AI data center funding tracker (57 deals)

Our funding tracker covering the AI data center market

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

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

Is the AI data center market really attracting $69 billion?

AI data center companies have attracted $69.22 billion in announced equity financing and commitments, but one enormous project makes the market look almost three times bigger than its underlying funding activity.

We identified 56 distinct financings across 38 companies. The Stargate Project alone accounts for $45 billion, or 65% of the total. Remove that transaction and the market comes down to $24.22 billion.

That remaining amount is still remarkable. Crusoe raised $5.275 billion during the period, Firmus $3.3 billion, and Nscale $3.255 billion. Fluidstack and Lambda also received more than $2 billion each.

Stargate deserves special treatment because its financing is different from a normal startup round. When OpenAI and SoftBank announced the project in January 2025, they described a plan to invest $500 billion in American AI infrastructure, beginning with $100 billion in deployment. The announcement named OpenAI, SoftBank, Oracle and MGX as initial equity funders but did not establish that $45 billion had already been paid into the company.

We retain the $45 billion reported commitment in our calculations while treating it separately whenever it could distort the conclusions. The more useful starting point for understanding ordinary AI data center financing is the $24.22 billion raised or committed outside Stargate.

How many AI data center startups are actually getting funded?

Only 38 companies received funding in the AI data center market during our research window, and most of them appeared just once.

Twenty-seven companies had one recorded financing, seven had two, and four had at least three. Altogether, 11 companies raised money more than once.

The companies returning to investors are especially important. Firmus completed six recorded financings, while Lambda and Nscale each completed three. Crusoe and Fluidstack raised enormous amounts through two transactions apiece.

We calculated that repeat fundraisers received $19.60 billion, compared with approximately $4.62 billion for single-round companies after excluding Stargate.

Part of the explanation is the nature of the business. Opening an AI cloud service is one investment decision; acquiring additional GPUs, securing megawatts of power and building new computing facilities requires another. The companies that can demonstrate progress have good reasons to return for significantly larger checks.

One caution is that several businesses entered the database late in the period. Their absence from subsequent funding announcements doesn't necessarily mean they failed to attract investors again.

Is most AI data center funding going to just a few companies?

AI data center investment is heavily concentrated among a few operators, with the ten largest funded companies capturing 94.6% of the recorded capital.

Stargate is the obvious outlier, but the concentration extends well beyond it. Crusoe, Firmus, Nscale, Fluidstack and Lambda collectively received more than $16 billion.

The largest individual rounds tell a similar story. Crusoe announced an initial closing of a $3.9 billion Series F in September 2026, following a $1.375 billion Series E in 2025. Firmus secured a $2 billion strategic equity investment in August 2026, while Nscale announced a $2 billion Series C in March.

These companies operate at a scale where financing a single expansion can cost more than the lifetime funding of dozens of smaller technology companies.

The concentration also shows which businesses investors are willing to back repeatedly. Established operators with access to computing hardware, infrastructure sites and customers have received far more money than companies still trying to establish those capabilities.

Company	Capital raised in window	Deals
The Stargate Project	$45.00B	1
Crusoe	$5.275B	2
Firmus	$3.300B	6
Nscale	$3.255B	3
Fluidstack	$2.330B	2
Lambda	$2.300B	3
CoreWeave	$1.100B	1
HyperVault	$1.000B	1
Helix Digital Infrastructure	$1.000B	1
Together AI	$906M	2

How much money does a typical AI data center startup raise?

The median AI data center financing is already $162.5 million, making this one of those markets where even a relatively ordinary round can involve nine figures.

The average is much higher, at $1.24 billion, but that number is heavily distorted by Stargate. The median gives a better idea of what the companies in our dataset are actually raising.

At company level, median cumulative financing is $125 million. That includes young businesses with only one round, established companies raising expansion capital and large infrastructure vehicles.

The variation is enormous. Starcloud began with a $2.4 million Seed round in March 2024. Gimlet Labs announced an $80 million Series A in March 2026, followed by a $300 million Series B in September. Fluidstack disclosed an $830 million Series A originally raised in January 2026.

So a founder looking at this market needs to distinguish between financing the initial business and financing the infrastructure required to serve large customers. The second can easily require hundreds of millions before the company reaches the age normally associated with late-stage venture funding.

Are investors still writing small checks for AI data center startups?

Small AI data center funding rounds still happen, but they account for a tiny fraction of the money flowing into the industry.

Only four recorded transactions were below $5 million. Another four fell between $5 million and $20 million. At the opposite end, 34 rounds reached at least $100 million, representing 60.7% of all financings.

Lumino's $2.8 million Seed, Starcloud's $2.4 million Seed and HIGHRESO's approximately $471,000 financing show that a company can enter the market with relatively modest capital. These examples are far removed from the financing requirements of Crusoe, Nscale or Lambda.

The economics change when a business needs to purchase substantial GPU capacity, prepare data center facilities or guarantee infrastructure availability to large customers.

We find that rounds worth $50 million or less contributed only about $344 million across the entire period. The bigger transactions accounted for nearly all the recorded funding.

Round size	Number of deals	Share of deals	Capital raised
Below $5M	4	7.1%	$7.4M
$5M–<$20M	4	7.1%	$38.4M
$20M–<$50M	7	12.5%	$198.4M
$50M–<$100M	7	12.5%	$432.0M
$100M+	34	60.7%	$68.54B
All rounds	56	100%	$69.22B

Which AI data center businesses are getting the most money?

Integrated AI infrastructure operators are attracting the broadest funding, with 34 transactions across 20 companies totaling $21.48 billion.

These businesses combine GPU computing capacity with services such as cloud provisioning, networking, storage and infrastructure operations. Crusoe, Lambda, Fluidstack, Nscale and Firmus are important examples.

Hyperscale AI Campuses technically lead the capital ranking at $46.15 billion, but Stargate supplies approximately 97.5% of that amount. Outside Stargate, the category recorded only four other financings worth around $1.15 billion.

The smaller AI Compute Data Centers category looks quite different. It includes nine companies and 12 rounds, with a median financing of just $33.5 million. TensorWave's $350 million Series B is its largest transaction.

Modular AI Data Centers received $411 million, much of it going to Armada, while AI Colocation Facilities recorded a single $52 million deal.

The category ranking makes the preference pretty clear: companies able to offer a complete infrastructure platform are absorbing substantially more capital than businesses focused on individual layers of AI computing.

AI data center category	Deals	Companies	Total funding	Median deal
Hyperscale AI Campuses	5	5	$46.15B	$100M
Integrated AI Infrastructure Operators	34	20	$21.48B	$310M
AI Compute Data Centers	12	9	$1.12B	$33.5M
Modular AI Data Centers	4	3	$411M	$85.5M
AI Colocation Facilities	1	1	$52M	$52M

Why are investors giving billions to AI infrastructure operators?

AI infrastructure operators are getting billion-dollar financings because building enough computing capacity for major AI customers requires enormous spending on equipment, electricity, buildings and operations.

Crusoe is a particularly clear example. Its September 2026 Series F announcement described a vertically integrated platform covering energy generation, AI infrastructure and cloud services. The company also reported more than $140 billion in total contracted value across its platform. That figure represents contracted value, rather than revenue already earned, but it helps explain why investors were prepared to support such an expensive expansion.

Nscale has a comparable ambition. Its March 2026 Series C was intended to accelerate infrastructure deployment across GPU computing, networking, data services and orchestration software.

Lambda's financing history also shows how quickly requirements can grow. After raising $320 million in February 2024 and $480 million in February 2025, the company announced more than $1.5 billion in additional funding in November 2025.

Aker's first-quarter 2026 shareholder letter describes the physical constraints particularly well: getting enough GPU clusters, specialized networking, electricity and cooling capacity has become a major challenge for AI infrastructure businesses.

Investors financing these operators are betting on their ability to secure and deploy those resources efficiently, often with substantial customer demand already contracted or anticipated.

Is Stargate actually comparable to other AI data center startups?

Stargate operates on such a different scale that we wouldn't use its financing as a benchmark for a normal AI data center startup.

The project brings together OpenAI, SoftBank, Oracle and MGX around a major infrastructure investment program. SoftBank has financial responsibility, while OpenAI has operational responsibility.

The scale of construction is equally unusual. In July 2025, OpenAI announced an agreement with Oracle to develop another 4.5 gigawatts of American data center capacity. By April 2026, OpenAI said it had secured more than 10 gigawatts of planned US AI infrastructure capacity, already surpassing its original target.

Those announcements explain why Stargate appears so prominently in funding statistics. It was designed from the beginning around infrastructure commitments measured in tens or hundreds of billions.

For a more realistic comparison, we would look at established operators such as Crusoe, Lambda, Nscale or Fluidstack. Their rounds remain extraordinarily large, but they give a clearer picture of how investors finance identifiable AI infrastructure businesses.

How young are AI data center startups when they raise big rounds?

AI data center companies are raising serious money surprisingly young, with a median age of roughly 2.8 years at the time of financing.

For transactions above $50 million, the median company age is about 3.6 years. Companies raising below $5 million are generally much younger, with a median age under one year.

TensorWave illustrates the speed of the funding cycle. Founded in 2023, it raised $100 million in Series A funding in May 2025 and $350 million in Series B funding in June 2026.

Gimlet Labs, also founded in 2023, announced an $80 million Series A in March 2026 and a $300 million Series B less than six months later.

Then there's Fluidstack, which announced that it had raised an $830 million Series A in January 2026 at a $7.5 billion valuation. The company publicly confirmed the financing in July.

These examples show that institutional investors are prepared to fund substantial infrastructure expansion long before an operator develops a lengthy corporate history. Of course, being young does not make an enormous round easy to obtain. The companies receiving that money still need to persuade investors they can deliver capacity at scale.

Where do AI data center startups start struggling to raise their next round?

The clearest gap in AI data center fundraising appears between Series A and Series B, where the number of recorded transactions falls from 16 to seven.

The amount of money needed also jumps sharply. Median Series A financing is $75 million, while Series B reaches $230 million. Series C climbs to $800 million.

Gimlet Labs and TensorWave demonstrate that companies can successfully make the transition, but the number of businesses reaching that level is much smaller than the number securing Series A funding.

By Series D+, the median recorded check exceeds $1.4 billion. These are financings associated with substantial infrastructure expansion, and the investor pool becomes very different from the one supporting early product development.

We should be careful about interpreting the counts as startup survival rates. The companies raising Series A and Series B during the same period are not necessarily members of the same founding cohort.

Still, the funding ladder shows where financial demands become much harder to meet. Investors are willing to finance new AI infrastructure businesses, but relatively few companies attract the capital needed to become large operators.

Funding stage	Deals	Total funding	Median round
Seed	7	$114.8M	$12M
Series A	16	$2.37B	$75M
Series B	7	$2.54B	$230M
Series C	5	$4.32B	$800M
Series D+	4	$7.26B	$1.438B
Growth Equity	9	$4.87B	$220.5M
Unknown / bespoke	8	$47.74B	$365.5M

Why are so few AI compute startups reaching Series C?

The AI Compute Data Centers category has no recorded Series C, Series D+ or Growth Equity financing, even though it contains nine funded companies.

Its transactions are concentrated at Seed, Series A and Series B. The median round is $33.5 million, far below the level seen among integrated infrastructure operators.

TensorWave is one of the more advanced businesses in the category, having raised a $350 million Series B. But even that transaction remains an earlier-stage financing by conventional labels.

We see two plausible explanations for the absence of later rounds.

First, the companies are relatively young. Median company age at financing is approximately 1.7 years, so many simply haven't had time to progress further.

Second, successful providers may eventually broaden their business models. A company that begins by selling GPU compute could add owned facilities, networking, storage and more extensive infrastructure services, becoming closer to the integrated-operator category.

We cannot establish how much category migration has happened from the funding records alone. The observed result is that AI Compute has a much thinner late-stage financing record than integrated infrastructure.

Are investors mostly backing new AI data center startups or existing ones?

Investors are overwhelmingly putting money into AI data center companies that have already raised capital before, with follow-on financings accounting for 46 of the 56 transactions.

Only nine deals are classified as first financings, while one has an unknown history.

The difference becomes particularly obvious when we examine capital. Excluding Stargate, first financings account for approximately $1.12 billion, compared with $23.08 billion for follow-on transactions.

Nscale, Lambda and Firmus provide three distinct examples. Each attracted successive investments to expand its business, often with much larger amounts than in previous rounds.

The pattern makes commercial sense. Expanding a data center or GPU cloud platform can require repeated fundraising even when the company is growing successfully.

The small number of first financings doesn't mean investors have stopped supporting new businesses. The larger financial opportunities are concentrated among operators that have already shown some ability to build and run infrastructure.

Which investors keep coming back to the AI data center market?

AI data center financing has a surprisingly varied group of lead investors, despite the heavy concentration of money among funded companies.

We identified 52 distinct disclosed lead or co-lead names across 42 financings. Only seven appeared more than once.

Valor Equity Partners and Mubadala Capital participated as leads or co-leads in two Crusoe rounds totaling $5.275 billion. Aker was involved in major Nscale financings, while 8090 Industries backed transactions across integrated and modular infrastructure.

Coatue, Magnetar and AMD Ventures also appeared repeatedly.

Large AI infrastructure operators can attract different kinds of capital as they grow. Early-stage investors may finance the initial business, while growth funds, industrial investors and other institutions become relevant when expansion requires several hundred million dollars or more.

There is an important distinction between appearing in a financing and committing the entire amount. The totals below represent the full round sizes associated with each lead or co-lead, rather than the investor's individual contribution.

Lead or co-lead investor	Deals	Combined round sizes
Valor Equity Partners	2	$5.275B
Mubadala Capital	2	$5.275B
Aker ASA	2	$3.10B
8090 Industries	2	$2.23B
Coatue	2	$1.605B
Magnetar	2	$450M
AMD Ventures	2	$450M

Why is NVIDIA involved in so many AI data center funding rounds?

NVIDIA has become one of the most important strategic investors in AI data centers, appearing in 12 of the financings we tracked.

Its involvement reflects the role of GPUs in the entire industry. Infrastructure companies need reliable access to accelerator hardware, while NVIDIA benefits from the growth of customers capable of deploying increasingly large computing clusters.

Nscale provides a useful example. Its March 2026 Series C included NVIDIA, Dell, Lenovo and Nokia. Those companies have commercial relationships with technologies and equipment that AI infrastructure operators need to build functioning systems.

Other strategic backers appear frequently too. Supermicro and Microsoft's M12 each feature in three transactions, while Pegatron, AMD Ventures, Aker and Nokia appear more than once.

Overall, we identified strategic participation in 32 of 56 financings, or 57.1%.

The difference in check sizes is striking. Rounds with a named strategic investor have a median of $240 million, compared with $51 million for those without one, a roughly 4.7-fold gap.

That doesn't establish that strategic investors cause larger rounds. Large, established operators may naturally attract both more money and more industrial partners. But financing and supply-chain relationships are clearly intertwined in the AI data center industry.

Is the United States taking most of the AI data center funding?

North America attracts most AI data center investment, accounting for 32 transactions and $59.73 billion of recorded capital.

Stargate inflates that figure considerably, but the region still accounts for approximately $14.73 billion after excluding the project, more than Europe or Asia-Pacific individually.

Crusoe, Lambda, CoreWeave and Together AI are important parts of the North American funding landscape. They also show that significant capital is reaching operating businesses beyond the biggest infrastructure initiative.

Asia-Pacific recorded 13 deals worth $5.06 billion. Firmus alone accounts for approximately $3.3 billion, making the region particularly dependent on one recurring fundraiser.

Europe registered 11 transactions and $4.42 billion. Its median financing is $52 million, below North America's $196.5 million, although major transactions involving Nscale and Fluidstack show that European operators can still attract exceptional amounts.

Nscale's involvement in Stargate Norway and Stargate UK also illustrates how AI infrastructure increasingly connects commercial computing demand with national and regional ambitions to secure local AI capacity.

For now, North America has the deepest observed pool of funded companies. Europe and Asia-Pacific have credible large operators but considerably fewer businesses attracting repeated scale capital.

Is AI data center fundraising getting busier right now?

AI data center fundraising has become noticeably busier, with September 2026 producing seven transactions, the highest monthly deal count in the period we studied.

Across the 33 completed months, the average was 1.7 deals per month and the median was just one. Several months in 2024 recorded no qualifying transactions.

September 2026 was different. The database includes Crusoe's $3.9 billion Series F, Fluidstack's $1.5 billion financing, Gimlet Labs' $300 million Series B, Verda's $189 million Series B, and transactions involving Helix and GMI Cloud.

Together, the month's seven deals represented approximately $7.15 billion.

The capital record still belongs to January 2025, when Stargate alone contributed $45 billion. But September's funding came from several separate companies, making it a better example of the market's growing transaction activity.

One unusually active month doesn't prove fundraising will keep accelerating. Still, the more frequent multi-deal months in 2026 suggest that financing is spreading across more established infrastructure businesses instead of appearing mainly through isolated announcements.

Does a billion-dollar AI data center funding round mean the company received a billion in cash?

A billion-dollar AI data center financing announcement doesn't always mean that the company received a billion dollars of new cash at the announcement date.

This distinction is particularly important because some large transactions combine cash investments with commitments, equity instruments, earlier financing or other corporate assets.

Nscale offers a concrete example. Its $2 billion Series C incorporated an earlier financing instrument and the roll-up of Aker's joint venture stake. In its first-quarter 2026 shareholder letter, Aker confirmed investing $350 million in cash while transferring its existing joint venture interest into Nscale.

Crusoe's September 2026 announcement described the initial closing of an anticipated $3.9 billion Series F. HyperVault's $1 billion entry also represents a financing commitment rather than proof that the entire amount was immediately paid.

There are similar qualifications around certain transactions involving Firmus and Helix.

These differences don't make the announcements meaningless. They do mean that we need to distinguish the reported size of an equity financing from cash that has already been funded.

For this market, announced and committed capital is a useful way to compare the scale of investment activity, provided we keep those differences visible.

What's the biggest problem for AI data center startups trying to scale?

The hardest part of scaling an AI data center startup appears to be securing enough capital and infrastructure to move from an initially funded compute business into a large, repeat-financed operator.

The funding ladder narrows sharply after Series A, and only 11 of the 38 companies raised more than once within the observed window.

Meanwhile, the companies that successfully attract repeated capital can move quickly into enormous financing requirements. The progression of Nscale, Lambda and Gimlet Labs illustrates how dramatically round sizes can increase.

Physical constraints add another layer of difficulty. An AI infrastructure company needs access to GPUs, power, cooling, networking and facilities while also finding customers willing to pay for the capacity.

Aker's description of these challenges in 2026 matches what we see in the financing records: once companies try to operate at scale, their capital needs begin to resemble those of major infrastructure businesses.

The open question is how many younger AI compute providers will eventually join that group. Currently, financing evidence supports a relatively small collection of businesses with the ability to secure major rounds repeatedly.

What should we actually take away from the AI data center funding data?

AI data center funding has reached enormous scale, but the money is concentrated in far fewer companies than the headline figures suggest. The market includes 38 funded businesses, and a small group of repeat fundraisers accounts for most of the capital outside Stargate.

Integrated infrastructure operators have become the main destination for recurring investment. Investors are giving them hundreds of millions or billions to expand computing capacity, acquire equipment and build the physical infrastructure required by AI customers.

The median financing is already exceptionally high, and check sizes rise dramatically after Series A. That creates a difficult transition for younger companies, especially those still focused on narrower compute services.

Strategic investors also play a major role. NVIDIA's repeated participation and the involvement of companies such as Dell, Nokia and Supermicro show how closely funding is tied to access to equipment and commercial partnerships.

The regional picture is uneven. North America attracts the most money and has the largest group of funded operators, while Europe and Asia-Pacific depend more heavily on a few companies raising very large amounts.

And we should keep the financing structures in perspective. Stargate's commitments, Nscale's combination of cash and corporate assets, and Crusoe's staged closing show why announced capital cannot always be treated as cash already received.

The most interesting question now is how many companies will manage to join the small group of established operators that can keep raising serious money. We already know that investors are willing to commit billions to AI data centers. What remains much less certain is how broadly that funding will spread as the market matures.