What’s funding like in the AI data center market right now?

All funding deals in the AI data center market over the last 24 months

Below is the table listing all the deals. You can find our methodology at the end of this page.

If you want a deeper understanding of the market and its current dynamics, get our report covering the AI data center market.

Is AI data center funding actually healthier now?

AI data center funding looks much healthier today on deal activity, company breadth and typical round size, even though total recorded capital is down 58%.

We counted 30 financings across 23 companies in the latest 12 months, up from 13 deals across 11 companies in the previous period. The median round jumped from $131 million to $312.5 million, and there was at least one financing in every month.

The weakness is at the entry point. Only 10% of recent deals were first financings, down from 15.4%, while follow-ons reached 90%. Seed activity also fell from two deals to one even as the whole market more than doubled in size by deal count.

So the market is healthier in the sense that many more companies can raise serious money. It is also more mature: most of that money is going into companies that investors already know.

Metric	Current L12M	Previous 12M	Change
Deals	30	13	+131%
Unique companies	23	11	+109%
Total capital	$19.87B	$47.58B	-58.2%
Median round	$312.5M	$131.0M	+138.5%
Average deals/month	2.50	1.08	+131%
First-financing share	10.0%	15.4%	-5.4 pp
Follow-on share	90.0%	84.6%	+5.4 pp
Later-stage deal share	53.3%	46.2%	+7.2 pp

Did AI data center funding really fall 58%?

The 58% drop in AI data center funding is technically correct, but it gives a terrible picture of what actually happened in the market.

The previous 12 months contained Stargate's $45 billion partner-equity commitment. That one transaction represented 94.6% of all capital in the period. OpenAI described Stargate as a much broader $500 billion four-year infrastructure program, with $100 billion expected to begin deploying immediately, so the database figure is already an unusual financing compared with a normal venture round.

Look below that outlier and today's market looks completely different. Capital in rounds of $50 million or less rose from $64.6 million to $172.6 million, a 167% increase. Capital in rounds of $100 million or less was essentially unchanged at $322.6 million versus $316.6 million. At the same time, the typical financing became much larger.

Someone looking only at aggregate dollars would therefore conclude that AI data center financing collapsed. Once Stargate stops dominating the comparison, that conclusion does not hold up.

Is today's AI data center boom still driven by a few giant rounds?

AI data center funding is far less dependent on one giant deal now, even though billion-dollar rounds have become remarkably common.

The largest recent financing, Crusoe's anticipated $3.9 billion Series F, represents 19.6% of current capital. A year earlier, Stargate represented 94.6%. The top five transactions now account for 54.9% of funding compared with 98.7% previously.

The more interesting change is the number of large deals underneath the leader. We counted 22 rounds above $100 million in the latest period, compared with seven previously. Crusoe, Firmus, Lambda, Fluidstack, Nscale, Together AI, TensorWave and several others have all reached nine- or ten-figure financing territory.

As seen above, this is now a much deeper financing market. Large checks are spread across a growing cohort rather than being almost entirely explained by one project.

Concentration	Current L12M	Previous L12M
Largest deal / total capital	19.6%	94.6%
Top 3 deals	39.8%	97.9%
Top 5 deals	54.9%	98.7%
Top 10 deals	80.0%	99.9%
Rounds above $50M	24	10
Rounds above $100M	22	7
$100M+ deal share	73.3%	53.8%
Capital in rounds of $100M or less	$322.6M	$316.6M

Are more AI data center companies actually getting funded now?

Yes, AI data center funding is reaching a much wider group of companies now.

We found 23 funded companies in the latest 12 months compared with 11 previously. That increase is almost as large as the rise in deal count, which went from 13 to 30.

The mix is also broader than a simple list of hyperscale campus developers. Recent financings span full-stack operators such as Crusoe and Firmus, GPU-cloud businesses such as TensorWave, modular infrastructure through Armada and Rune, European platforms such as Verda, and even Starcloud's orbital data-center model.

That breadth is one of the clearest improvements in the market. More companies have managed to turn AI-compute demand into something investors consider financeable at scale.

Are investors still funding new AI data center startups?

New AI data center startups are still getting funded, but the current boom is overwhelmingly about scaling companies that already exist.

Only three recent transactions were first financings. The absolute number increased from two, yet first financings fell from 15.4% to 10% of deal activity because everything else grew much faster.

Follow-on capital tells the story more clearly. Existing companies raised $18.81 billion in the latest period versus $2.57 billion previously, and the median follow-on increased from $131 million to $324.9 million.

The entry point therefore looks relatively thin these days. Investors are putting enormous amounts behind the cohort that has already made it through the first stages of company formation, while the flow of freshly funded businesses has barely moved.

New entrants vs existing companies	Current L12M	Previous L12M
First financings	3	2
First-financing share	10.0%	15.4%
Follow-on financings	27	11
Follow-on share	90.0%	84.6%
Follow-on capital	$18.81B	$2.57B
Median follow-on	$324.9M	$131M
Companies with multiple financings	5	2
Repeat-fundraising rate	21.7%	18.2%

Have $100 million AI data center rounds become normal?

Yes. In AI data center funding today, a $100 million round barely qualifies as unusually large anymore.

Twenty-two of the 30 recent deals cleared $100 million. That means 73.3% of the market by transaction count sits in nine-digit territory.

Meanwhile, the conventional $5 million to $20 million venture round almost disappeared, falling from 15.4% to 3.3% of deals. Even Series A rounds had a $125 million median in the latest period.

The economics explain a lot of this. These companies are buying GPUs, securing power, fitting out high-density sites, building cooling systems, manufacturing modular data centers and committing to large capacity deployments. A financing that would look enormous for a software startup can be relatively ordinary when the product involves hundreds of megawatts of physical infrastructure.

Which AI data center business model is getting the most investor attention?

Vertically integrated AI infrastructure operators are winning the funding race right now.

Their share of deals rose from 38.5% to 63.3%, while their median financing increased from $220.5 million to $600 million. They captured $17.6 billion, or 88.6% of all capital in the latest period.

Crusoe is perhaps the clearest example. The company combines energy, data-center development and AI cloud services, and its latest financing followed more than $140 billion in reported contracted value across its platform. Firmus combines AI factories, proprietary infrastructure and cloud capacity. Nscale describes its model as spanning power, data centers, GPU compute, networking and orchestration. Verda's latest $189 million financing similarly supports a full-stack AI cloud rather than a narrow infrastructure layer.

Investors increasingly seem to prefer companies that can control more of the deployment chain. With power, GPUs, networking and customer delivery all potential bottlenecks, owning more of the stack gives an operator more ways to keep projects moving.

Category	Current deal share	Current median	Previous deal share	Previous median
Integrated AI Infrastructure Operators	63.3%	$600M	38.5%	$220.5M
AI Compute Data Centers	20.0%	$160M	30.8%	$28.5M
Hyperscale AI Campuses	10.0%	$50M	15.4%	$22.55B
Modular AI Data Centers	6.7%	$135M	7.7%	$131M
AI Colocation Facilities	0%	-	7.7%	$52M
Integrated operator capital share	88.6%	-	Lower	-
Overall median financing	-	$312.5M	-	$131M

Is AI data center funding moving later-stage?

Yes, AI data center funding has moved further into the scaling phase over the last year.

Early-stage deals fell from 38.5% to 30% of transactions while later-stage deals rose from 46.2% to 53.3%. Seed was especially quiet, with only one deal.

Series B went the other way, growing from one financing to six. Series C reached a $1.4 billion median, while Series D+ reached $1.5 billion.

This looks like a cohort graduating. A meaningful number of AI infrastructure companies that were proving a concept a few years ago are now financing actual deployments, larger sites and commercial expansion.

The shift also explains why overall company age barely changed while round sizes exploded. Investors do not simply seem to be financing much older businesses; they are writing much bigger checks to companies that have reached a more capital-intensive phase.

Do AI data center companies need more proof to raise big rounds now?

For the biggest AI data center rounds, proof of deployment and customer demand now looks almost as important as the technology itself.

We cannot build a clean market-wide milestone comparison because public disclosure is too inconsistent. But the companies attracting the biggest checks increasingly arrive with something tangible behind the story.

Crusoe reported more than $140 billion in contracted value when it announced its anticipated $3.9 billion Series F. Two days earlier, it had announced a multi-year agreement with Perplexity covering both frontier-model training and production inference.

Firmus has gone even further into industrial-scale proof. Its recent financing cycle sits alongside large Australian deployments, a 360 MW campus in Indonesia built around as many as 170,000 NVIDIA accelerators, and new GPU-compute agreements with Meta across Asia-Pacific.

That is a much higher bar than proving that a GPU cloud can technically work. The large-round market increasingly rewards companies that can show customers, hardware access, power and credible deployment capacity at the same time.

What are AI data center companies actually raising all this money for?

AI data center funding is increasingly paying for physical rollout: more power, more GPUs, more facilities and more manufacturing capacity.

Lambda said its more-than-$1.5 billion Series E would accelerate gigawatt-scale AI factories and supercomputers. Armada paired a $230 million Series B with plans for a dedicated Arizona manufacturing facility for its modular Galleon data centers.

Firmus said its $2 billion strategic equity round would accelerate Project Southgate in Australia and prepare expansion into other Asia-Pacific markets. Crusoe said its latest capital would support AI factories ranging from large campuses to modular units while expanding Crusoe Cloud.

Nscale's $2 billion Series C has a similar flavor, with the company explicitly targeting vertically integrated deployments across Europe, North America and Asia.

The practical center of the market has moved well beyond R&D. These days, the expensive problems are deployment problems.

Are strategic investors taking over AI data center funding?

Strategic investors are becoming much more influential in AI data center financing, especially where hardware and deployment relationships matter.

Strategic participation rose from 46.2% of deals to 66.7%. More strikingly, it now appears across Series A, Series B, Series C, later-stage and growth financings rather than being confined to a few mature companies.

NVIDIA appears repeatedly because access to accelerated computing sits at the center of these businesses. AMD Ventures backed TensorWave, whose infrastructure is built around AMD Instinct GPUs. Johnson Controls invested alongside an agreement to manufacture Armada's modular systems. Samsung committed $1 billion to Helix while bringing semiconductor, construction, cloud and energy capabilities into the relationship.

A corporate investor can bring much more than cash in this market. Hardware supply, networking, manufacturing, site development and customer access can all become part of the same financing relationship, which helps explain why strategic participation has climbed so quickly.

Is Asia-Pacific becoming a serious AI data center funding hub?

Yes, Asia-Pacific has become one of the clearest areas of new momentum in AI data center funding.

The region's deal share increased from 15.4% to 26.7%, and the median transaction reached roughly $415 million. Australia alone contributed around $2.9 billion to the latest period.

Firmus is the standout example, but the regional story goes beyond one company. HyperVault brought a large Indian infrastructure commitment into the dataset, Nava is targeting India and Southeast Asia, and Japanese operators are also building sovereign AI-compute platforms.

The newest company announcements make the trend harder to dismiss as a temporary financing burst. Firmus has since announced a 360 MW NVIDIA-linked campus in Indonesia and GPU-compute agreements with Meta across Southeast Asia.

North America still leads the market, but AI data center capital is increasingly following power availability, sovereign-compute ambitions and customer demand across Asia-Pacific.

Is Europe falling behind in AI data center funding?

Europe is getting fewer AI data center deals, but the companies that do raise are attracting much larger rounds than before.

Europe's deal share fell to 13.3% from 46.2%. Its median financing, however, rose from $48.5 million to roughly $509.5 million.

Nscale captures the change well. Its $2 billion Series C is funding full-stack infrastructure across Europe, North America and Asia, and the company subsequently announced another $790 million financing package for its Narvik data-center buildout in Norway. Verda's recent $189 million financing gives Europe another sizeable full-stack AI cloud platform.

The European picture is more concentrated, but it does not look starved of capital. Money is clustering around a smaller number of companies trying to operate at hyperscaler-like scale.

Are AI data center investors becoming more selective?

AI data center investors look more demanding about company maturity, even while they are funding far more companies overall.

We found roughly twice as many disclosed lead investors as in the previous period, and unique funded companies also doubled. Capital has clearly not shut down.

Yet the market's center of gravity moved toward follow-ons, Series B and later-stage rounds. Seed barely participated in the boom, and first-time financings lost share.

The bar has moved higher, rather than investor appetite moving lower. Investors are paying much more attention to AI infrastructure, but the largest checks increasingly go to companies that can show they have somewhere productive to put hundreds of millions of dollars.

Has AI data center funding become more consistent through the year?

Yes, AI data center fundraising now happens much more consistently instead of arriving in a few isolated bursts.

The latest period had no zero-deal months, compared with two previously. Median monthly deal count doubled from one to two, and median monthly capital rose from $115.5 million to about $895 million.

September was particularly strong, with seven deals and roughly $7.15 billion. But funding was already active across the rest of the year through companies such as Lambda, Firmus, Fluidstack, Nscale, Armada, TensorWave and Together AI.

The difference with the previous period is straightforward: Stargate created one enormous month, while the current market produces large transactions again and again.

Are AI data center companies raising money faster than before?

We can say that repeat fundraising is more common in AI data centers now, but we cannot honestly claim that the whole market is raising faster.

Five current-period companies raised more than once, compared with two previously, and the repeat-fundraising rate increased from 18.2% to 21.7%.

Firmus, Crusoe, Fluidstack and Starcloud all show how quickly capital can come back to companies with large deployment plans. Fluidstack is a useful reminder to be careful, though: its $830 million Series A was publicly announced in July even though the company says it actually closed in January, so counting both dates would create a fake extra transaction.

A proper funding-velocity statistic requires reliable previous-round dates for the entire sample, and public records do not provide that consistently enough. We therefore stop at the defensible conclusion: repeat financing broadened, while the exact speed of market-wide fundraising remains uncertain.

Has AI data center funding become more like infrastructure finance than venture capital?

Yes, today's AI data center funding increasingly looks like industrial infrastructure finance delivered through startup equity.

The clearest clue is scale. Nearly three-quarters of recent transactions exceeded $100 million, the median round reached $312.5 million, and strategic investors participated in roughly two-thirds of deals.

Then look at what companies are actually financing: gigawatt-scale facilities, hundreds of thousands of accelerators, manufacturing lines, energy agreements, high-density cooling and multi-region deployments. Those are industrial problems as much as startup problems.

The companies are still venture-backed and their technology, execution speed and market position matter enormously. But the capital intensity increasingly resembles infrastructure development.

That hybrid model helps explain why full-stack operators are doing so well right now. The company that can coordinate capital, power, hardware and customers has a much better chance of actually getting capacity online.

What really changed in AI data center funding over the last 12 months?

AI data center funding became broader, more mature and much more industrial over the last 12 months, while headline capital fell for a reason that says very little about current market health.

The biggest change is breadth. We found more than twice as many financings and more than twice as many funded companies, with activity spread through every month rather than depending on one extraordinary announcement.

The second change is the normalization of very large rounds. Nine-digit financings now dominate the market, while integrated infrastructure operators have become the main destination for capital.

The third change is who benefits. Existing companies captured the overwhelming majority of recent financing, while first-time funding and seed activity failed to keep up. AI data centers currently have a much stronger scale-up engine than startup-formation engine.

Strategic investors also have a bigger role because funding increasingly sits alongside GPU supply, networking, manufacturing, energy and customer agreements. The companies that can coordinate those pieces are attracting the largest checks.

Finally, total funding is the most misleading statistic in this comparison. Stargate made the previous period look vastly bigger than the current one, but underneath that $45 billion commitment we find a market that is currently financing more companies, writing much larger typical checks and supporting a substantially deeper group of scaled operators.