AI Data Center Spending Grew 60% Last Year, Could Hit $1T Next Decade, Intersect360 Says
The world spent more than $300 billion on AI data centers in 2025, representing a 60% increase over the prior year, according to the latest report from Intersect360 Research. While it expects the growth rate go slow in the years to come as AI business models shake out, Intersect360 still expects total AI data center spending to top more than $1 trillion by 2040.
To gauge market activity for this report, Intersect360 Research gathered data from two main sources: survey data from its HPC-AI Leadership Organization (HALO) members, as well as a separate survey of 458 enterprises around the world with revenues of at least $10 million. “We combine that…to get a total picture of the market,” CEO Addison Snell said during a webinar presenting the research last week.
Fig. 1: Total data center spending is projected to slow at the end of the decade (Image courtesy Intersect360 Research)
The research shows what most people in the industry already know: The AI boom is driving massive spending on everything AI-related. That includes data centers, racks to house servers, AI accelerators to speed training and inference, SSDs to store the data, networking to move the data, software to make the data useful, services to bring it all together, and enough power and water to keep it firing but not melting down.
The growth figures are down from 2024, which will likely be the high-water mark for one-year changes, until the next new new thing arrives. At the same time, it’s important to note that 60% is still a fairly sizable one-year increase, and $300 billion is serious money. “We talk about big numbers all the time in this industry,” said Intersect360 Research Analyst Antonia Maar during the webinar. “But I would say that’s still really significant growth.”
As seen in Fig. 1, the majority of the spending and the growth on AI data centers (which excludes spending on regular non-accelerated servers) came from the hyperscale AI segment, which includes industry giants like Google and Meta in the US, ByteDance and Alibaba out of China, as well as the neo-clouds, like Nebius and CoreWeave. Together, the hyperscale AI segment accounted for $216 billion in spending in 2025, up from about $120 billion or so in 2024.
The second big bucket is enterprise AI, which includes AI spending by private companies as well as spending on what could be called traditional HPC. Previously, Intersect360 listed these categories separately, but Snell decided to combine them for the 2025 analysis. The reason for the change is it better captures the underlying spending trends, as very few survey-takers are doing straight HPC anymore.
“When we take those 458 people…94% of them today have spending on HPC or AI or both. Usually it’s both,” he said. Fewer than 1% of respondents say they’re doing pure, straight HPC. “It’s a rare special project where you find HPC but not AI,” he added.
Fig 2: Enterprise AI spending by sub-category (Image courtesy Intersect360 Research)
Intersect360 is predicting growth to slow. Total data center market growth is forecast to be 7% CAGR through 2030, while the accelerated computing segment will be a bit faster, growing at an 11% CAGR pace through 2030. “What starts to change is hyperscale, not because spending stops and also not because AI goes away, but because at some point, even the largest organization starts to run into practical limits,” Maar said. “So we’re expecting hyperscale growth to moderate after 2027.”
When you step back from the AI boom and take a hard look at the inflows and the outflows, it’s clear that existing business models cannot support the status quo, Snell said.
“It’s not that we don’t believe in AI, but we do look at where the hyperscalers make their money and how much return you can conceivably get,” Snell said. “So if you’re a company like Meta investing in AI, where does that return come from? And if you’re Microsoft or Amazon and Google…most of their AI is still internal in its consumer model. And there’s a limit to how far that can go. So that’s why we look at this flattening. We don’t really have a big bubble pop where the market is going to crash. But we do think that there’s a kind of a reckoning by the end of this decade around how much goes into that.”
Intersect 360 predicts the Enterprise AI segment (which includes HPC) to grow at a 13% CAGR through 2030 (see Fig. 3). The vast majority of that growth will come from spending on AI and machine learning initiatives in the enterprise, and little of that growth will come from what some would traditionally think of as HPC. “HPC continues to grow in absolute terms, [but] it’s becoming a smaller share of the overall AI market simply because AI is growing so quickly,” Maar said.
Interestingly, Intersect360 sees the potential for the cloud segment of the market continuing to grow, while on-prem begins to stagnate. However, it adds a caveat. “This trend is not certain,” the company said in its report. “Users would prefer to keep hybrid with on-prem for cost and data sovereignty, but supplier trends favor hyperscale and cloud.”
Fig. 3: Cloud grows to dominate the enterprise AI segment (Image courtesy Intersect360 Research)
Some of the 2025 concerns over cutbacks in government spending on scientific computing turned out to be not as bad as expected, Snell said. “I think I overreacted to a lot of the announcements early in the Trump administration, which is where we were a year ago, and thought that it would drive some more changes in spending than it actually drove,” he said. “And we’re looking ahead toward more public private partnerships that will continue to do national sovereign AI going forward. So it is a high growth segment, but the nature of that growth kind of changed a little bit in terms of where it is.
In terms of systems, the high-end of the enterprise AI market (including HPC) will do better than the entry-level and midrange parts of the enterprise AI market, Intersect360 said. The company sees total enterprise AI system spending rising in 2026 to about $27.5 billion, up from about $22.5 billion in 2025. The company sees spending on enterprise AI systems topping out at a hair over $40 billion in 2029, before declining in 2030, which the company says is “a combination of a slowdown in AI spending, combined with the long-term trend favoring cloud over on-prem computing.”
There are some significant headwinds limiting long-term growth, according to Intersect360, including power generation, geopolitical stability, data sovereignty, and the technical requirements of consumer-grade AI and its impact on scientific computing (namely, is FP8 all we need?) The emergence of quantum computing, which Intersect360 says could be an $80 billion-per-year business by 2040, could also impact this market, just as the growth of cloud computing currently is.
While Intersect360 is predicting $1 trillion in spending by 2040, it will have to go through some AI doldrums before it gets to the promised land.
“We think this lull in 2029, 2030, 2031 is exactly that,” Snell said. “It’s a lull, just like we saw with the adoption of the World Wide Web. As we went into the dot-com era, it didn’t stop the Internet, right? There was a period of big, enthusiastic investment and then a right-sizing of expectations and where it fits business models, followed by long-term, sustainable growth. And I would expect that we see the same pattern here with respect to AI.”
The Intersect360 team is attending ISC26 in Hamburg and will be discussing these findings, and among other topics, throughout the week.
Related
Broadcom Announces VMware Cloud Foundation 9.1
PALO ALTO, Calif., May 5, 2026 — Broadcom Inc., a global technology leader that designs,…
NVIDIA Announces Financial Results for 1st Quarter Fiscal 2027
SANTA CLARA, Calif., May 21, 2026 — NVIDIA (NASDAQ: NVDA) has reported record revenue for…
Anthropic Unveils ‘Project Glasswing’ as Claude Mythos Targets Software Vulnerabilities
April 9, 2026 — Anthropic has announced Project Glasswing, a new initiative that brings together Amazon…
Google Unveils Gemini Enterprise Agent Platform, Expands Vertex AI into Full Agent Stack
At Google Cloud Next 2026 in Las Vegas this week, the company announced major changes…
OpenAI Shutters Sora, Shifts Business Strategy Ahead of IPO
Back in 2022, OpenAI set off a chain reaction in the tech world when it…
Cerebras Systems Announces Launch of Initial Public Offering
SUNNYVALE, Calif., May 4, 2026 — Cerebras Systems, Inc. today announced that it plans to commence…
NVIDIA Announces BioNeMo Agent Toolkit — Tools for Agents to Accelerate Scientific Discovery
SAN DIEGO, June 23, 2026 — NVIDIA today announced NVIDIA BioNeMo Agent Toolkit, which provides…
Multiverse Computing Launches Pulsar 16B in Collaboration with NVIDIA
The new open reasoning model delivers 30B-class intelligence in a 16B-parameter footprint, with 3.1B active…
AI Data Center Spending Grew 60% Last Year, Could Hit $1T Next Decade, Intersect360 Says
The world spent more than $300 billion on AI data centers in 2025, representing a…
Five Things to Watch as ISC 2026 Kicks Off in Hamburg
he International Supercomputer Conference is upon us, featuring four days of HPC education, collaboration, and…
Upbound Launches Modelplane: The Open Source Control Plane for AI Inference
SAN FRANCISCO, June 23, 2026 — Upbound, the company behind Crossplane, today released Modelplane, an…
NVIDIA Announces Halos for Robotics, the Industry’s 1st Full-Stack Safety System for Physical AI
CHICAGO, June 23, 2026 — NVIDIA has announced NVIDIA Halos for Robotics, the industry’s first…
