OpenAI's Ambitious Plan: 100 Million GPUs and the Future of AI Infrastructure

Sam Altman teases 100 million GPU scale for OpenAI that could cost $3 trillion -- ChatGPT maker to cross 'well over 1 million' by end of year

OpenAI CEO Sam Altman isn't exactly known for thinking small, but his latest comments push the boundaries of even his usual brand of audacious tech talk. In a new post on X, Altman revealed that OpenAI is on track to bring "well over 1 million GPUs online" by the end of this year. That alone is an astonishing number -- consider that Elon Musk's xAI, which made waves earlier this year with its Grok 4 model, runs on about 200,000 Nvidia H100 GPUs. OpenAI will have five times that power, and it's still not enough for Altman going into the future. "Very proud of the team..." he wrote, "but now they better get to work figuring out how to 100x that lol." The "lol" might make it sound like he's joking, but Altman's track record suggests otherwise. Back in February, he admitted that OpenAI had to slow the rollout of GPT‑4.5 because they were literally "out of GPUs." That wasn't just a minor hiccup; it was a wake-up call considering Nvidia is also sold out till next year for its premier AI hardware. Altman has since made compute scaling a top priority, pursuing partnerships and infrastructure projects that look more like national-scale operations than corporate IT upgrades. When OpenAI hits its 1 million GPU milestone later this year, it won't just be a social media flex -- it'll be cementing itself as the single largest consumer of AI compute on the planet. Anyhow, let's talk about that 100x goal, because it's exactly as wild as it sounds. At current market prices, 100 million GPUs would cost around $3 trillion -- almost the GDP of the UK -- and that's before factoring in the power requirements or the data centers needed to house them. There's no way Nvidia could even produce that many chips in the near term, let alone handle the energy requirements to power them all. Yet, that's the kind of moonshot thinking that drives Altman. It's less about a literal target and more about laying down the foundation for AGI (Artificial General Intelligence), whether that means custom silicon, exotic new architectures, or something we haven't even seen yet. OpenAI clearly wants to find out. The biggest living proof of this is OpenAI's Texas data center, now the world's largest single facility, which consumes around 300 MW -- enough to power a mid-sized city -- and is set to hit 1 gigawatt by mid-2026. Such massive and unpredictable energy demands are already drawing scrutiny from Texas grid operators, who warn that stabilizing voltage and frequency for a site of this scale requires costly, rapid infrastructure upgrades that even state utilities struggle to match. Regardless, innovation must prevail, and the bubble shouldn't burst. The company isn't just hoarding NVIDIA hardware, either. While Microsoft's Azure remains its primary cloud backbone, OpenAI has partnered with Oracle to build its own data centers and is rumored to be exploring Google's TPU accelerators to diversify its compute stack. It's part of a larger arms race, where everyone from Meta to Amazon is building in-house AI chips and betting big on high-bandwidth memory (HBM) to keep these monster models fed. Altman, for his part, has hinted at OpenAI's own custom chip plans, which would make sense given the company's growing scale. Altman's comments also double as a not-so-subtle reminder of how quickly this field moves. A year ago, a company boasting 10,000 GPUs sounded like a heavyweight contender. Now, even 1 million feels like just another stepping stone toward something much bigger. OpenAI's infrastructure push isn't just about faster training or smoother model rollouts; it's about securing a long-term advantage in an industry where compute is the ultimate bottleneck. And, of course, Nvidia would be more than happy to provide the building blocks. Is 100 million GPUs realistic? Not today, not without breakthroughs in manufacturing, energy efficiency, and cost. But that's the point. Altman's vision isn't bound by what's available now but rather aimed at what's possible next. The 1 million GPUs coming online by year's end are a real catalyst for marking a new baseline for AI infrastructure, one that seems to be diversifying by the day. Everything beyond that is ambition, and if Altman's history is any guide, it might be foolish to dismiss it as mere hype.

Elon Musk: 1M Nvidia GPUs? Nah, My Supercomputers Need the Power of 50M

Elon Musk isn't stopping at acquiring 1 million Nvidia GPUs for AI training. The billionaire wants millions more as his startup xAI races to beat the competition on next-generation AI systems. Musk today tweeted that xAI aims for compute power that's on par with 50 million Nvidia H100 GPUs, the enterprise-grade graphics chip widely used for AI training and running chatbots. "The xAI goal is 50 million in units of H100 equivalent-AI compute (but much better power-efficiency) online within 5 years," he said. Musk's tweet comes a day after rival Sam Altman, the CEO of OpenAI, wrote in his own post about plans to run "well over 1 million GPUs by the end of this year," with the goal of exponentially scaling up the compute power by "100x." Meta CEO Mark Zuckerberg, meanwhile, has a similar goal; he wants mega data centers devoted to developing AI super intelligence. These growing AI investments underscore how expensive it is to scale up (and attract top talent). Musk's tweet doesn't mean he'll try to buy 50 million GPUs, though. The H100 was introduced in 2022 before Nvidia began offering more powerful models, including in the GB200, which can reportedly deliver an up to 2.5 times performance boost. Nvidia has also released a roadmap that outlines two additional GPU architectures, Rubin and Feynman, which promise to unleash more powerful AI chips in the coming years with improved power efficiency. Still, Musk's xAI will likely need to buy millions of Nvidia GPUs to reach his goal. In the meantime, Musk said in another tweet that xAI's Colossus supercomputer in Memphis, Tennessee, has grown to 230,000 GPUs, including 30,000 Nvidia GB200s. His company is also building a second Colossus data center that'll host 550,000 GPUs made up of Nvidia's GB200s and more advanced GB300 chips. This compute power requires enormous amounts of electricity; xAI is using gas turbines at the Colossus site, which environmental groups say are worsening the air pollution in Memphis.

Elon Musk says xAI is targeting 50 million 'H100 equivalent' AI GPUs in five years -- 230k GPUs, including 30k GB200s already reportedly operational for training Grok

Leading AI companies have been bragging about the number of GPUs they use or plan to use in the future. Just yesterday, OpenAI announced plans to build infrastructure to power two million GPUs, but now Elon Musk has revealed even more colossal plans: the equivalent of 50 million H100 GPUs to be deployed for AI use over the next five years. But while the number of H100 equivalents looks massive, the actual number of GPUs to be deployed may not be quite as great. Unlike the power they will consume. "The xAI goal is 50 million in units of H100 equivalent-AI compute (but much better power-efficiency) online within 5 years," Elon Musk wrote in an X post. One Nvidia H100 GPU can deliver around 1,000 FP16/BF16 TFLOPS for AI training (these are currently the most popular formats for AI training), so 50 million of such AI accelerators will have to deliver 50 FP16/BF16 ExaFLOPS for AI training by 2030. Based on the current performance improvement trends, this is more than achievable over the next five years. Assuming that Nvidia (and others) will continue to scale BF16/FP16 training performance of its GPUs at a pace slightly slower than with the Hopper and Blackwell generations, 50 BF16/FP16 ExaFLOPS will be achievable using 1.3 million GPUs in 2028 or 650,000 in 2029, based on our speculative guesses. If xAI has enough money to spend on Nvidia hardware, it is even possible that the goal of getting to 50 ExaFLOPS for AI training will be achieved even earlier. Elon Musk's xAI is already among the fastest companies to deploy the latest AI GPU accelerators to boost its training capability. The company already runs its Colossus 1 supercluster that uses 200,000 H100 and H200 accelerators based on the Hopper architecture, as well as 30,000 GB200 units based on the Blackwell architecture. In addition, the company aims to build its Colossus 2 cluster consisting of 550,000 GB200 and GB300 nodes (each of such nodes has two GPUs, so the cluster will feature over a million GPUs) with the first nodes set to come online in the coming weeks, according to Musk. Nvidia (and other companies) recently switched to a yearly cadence of new AI accelerators release and Nvidia's schedule now resembles Intel's Tick-Tock model from back in the day though in this case we are talking about an architecture -> optimization approach using a single production node (e.g., Blackwell -> Blackwell Ultra, Rubin -> Rubin Ultra) rather than switching to a new process technology for a known architecture. Such an approach ensures significant performance increases every year, which in turn ensures dramatic longer-term performance gains. For example, Nvidia claims its Blackwell B200 delivers 20,000 times higher inference performance than the 2016 Pascal P100, offering around 20,000 FP4 TFLOPS versus the P100's 19 FP16 TFLOPS. Though not a direct comparison, the metric is relevant for inference tasks. Blackwell is also 42,500 times more energy efficient than Pascal when measured by joules per generated token. Indeed, Nvidia and others are not slowing down with performance advancements. The Blackwell Ultra architecture (B300-series) offers a 50% higher FP4 performance (15 FPLOPS) compared to the original Blackwell GPUs (10 FPLOPS) for AI inference, as well as two times higher performance for BF16 and TF32 formats for AI training, yet at the cost of lower INT8, FP32, and FP64 performance. For reference, BF16 and FP16 are typical formats used for AI training (though FP8 seems to be evaluated as well), so it is reasonable to expect Nvidia to boost performance in these formats with its next-generation Rubin, Rubin Ultra, Feynman, and Feynman Ultra GPUs. Depending on how we count, Nvidia increased FP16/BF16 performance by 3.2 times with H100 (compared to A100), then by 2.4 times with B200 (compared to H100), and then by 2.2 times with B300 (compared to B200). Actual training performance of course depends not only on pure math performance of new GPUs, but also on memory bandwidth, model size, parallelism (software optimizations and interconnect performance), and usage of FP32 for accumulations. Yet, it is safe to say that Nvidia can double the training performance (with FP16/BF16 formats) of its GPUs with each new generation. Assuming that Nvidia can achieve the aforementioned performance increases with its four subsequent generations of AI accelerators based on the Rubin and Feynman architectures, it is easy to count that around 650,000 Feynman Ultra GPUs will be needed to get to around 50 BF16/FP16 ExaFLOPS sometime in 2029. But while Elon Musk's xAI and probably other AI leaders will probably get their 50 BF16/FP16 ExaFLOPS for AI training over the next four or five years, the big question is how much power will such a supercluster consume? And, how many nuclear power plants will be needed to feed one? One H100 AI accelerator consumes 700W, so 50 million of these processors will consume 35 gigawatts (GW), which is equal to the typical power generated by 35 nuclear power plants, making it unrealistic to power such a massive data center today. Even a cluster of Rubin Ultra will require around 9.37 GW, which is comparable to the power consumption of French Guiana. Assuming that the Feynman architecture doubles performance per watt for BF16/FP16 compared to the Rubin architecture (keep in mind that we are speculating), a 50 ExaFLOPS cluster will still need 4.685 GW, which is well beyond 1.4 GW - 1.96 GW required for xAI's Colossus 2 data center with around a million AI accelerators. Can Elon Musk's xAI get 4.685 GW of power to feed a 50 ExaFLOPS data center in 2028 - 2030? That is something that clearly remains to be seen.

Sam Altman's trillion-dollar AI vision starts with 100 million GPUs. Here's what that means for the future of ChatGPT (and you)

ChatGPT's CEO Sam Altman has a bold vision for the future of AI, something other big tech can't compete with: one powered by 100 million GPUs. That jaw-dropping number, casually mentioned on X just days after ChatGPT Agent launched as we await ChatGPT-5, is a glimpse into the scale of AI infrastructure that could transform everything from the speed of your chatbot to the stability of the global energy grid. Altman admitted the 100 million GPU goal might be a bit of a stretch -- he punctuated the comment with "lol" -- but make no mistake, OpenAI is already on track to surpass 1 million GPUs by the end of 2025. And the implications are enormous. For those unfamiliar, I'll start by explaining the GPU, or graphics processing unit. This is a specialized chip originally designed to render images and video. But in the world of AI, GPUs have become the powerhouse behind large language models (LLMs) like ChatGPT. Unlike CPUs (central processing units), which handle one task at a time very efficiently, GPUs are built to perform thousands of simple calculations simultaneously. That parallel processing ability makes them perfect for training and running AI models, which rely on massive amounts of data and mathematical operations. So, when OpenAI says it's using over a million GPUs, it's essentially saying it has a vast digital brain made up of high-performance processors, working together to generate text, analyze images, simulate voices and much more. To put it into perspective, 1 million GPUs already require enough energy to power a small city. Scaling that to 100 million could demand more than 75 gigawatts of power, around three-quarters of the entire UK power grid. It would also cost an estimated $3 trillion in hardware alone, not counting maintenance, cooling and data center expansion. This level of infrastructure would dwarf the current capacity of tech giants like Google, Amazon and Microsoft, and would likely reshape chip supply chains and energy markets in the process. While a trillion-dollar silicon empire might sound like insider industry information, it has very real consequences for consumers. OpenAI's aggressive scaling could unlock: In short, the more GPUs OpenAI adds, the more capable ChatGPT (and similar tools) can become. But there's a tradeoff: all this compute comes at a cost. Subscription prices could rise. Feature rollouts may stall if GPU supply can't keep pace. And environmental concerns around energy use and emissions will only grow louder. Altman's tweets arrive amid growing competition between OpenAI and rivals like Google DeepMind, Meta and Anthropic. All are vying for dominance in AI model performance, and all rely heavily on access to high-performance GPUs, mostly from Nvidia. OpenAI is reportedly exploring alternatives, including Google's TPUs, Oracle's cloud and potentially even custom chips. More than speed, this growth is about independence, control and the ability to scale models that could one day rival human reasoning. Whether OpenAI actually hits 100 million GPUs or not, it's clear the AI arms race is accelerating. For everyday users, that means smarter AI tools are on the horizon, but so are bigger questions about power, privacy, cost and sustainability. So the next time ChatGPT completes a task in seconds or holds a surprisingly humanlike conversation, remember: somewhere behind the scenes, thousands (maybe millions) of GPUs are firing up to make that possible and Sam Altman is already thinking about multiplying that by 100.

What does OpenAI want with 100 million GPUs? Altman just made the most expensive tech bet yet

OpenAI's expansion into Oracle and TPU shows growing impatience with current cloud limits OpenAI says it is on track to operate over one million GPUs by the end of 2025, a figure that already places it far ahead of rivals in terms of compute resources. Yet for company CEO Sam Altman, that milestone is merely a beginning, "We will cross well over 1 million GPUs brought online by the end of this year," he said. The comment, delivered with apparent levity, has nonetheless sparked serious discussion about the feasibility of deploying 100 million GPUs in the foreseeable future. To put this figure in perspective, Elon Musk's xAI runs Grok 4 on approximately 200,000 GPUs, which means OpenAI's planned million-unit scale is already five times that number. Scaling this to 100 million, however, would involve astronomical costs, estimated at around $3 trillion, and pose major challenges in manufacturing, power consumption, and physical deployment. "Very proud of the team but now they better get to work figuring out how to 100x that lol," Altman wrote. While Microsoft's Azure remains OpenAI's primary cloud platform, it has also partnered with Oracle and is reportedly exploring Google's TPU accelerators. This diversification reflects an industry-wide trend, with Meta, Amazon, and Google also moving toward in-house chips and greater reliance on high-bandwidth memory. SK Hynix is one of the companies likely to benefit from this expansion - as GPU demand rises, so does demand for HBM, a key component in AI training. According to a data center industry insider, "In some cases, the specifications of GPUs and HBMs...are determined by customers (like OpenAI)...configured according to customer requests." SK Hynix's performance has already seen strong growth, with forecasts suggesting a record-breaking operating profit in Q2 2025. OpenAI's collaboration with SK Group appears to be deepening. Chairman Chey Tae-won and CEO Kwak No-jung met with Altman recently, reportedly to strengthen their position in the AI infrastructure supply chain. The relationship builds on past events such as SK Telecom's AI competition with ChatGPT and participation in the MIT GenAI Impact Consortium. That said, OpenAI's rapid expansion has raised concerns about financial sustainability, with reports that SoftBank may be reconsidering its investment. If OpenAI's 100 million GPU goal materializes, it will require not just capital but major breakthroughs in compute efficiency, manufacturing capacity, and global energy infrastructure. For now, the goal seems aspirational, an audacious signal of intent rather than a practical roadmap.

xAI's significant expansion in AI computing capabilities sparks concerns over energy supply

Elon Musk's AI company, xAI, aims to exponentially increase its computing power to the equivalent of 50 million Nvidia H100 GPUs within five years. This ambitious plan raises concerns about the feasibility of managing the energy demands involved. Musk recently revealed that xAI's second Colossus compute cluster will soon launch with 550,000 Nvidia Blackwell AI accelerators, incorporating GB200 and GB300 chips. This development follows the existing Colossus 1 cluster, which operates over 200,000 H100 and H200 GPUs along with 30,000 GB200 units. According to Wccftech, it's estimated that the creation of Colossus 2 could cost up to US$2 trillion, marking an unprecedented investment in AI development and highlighting a significant increase in industry hardware spending. Tom's Hardware reports that 50 million H100 GPUs can achieve about 50 exaFLOPS for FP16/BF16 tasks. Considering efficiency improvements in new GPUs, xAI aims to use around 650,000 next-gen Feynman Ultra GPUs to reach its computing targets, significantly advancing AI hardware capabilities. Energy demand is rising as industries develop rival plans to address efficiency and sustainability. Businesses are competing to adopt innovative technologies and renewable resources, aiming to reduce costs and carbon footprints. This race is reshaping markets and influencing global energy strategies. The energy demand for 50 million H100 GPUs is expected to match the output of 35 nuclear power plants. Even with the efficient Feynman Ultra GPUs, xAI would still require approximately 4.7GW of power, significantly surpassing the 1.4 to 1.96GW used by the current Colossus 2 cluster. This gap underscores the difficulty of expanding AI computing power without straining energy resources. Key players in AI are preparing significant hardware expansions. OpenAI CEO Sam Altman aims to deploy one million GPUs by the end of 2025, with a long-term goal of up to 100 million. Meta CEO Mark Zuckerberg announced plans for AI data centers at a gigawatt scale, starting in 2026, with investments reaching hundreds of billions of dollars. Article edited by Jack Wu

'You're going to need more electricity than any human beings ever... Jensen, you're gonna have to explain that to me someday' says Trump to the Nvidia head honcho as he rolls out his mega-AI expansion plan

US President Donald Trump delivered the keynote speech at a summit in Washington DC yesterday entitled "Winning the AI Race", in which he discussed his newly-announced AI action plan. Addressing the crowd on a variety of topics, Trump singled out his apparent disbelief at the power requirements needed to build his AI vision of the future. "We wanna have very inexpensive electricity, so that you can power up the plants. You're going to need more electricity than any human beings ever in the history of the world" said Trump. "When I heard what you really need, I said you gotta be kidding me. "Double what we produce right now for everything, right? Jensen, you're gonna have to explain that to me someday, why they need so damn much. Couldn't you do it with a little bit less? My father always used to say, 'turn off the lights son.' But you guys are turning up the lights." President Trump's reaction at the level of power needed for AI data center expansion is well-warranted, if his estimate proves to be correct. According to the US Energy Information Administration, in 2023 the US produced 4,178 billion kWh of electricity, 60% of which came from fossil fuels, 21.4% from renewable sources like wind and solar power, and 18.6% from nuclear power plants. Doubling that figure would be a truly monumental task. It's not clear if Trump was referring to the sum total of what is needed for data centers alone, ie double the current US output just for AI expansion, or whether his estimates include existing US power requirements in the mix. Whichever, it's still a staggering amount of juice, and it's difficult to see how the US power grid could be expanded to meet that sort of demand. While the second point of Trump's AI plan includes a proviso for "creating new national initiatives to increase high-demand occupations like electricians", the US is going to need more than a few extra sparkies to double its output. Of course, President Trump has been known to speak in hyperbole, so perhaps this estimate should be taken with a pinch of salt. That being said, these new data centers will have to be powered by something, even if that doubled figure is an overstatement. Nvidia's new Blackwell AI GPUs can draw up to 1200 W each depending on the configuration, while the ever-popular H100 GPUs used in many data centers top out at 700 W a piece. Multiply that by the ever-increasing numbers hooking into the grid, and boy is that a lot of juice coming down the wires. Prior to his comments, Trump credited US secretary of the interior Doug Burgum with "producing low-cost energy" stating that "we're down to $64, I want to get it down a little bit further if we can... I don't know if the oil companies love that or not, but we wanna have very inexpensive electricity." US power demands were already expected to grow by 2% over the 2025-2027 period, and while renewable power plants continue to expand, it's difficult to see how huge amounts of extra capacity could be added to the system in a short period of time without a huge infrastructure push, and potentially the reliance on environmentally-unfriendly solutions like methane gas turbines and the restarting of coal power production to meet these new needs. That being said, clean energy expansion proposals put into place by the Biden administration appear to be bearing fruit. Wind and solar power is estimated to have overtaken coal electricity production in 2024, reaching a record 17% of US electricity production at its peak, while states like California and Nevada were said to have surpassed their 30% annual share of solar in their electricity mix in the same year. However, while this rate of expansion is admirable, US power companies continue to announce the closure or gas conversion of coal-powered plant facilities, so what could potentially fill the gap? Nuclear, perhaps? New York recently announced it would be building a the first major new US nuclear power plant in 15 years, so maybe this is merely the beginning of a new wave of nuclear facilities to power all these AI data demands. If I were a betting man, I'd put good money on some significant power infrastructure plans being announced following the emergence of this new AI push -- although it's still difficult to see how such a massive power generation expansion could keep up with the vast numbers of AI GPUs coming online in new facilities currently under construction by Meta, xAI, OpenAI, and others. In the meantime, if you're a US resident, it might be time to consider turning out the lights a little earlier. Every little helps, right?

OpenAI to have one million GPUs online by the end of the year, CEO Sam Altman wants 100 million

TL;DR: OpenAI CEO Sam Altman confirmed that the company will have over 1 million GPUs online by year-end, highlighting the massive growth of its AI infrastructure. While he joked that he'd like to see 100 million GPUs, this ambition underscores the industry's push for advanced AI hardware and the potential for in-house chip development. OpenAI CEO Sam Altman took to X to confirm that the AI firm will have over 1 million GPUs online by the end of the year, which is an impressive statistic to visualize. However, as you try to picture what 1 million cutting-edge GPUs looks like, Sam Altman added that he'd much rather see 100 million GPUs go online. After saying that he's "very proud of the team" for reaching the 1 million GPU milestone, he joked that they "better get to work figuring out how to 100x that." To put this figure into perspective, xAI's headline-grabbing Grok 4 model is powered by around 200,000 NVIDIA H100 GPUs, which suggests OpenAI is working with five times the GPU power as xAI. The 100 million GPU figure is not currently feasible, as earlier this year, Sam Altman announced that OpenAI was delaying the release of its GPT 4.5 model because it was "out of GPUs." Which is a good problem to have if you're NVIDIA, as companies like OpenAI, xAI, Microsoft, and others are buying up GPUs as quickly as they can be produced. Although the 100 million GPU figure is widely regarded as a joke, as it would require an incredible amount of power and physical space to be realized, this hasn't stopped people from speculating about the cost of just the hardware alone. According to The AI Investor on X, 100 million NVIDIA Blackwell GPUs for around $30,000 each would cost around $3 trillion - assuming NVIDIA doesn't give OpenAI a discount for buying so many at once. This insatiable desire for GPU hardware is one of the reasons companies like Oracle, Google, AMD, Microsoft, and others are creating their own AI chips. Although there hasn't been any announcement, Sam Altman has hinted that OpenAI could be on the path to developing its own AI chips. In a world where 1 million GPUs isn't enough and you're looking to scale that by a factor of 100X, this would make a lot of sense.

Elon Musk: 230K AI GPUs train Grok at Colossus 1: 550K GB200, GB300s at Colossus 2 coming soon

TL;DR: Elon Musk's xAI is investing up to $2 trillion to build Colossus 2, a supercomputer with 550,000 NVIDIA GB200 and GB300 AI GPUs, aiming for 50 million H100-equivalent GPUs and 200 exaFLOPs compute power within five years. This massive AI hardware expansion outpaces global supercomputers. The AI industry is reportedly preparing to spend "trillions of dollars" securing AI hardware, with Elon Musk's xAI planning to acquire the compute power equivalent to 50 million NVIDIA H100 AI GPUs. In a new post on X, Musk said: "230k GPUs, including 30k GB200s, are operational for training Grok @xAI in a single supercluster called Colossus 1 (inference is done by our cloud providers). At Colossus 2, the first batch of 550k GB200s & GB300s, also for training, start going online in a few weeks. As Jensen Huang has stated, @xAI is unmatched in speed. It's not even close". xAI's massive AI supercomputer cluster is called the Colossus 2, and it'll be online in the coming weeks, powered by NVIDIA GB200 and GB300 AI servers, with a total count of 550,000 units. Just on these numbers alone, that means xAI has spent around $2 trillion getting Colossus online, which is a colossal (pun intended) amount of money. Elon says that the goal is to get up to 50 million H100-equivalent AI compute power online in the next 5 years, where depending on the price -- and we're sure xAI is getting a discount from NVIDIA because of the mass AI hardware orders -- this would be worth somewhere between $1.5 trillion and $2 trillion. As for the AI compute power, we're talking about 200 exaFLOPs, which is 20x more compute power than the world's fastest supercomputer. Elon isn't the only one salivating over AI hardware, with OpenAI boss Sam Altman also announcing plans to secure over 100 million AI chips in the coming years, as well as Meta boss Mark Zuckerberg who is spending big money getting GW-scale AI clusters up and operational.

Thought the AI Hype Was Fading? Think Again; OpenAI's Sam Altman Signals to Buy Up to 100 Million AI Chips; A Move Potentially Worth Trillions

The demand for AI computing power isn't stopping at all, as Sam Altman reveals rather "shady" plans to acquire up to a million AI chips moving in the future. There's no stopping the AI train right now, since it is racing to new levels with each passing day. We have heard companies like Microsoft, Google, and Meta building up large-scale AI clusters, but despite that, organizations still report a lack of computing capabilities. OpenAI seems to be facing a similar situation, as Sam Altman claims that he has massive plans ahead, part of which includes acquiring a whopping one million AI chips. While the statement does seem a bit far-fetched, nothing seems impossible for OpenAI and its elite AI team. OpenAI's CEO is known for his hilarious investment figures. Just a few months ago, Altman was running around the world, raising trillions of dollars to build his network of chip facilities, but the project is nowhere to be seen for now. Even if, for some reason, OpenAI needs the power of 100 million AI chips, the firm needs trillions of dollars in capital on board, which is worth almost what NVIDIA is currently valued for. So, getting such a high count of AI chips seems impossible for now, but considering that GW-level AI clusters are starting to become a lot more common now, we cannot rule it out entirely. Interestingly, I decided to calculate the energy needed to power up 100 million AI GPUs. If we consider that each chip is rated to run at 750W, that puts it up to a 75 GW cluster, which constantly accounts for 75% of the UK's entire grid capacity. Unless, for some reason, OpenAI manages to get its hands on nuclear plants, it will still need 75 nuclear reactors, so good luck to Altman in scaling up towards a 100 million AI chip count. It seems like the industry thinks that racking up AI clusters would lead them to AGI, which is why the focus is towards getting a massive count of AI chips. There's no doubt that companies are involved in the race for AI infrastructure; with that, they are ready to spend hundreds of billions. The AI CapEx for Big Tech keeps growing, and it's safe to say that companies like NVIDIA are in for a treat.