What are the latest funding trends in AI data centers?

All funding deals in the AI data center market over the last years

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 really booming right now?

AI data center funding is booming in 2026, and the increase is much broader than a single giant round.

We count 25 financings in the current comparable period, up from 11 in 2025 and 13 in 2024. Twenty different companies raised money, twice as many as a year earlier. Audited capital reached $15.62 billion, compared with $2.42 billion in 2025 and $1.76 billion two years ago.

What makes this year unusual is that deal volume and round size jumped together. The median financing is now $250 million, more than twice the $115.5 million median last year and almost twelve times the $21.4 million level in 2024. Usually, a big increase in deal count brings more smaller transactions into the mix. Here, investors are funding more companies while also writing much larger checks.

By now, the AI data center market looks less like a niche venture category and more like a capital-intensive infrastructure market where nine-figure private rounds have become common.

Metric	2026 YTD	2025 comparable	2024 comparable	2026 vs 2025
Deals	25	11	13	+127%
Unique funded companies	20	10	13	+100%
Audited capital	$15.62B	$2.42B	$1.76B	+545%
Median round	$250M	$115.5M	$21.4M	+117%
Average round	$624.8M	$242.1M	$135.6M	+158%
Largest usable round	$3.90B	$1.10B	$1.10B	+255%
Companies raising multiple times	4	1	0	+300%

Is the AI data center funding boom just Crusoe, Nscale and a few huge rounds?

No. The biggest AI data center rounds are enormous, but the market is actually less dependent on its top financings than it was in either of the previous two periods.

Crusoe's $3.9 billion Series F, Nscale's $2 billion Series C and Firmus's $2 billion strategic equity financing obviously move the total. Yet the top five rounds account for 66.6% of current capital, down sharply from 87.2% last year and 93.9% in 2024.

The bigger change sits below those headline deals. Eighteen financings now exceed $100 million. We counted only five at that level in comparable 2025 and three in 2024.

This is why calling the current boom "megadeal-driven" misses part of the story. Megadeals are bigger than ever, but there are also far more companies capable of raising serious nine-figure capital.

Funding concentration	2026 YTD	2025 comparable	2024 comparable
Top 1 share of capital	25.0%	45.4%	62.4%
Top 3 share	50.6%	74.4%	86.6%
Top 5 share	66.6%	87.2%	93.9%
Rounds above $50M	20	8	5
Rounds above $100M	18	5	3
Median round	$250M	$115.5M	$21.4M
Audited capital	$15.62B	$2.42B	$1.76B

Would AI data center funding still look strong without the giant rounds?

Yes, although the underlying AI data center market is much stronger at the very top and the small end than in the middle.

Rounds of $50 million or less brought in $122.6 million this year, against $57 million during the same period of 2025. Smaller financings have therefore more than doubled in aggregate.

The picture changes when we include everything up to $100 million. That pool represents $272.6 million today versus $309 million last year. In other words, the $50 million to $100 million bracket has been relatively weak.

So the market has developed an odd shape. Small deals are still happening, and $100 million-plus rounds have exploded. The middle has not kept pace.

Are more AI data center companies actually getting funded?

Yes. More AI data center companies are getting funded today, so the boom cannot be explained by the same few companies repeatedly coming back for money.

The number of funded companies doubled from 10 to 20 year over year. Even with four companies completing multiple financings this year, the breadth increase remains obvious.

At the same time, funding is becoming deeper for the companies that investors already know. Four companies raised more than once in the current period, compared with one last year and none in comparable 2024.

Investor interest has spread across a bigger group of operators, but that does not mean a new generation of startups is flooding into the market. In fact, the opposite is happening.

Are new AI data center startups struggling to get funded?

Yes. New AI data center companies are barely showing up in this year's financing activity, even though overall funding has never been stronger in our comparison.

Only one of the 25 current financings qualifies as a first financing under our financing-history check. There were two in comparable 2025 and four in 2024. First financings now represent only 4% of deals, down from 18.2% last year and 30.8% two years ago.

Follow-on rounds dominate today's market. They account for 24 of the 25 transactions, and their median size has climbed to about $275 million from $131 million last year.

The financing-history check is deliberately lightweight rather than a perfect reconstruction of every company's capitalization since formation. Even with that caveat, the gap is too wide to dismiss. These days, most AI data center money is going to companies that have already proved they can raise institutional capital.

Financing history	2026 YTD	2025 comparable	2024 comparable
First financings	1	2	4
First-financing share	4.0%	18.2%	30.8%
Follow-ons	24	9	8
Follow-on share	96.0%	81.8%	61.5%
Usable first-financing capital	$8.8M	$12.0M plus Stargate amount excluded	$46.6M
Follow-on capital	$15.61B	$2.41B	$1.70B
Median follow-on	$275M	$131M	$65M

Has AI data center funding moved from startup creation to scaling?

Yes. AI data center funding is now much more about building out proven operators than creating new ones.

Early-stage transactions represented 69.2% of deals in 2024. Today that share is 32%. Later-stage and growth financings moved the other way, from 23.1% two years ago to 52% now.

Even "early stage" means something different in this market these days. Seven of the eight current seed or Series A transactions are follow-ons rather than first financings. A company can still carry a Series A label while already needing substantial capital for GPUs, data-center capacity and physical infrastructure.

Company age points in the same direction. The median company receiving funding was about 2.2 years old in 2024, roughly 3.0 years old last year and 3.6 years old now.

Investors are clearly more comfortable putting very large amounts behind companies that already have operating history, customers or infrastructure under development.

Why are AI data center funding rounds getting so huge?

AI data center rounds are getting huge because today's leading companies need to finance power, land, buildings, GPUs and cloud capacity at the same time.

Crusoe is a good example. Its latest Series F was announced at $3.9 billion, and the company said it had more than $140 billion in total contracted value across its platform. Days later, Crusoe disclosed a $65 million annual inference agreement with Thinking Machines Lab, followed by confirmation that it is developing Google's AI data center campus in Armstrong County, Texas. Those are useful recent signs that the financing is being followed by actual customer and construction activity.

Nscale's $2 billion Series C has a similar logic. The company says the money will accelerate a vertically integrated infrastructure platform spanning GPU compute, networking, data services and orchestration.

Firmus has gone even further into physical deployment. Its $2 billion strategic equity financing supports its AI Factory rollout, and the company now says it has more than 900 MW of contracted capacity across customers. Its plans include a 360 MW NVIDIA-based campus in Indonesia designed around roughly 170,000 GPUs.

The size of these rounds makes more sense once we look at what the companies are actually trying to build.

Which part of the AI data center market is hottest right now?

Integrated AI infrastructure operators are currently attracting the most attention by a wide margin.

They account for 16 of the 25 current financings and about $14.4 billion of capital. Their deal share jumped from 36.4% last year to 64% today, while their share of audited capital reached 92.2%.

The names also show that this is broader than a single funding story. Crusoe, Nscale, Firmus, Fluidstack, Neysa, Together AI, Groq, Gimlet Labs, Verda, Helix and GMI Cloud all raised capital in the current period.

AI compute data centers are growing too. We count six deals today versus three last year and two in 2024. Their relative share slipped slightly because integrated operators grew much faster.

Investors are increasingly backing companies that can connect several pieces of the stack rather than just provide a building or a narrow layer of compute.

Category	2026 deals	Deal share	2026 capital	Capital share	2025 deals
Integrated AI infrastructure operators	16	64.0%	$14.40B	92.2%	4
AI compute data centers	6	24.0%	$950.8M	6.1%	3
Modular AI data centers	2	8.0%	$270M	1.7%	1
Hyperscale AI campuses	1	4.0%	$1.75M	~0%	2
AI colocation facilities	0	0%	$0	0%	1
Total	25	100%	$15.62B	100%	11
Integrated-operator median round	-	-	$552.5M	-	-

Why are investors suddenly obsessed with full-stack AI infrastructure?

Investors are paying more attention to full-stack AI infrastructure because getting compute online quickly has become a business advantage of its own.

Crusoe now describes itself as vertically integrated from energy through AI cloud services. Nscale is explicitly building across GPU compute, networking, data services and orchestration. Firmus combines data-center infrastructure with NVIDIA systems and energy planning. Together AI sits further up the software stack but still raised $800 million while building a full-stack AI platform serving thousands of customers.

NVIDIA's latest moves make the shift even clearer. The company is working with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR on financing platforms intended to mobilize more than $500 billion for AI infrastructure over time. NVIDIA describes AI-factory compute as an emerging investable infrastructure asset rather than something customers have to finance project by project.

Compute capacity is increasingly being treated like infrastructure that can support long-duration institutional capital. That's a pretty big change in how this market gets financed.

Are hyperscale AI campuses falling out of favor?

Hyperscale AI campus fundraising has cooled inside this startup funding dataset, but demand for giant AI campuses clearly has not disappeared.

Comparable 2025 included Stargate and Fermi America, while the current period contains only the much smaller StratGrid financing in this category. Hyperscale-campus deal share consequently fell from 18.2% to 4%.

Stargate itself shows why we should be careful with that comparison. OpenAI initially described a $500 billion four-year infrastructure ambition rather than a conventional $45 billion financing round, which is why we excluded the database's $45 billion figure from capital calculations. Since then, OpenAI has continued adding capacity. It said in 2026 that Stargate had already passed its original 10 GW U.S. infrastructure target.

The activity has shifted more than it has vanished. Today, a larger part of the fundraising sits directly inside operators building and selling compute rather than separate headline campus vehicles.

Are strategic investors taking over AI data center funding?

Strategic investors are becoming much more important in AI data center funding, especially in the larger scaling rounds.

We identify strategic participation in 16 current financings, equal to 64% of deals. That compares with 54.5% last year and 46.2% in 2024.

NVIDIA is the clearest recurring example. It appears around Firmus, Nscale, Together AI, Crusoe, GMI Cloud and other infrastructure businesses. AMD Ventures has backed TensorWave, while Johnson Controls and Singtel Innov8 participated in Armada's $230 million Series B. Hardware, energy, telecom and industrial companies are increasingly investing alongside traditional financial investors.

The commercial logic is straightforward. Every successful new AI cloud or factory can become a major buyer of accelerators, servers, cooling, power equipment and connectivity.

As seen above, the median strategic-participation financing has also risen sharply, from $65 million in 2024 to $275 million today. Strategic capital is showing up where the infrastructure bill is largest.

Are AI data center investors becoming more selective?

Yes, but in a very specific way: more companies are getting funded, while new entrants are getting far less of the money.

If investors were simply retreating into a handful of winners, we would expect the number of funded companies to shrink. Instead, it doubled.

The filter appears later in the lifecycle. Investors are comfortable funding a growing set of operators once those companies have customers, infrastructure plans, strategic partners or earlier financing behind them.

Groq offers a useful recent example. The company raised $650 million in growth capital and then another $350 million a few weeks later as it expanded an inference cloud operating across 13 data centers. Firmus completed several financings while building contracted capacity across four countries. These are companies raising against deployment plans rather than PowerPoint-stage market entry.

For a new AI data center startup, today's market may therefore feel much tougher than the headline $15.62 billion total suggests.

Is North America still winning the AI data center funding race?

North America still has the broadest AI data center funding market, while Europe and Asia-Pacific are producing big totals from a much smaller group of companies.

North America accounts for 15 current deals and roughly $9.34 billion of capital. Its deal share rose from 45.5% last year to 60% today, with financing spread across companies including Crusoe, Groq, Together AI, TensorWave, Armada, Starcloud and others.

Europe raised around $3.03 billion through only four deals. Nscale and Fluidstack alone contribute about $2.83 billion, so Europe's total looks much broader in dollars than it really is in company count.

Asia-Pacific recorded six financings and about $3.25 billion. Firmus accounts for roughly $2.575 billion of that amount across three rounds, while Neysa contributes another $600 million of equity.

APAC does have a real capacity buildout behind the funding. NVIDIA recently said Australian infrastructure partners including Firmus, IREN, NEXTDC and AirTrunk are working toward as much as 2 GW of AI infrastructure capacity. Still, today's private funding total remains heavily concentrated around Firmus.

Region	Current deals	Deal share	Current capital	Capital share	Comparable 2025 deals	Deal-share change
North America	15	60.0%	$9.34B	59.8%	5	+14.5 pp
Asia-Pacific	6	24.0%	$3.25B	20.8%	2	+5.8 pp
Europe	4	16.0%	$3.03B	19.4%	4	−20.4 pp
North America top-level interpretation	Broad	-	Broad	-	Narrower	Improving
Europe top-level interpretation	Narrow	-	Large	-	Broader	Weaker breadth
APAC top-level interpretation	Moderate	-	Firmus-heavy	-	Narrower	Improving
Total	25	100%	$15.62B	100%	11	-

Is the AI data center investor base getting broader too?

Yes. The AI data center investor base has widened substantially alongside the rise in deal activity.

We count 25 different disclosed lead or co-lead investors in the current period, compared with 11 last year and 17 in 2024. Lead disclosure itself only improved slightly, so better reporting does not explain the jump.

8090 Industries is the only current investor that appears as lead or co-lead on more than one deal, with two appearances. There is no single VC or private-equity firm repeatedly leading a dominant share of the market.

That breadth fits what we see elsewhere: more funded companies, lower top-five capital concentration and more corporate participation. The financing ecosystem around AI infrastructure is getting larger along with the rounds themselves.

Has 2026 already beaten previous full years for AI data center funding?

Yes. AI data center funding in the current period has already moved well past the complete 2025 and 2024 totals on both audited capital and deal count.

Full-year 2025 produced 16 deals and about $6.67 billion of usable capital after applying the same treatment to Stargate. Full-year 2024 had 15 deals and approximately $1.93 billion.

The current period already stands at 25 deals and $15.62 billion. We are therefore at about 156% of the entire 2025 deal count and 234% of its usable capital before considering the rest of the year.

We would not annualize that figure. Infrastructure funding is too lumpy, and another multibillion-dollar round can distort a quarterly comparison instantly. This year's strength has already become too large to dismiss as timing noise.

What is the biggest change in AI data center funding right now?

The biggest change in AI data center funding is that raising $100 million or more has become normal for a much wider group of established operators.

There were only three $100 million-plus deals in comparable 2024. Today we count eighteen.

At the same time, those large rounds are spread widely enough that top-five capital concentration has dropped, the number of funded companies has doubled, integrated operators are gaining market share and strategic investors are appearing more often.

The financing market has effectively moved up a weight class. A $100 million AI infrastructure round no longer automatically looks exceptional; for companies trying to deploy GPUs and power at meaningful scale, it increasingly looks like ordinary expansion capital.

What could go wrong with the AI data center funding boom?

The biggest weakness in today's AI data center funding boom is the lack of new-company formation, followed by the enormous execution burden attached to these giant rounds.

One first financing out of 25 deals is thin. It means today's funding strength tells us much more about investors' willingness to scale existing operators than about the health of the next generation of entrants.

There is also a huge difference between raising billions and earning attractive returns on billions of deployed infrastructure. Power availability, utilization, customer concentration, GPU economics and construction schedules can all change the economics quickly.

The financing structures themselves are becoming harder to compare too. Crusoe describes its $3.9 billion transaction as the initial closing of an anticipated round. Firmus's $2 billion financing was announced as fully committed strategic equity. Stargate mixes equity funding, infrastructure commitments and a multiyear investment program. Treating every headline amount as identical cash-on-balance-sheet venture funding would make the market look cleaner than it really is.

For now, the funding evidence is exceptionally strong. The harder test comes next: whether all this new capacity earns enough money to justify the amount of capital going into it.

What is actually changing in AI data center funding?

AI data center funding is much stronger today, but the important change is deeper than the sixfold increase in audited capital.

More companies are raising money, so this is a genuine expansion in funding activity. At the same time, the market has moved sharply toward much larger rounds, with eighteen current financings above $100 million.

Existing operators are getting most of the benefit. First financings have almost disappeared from the current mix, while repeat issuers, older companies and later-stage businesses absorb nearly all the capital.

Integrated AI infrastructure companies are emerging as the clear funding center of gravity. Investors increasingly want operators that can connect power, facilities, accelerators and cloud services and then bring that capacity online quickly.

Strategic capital is also becoming part of the market's structure. NVIDIA, AMD, industrial groups, telecom companies and infrastructure players have direct economic reasons to help expand the companies that will buy and operate their technology.

And the latest operating evidence gives the boom more substance than fundraising announcements alone. Crusoe is signing large inference contracts and developing new campuses; Firmus says it has more than 900 MW contracted; Groq is scaling a 13-data-center inference footprint; and NVIDIA is helping create institutional financing platforms for AI-factory infrastructure.

The part we would watch most carefully is startup formation. Today's AI data center market looks extremely healthy if we judge it by capital available to proven operators. It looks much less open if we ask how many genuinely new companies are breaking into the funding cycle.