department or people. GenAI is used to interpret the call or email to capture not only the content but also the sentiment and tone, he said. An AI agent then routes the call or email to the right place or party. As for tackling hallucinations, Wells said techniques are being used to mitigate them, such as GraphRAG (retrieval augmented generation). Companies could also train a small language model on internal data and employ fewer parameters in the AI model and then deeply training it on corporate datasets, to reduce hallucinations. To ensure accuracy in AI agents, one tactic is to create governance layers -- inserting a safety AI agent in the agentic workflow or a human in the loop, Wells said. Another way is to audit the AI agent's decisions, among other techniques. "There are guardrails you can put in place, but a lot of it depends on who's architecting the solution, and it's not going to be perfect," Wells said. "We're going to see silliness, and that's where you as the person who's doing the solution ... has to be very purposeful and pedantic to make sure that you're putting in the safeguards." But C-suite's focus remains on innovation. Notably, 60% of executives plan to innovate first and then ensure GenAI is deployed responsibly, Wells said their survey showed. Only 30% was the other way around. "It tells you the pressure that a lot of businesses are under -- you've got to get out there and adopt this technology, and not necessarily for cost savings," Wells said. "It's really for competitive differentiation and driving better products and services in the market." "The people that run to AI and start using these tools are going to be the ones that capitalize on them" best, Wells concluded. "The ones that don't use it are going to be the ones that get left behind."