AI Infrastructure Funding Tracker (267 deals)

Our funding tracker covering the AI infrastructure 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 on AI infrastructure funding, valuations and who's ahead.

Is AI infrastructure really a $52.5 billion funding market right now?

Yes, AI infrastructure funding really reached $52.54 billion across the period we studied, although almost all of that money came from the market’s large-scale financing layer.

We tracked 266 financings across 193 companies. Once rounds above $50 million are removed, quantifiable funding falls to $2.53 billion. Deals of $100 million or more account for 93% of all disclosed capital. With 93% of capital in $100 million-plus rounds, we follow how much money came in, quarter by quarter, in our report on AI infrastructure.

That gives the market a very unusual shape. There is a large base of Seed, Series A and mid-sized venture rounds, then a much narrower group of companies raising hundreds of millions or several billion dollars at a time.

The biggest recent deal makes the scale clear. Crusoe’s latest Series F initial closing came in at $3.9 billion. The company says it now has more than $140 billion of total contracted value across its infrastructure platform. A financing that large makes sense in a business building data centers, manufacturing infrastructure and cloud capacity, while it says very little about what an ordinary AI infrastructure startup can raise.

How many AI infrastructure startups are actually getting funded?

The AI infrastructure funding market currently consists of 193 funded companies in our research, and three quarters of them appear only once.

We found 145 companies with a single financing during the period, 31 with two and 17 with three or more. Only 48 companies raised multiple times.

Those 48 repeat fundraisers absorbed roughly $41.09 billion. The other 145 companies collectively raised about $11.45 billion.

So there are plenty of companies entering AI infrastructure, yet the really deep pools of money are already clustering around a fairly small set of businesses that investors are willing to fund again and again.

Is most AI infrastructure funding ending up with the same few companies?

Very much so: the ten best-funded AI infrastructure companies collected 52.6% of all quantifiable capital we tracked.

Crusoe alone reached roughly $5.28 billion of cumulative financing inside the period. Lambda, Nscale, Firmus, Groq, Fluidstack, Cerebras and Baseten also raised multiple large rounds.

The concentration becomes clearer when we compare companies with individual deals. The ten largest financings represent 37.2% of the market’s capital, while the ten largest companies reach 52.6%. The biggest companies keep coming back for another large check rather than relying on one exceptional round.

That is one of the clearest features of AI infrastructure funding today. Scale compounds quickly once a company gets into the group that investors trust with billion-dollar infrastructure programs.

Company	Financings	Cumulative capital
Crusoe	2	$5.28B
Lambda	3	$3.30B
Nscale	3	$3.26B
Firmus	4	$2.95B
Groq	4	$2.39B
Fluidstack	2	$2.33B
Cerebras	2	$2.10B
Baseten	5	$2.07B
HyperVault AI Data Center	1	$2.00B
Tenstorrent	2	$1.98B

What can a normal AI infrastructure startup realistically raise?

A typical AI infrastructure financing is around $38.5 million today, far below the $202.1 million average created by the biggest rounds.

The distribution is quite broad. We found 24 disclosed financings below $5 million, 65 between $5 million and $20 million, and 54 between $20 million and $50 million. More than half of the disclosed deals therefore stay below $50 million.

Company-level funding looks similar. The median company raised $29 million across the whole period, while the average reached $272.2 million.

For a founder trying to benchmark a realistic round, the median gives the more useful reference point. A $200 million financing already puts a company deep into the unusually well-capitalized end of this market.

Which parts of AI infrastructure are getting the most money?

AI Cluster Cloud currently absorbs by far the most AI infrastructure funding, with $24.86 billion raised across 53 deals.

AI Accelerators actually produced more financings, with 72, yet collected $13.04 billion. Network Fabric reached $5.69 billion and Compute Platforms $5.62 billion.

The difference in median checks is just as revealing. Cluster Cloud sits at $100 million, Accelerators at $69 million, Network Fabric at $58 million and Compute Platforms at $19 million. Medians from $19 million to $100 million by category show where the money concentrates; we follow the biggest rounds in our AI infrastructure market report.

That last category is especially interesting. Compute Platforms generate plenty of companies and deals, while their funding profile still looks much closer to a conventional software market. At the other end, Cluster Cloud companies quickly run into the cost of GPUs, power, buildings, cooling and network infrastructure.

Category	Deals	Companies	Capital	Median round
AI Cluster Cloud	53	36	$24.86B	$100M
AI Accelerators	72	55	$13.04B	$69M
AI Network Fabric	51	37	$5.69B	$58M
AI Compute Platforms	68	51	$5.62B	$19M
AI Server Systems	7	7	$1.82B	$21M
AI Storage Systems	3	3	$1.02B	$11M
Logic Semiconductors	9	9	$313M	$6M
Memory Semiconductors	1	1	$135M	$135M
Sensor Semiconductors	1	1	$47M	$47M
Power Semiconductors	1	1	$1.7M	$1.7M

Why are AI cloud companies suddenly raising billions?

AI cloud companies are raising billions because their growth increasingly requires physical infrastructure at data-center scale.

Nscale shows how quickly the numbers can escalate. The company went from a $155 million Series A to a $1.1 billion Series B and then a $2 billion Series C. Alongside the equity, Nscale also signed a $1.4 billion GPU-backed loan to finance cluster deployments across Europe.

Firmus has followed an equally capital-heavy path. We recorded four equity financings worth roughly $2.95 billion in total, including a $2 billion strategic round. Separately, Firmus secured a $10 billion debt facility for Project Southgate.

The latest commercial activity helps explain why investors are comfortable supplying so much capital. Firmus now says it has more than 900 MW of contracted capacity and has added OpenAI as an anchor customer. Crusoe says its contracted value has passed $140 billion, and the company recently announced a multi-year partnership with Perplexity covering training and production inference.

Lambda gives us another example. Its multibillion-dollar agreement with Microsoft covers infrastructure powered by tens of thousands of NVIDIA GPUs.

These companies can have enormous capital needs before revenue is fully recognized because capacity has to be secured and built ahead of delivery. In that part of AI infrastructure, funding starts to resemble industrial financing remarkably quickly.

The wider spending environment is moving in the same direction. The International Energy Agency reported that capital expenditure by five large technology companies exceeded $400 billion in 2025 and could rise another 75% in 2026, largely as companies race to add data-center and AI capacity.

Are AI chip startups still attracting serious money?

AI accelerator startups are attracting very serious funding today, with $13.04 billion spread across 72 financings in our research.

The category has 55 funded companies and a $69 million median round. That already makes it one of the deeper parts of the market before we even reach the largest deals.

Positron AI is a good example of how quickly the curve can steepen. We tracked a $23.5 million Seed financing, a $51.6 million Series A, a $230 million Series B and then an announced $875 million Series C financing. The final figure includes a tranche of up to $500 million, so we preserve that qualification rather than treating every dollar as already closed cash.

The reason investors can write these checks fairly early is straightforward. Building competitive accelerator hardware means paying for chip design, tape-outs, memory, packaging, manufacturing capacity and complete systems long before the business reaches software-like scale.

There is still a wide field of accelerator companies, but the funding requirements rise quickly once a startup moves from proving a chip architecture to putting systems into production.

Why are AI networking and optics startups suddenly attracting strategic investors?

AI networking and optical infrastructure have become strategic funding targets because moving data between accelerators is now one of the hardest constraints in large AI systems.

We counted 51 Network Fabric financings worth $5.69 billion. More strikingly, 58.8% of those deals involved an identifiable strategic investor, the highest rate among the major categories.

Ayar Labs captures the trend well. Its $500 million Series E brought in AMD, Alchip, MediaTek and NVIDIA alongside financial investors. Ayar is building co-packaged optical I/O designed to connect large numbers of AI accelerators with higher bandwidth and lower power consumption. Chipmakers writing checks next to financial investors is a pattern; we look at which investors make the biggest bets in our report on AI infrastructure.

The activity around Ayar also fits what larger incumbents are doing directly. NVIDIA has announced multibillion-dollar commercial and investment agreements with Coherent and Lumentum to expand advanced optics capacity, manufacturing and R&D.

Once clusters contain thousands of costly accelerators, poor interconnect performance wastes expensive compute. That puts networking companies unusually close to the economics of the whole AI factory, which helps explain why chipmakers, systems companies and other industrial players want a seat at the table.

How far along are AI infrastructure companies when the big checks arrive?

The really large AI infrastructure checks usually arrive after a few years of company building, even though investors start backing the sector very early.

The median company age at financing is three years. Seed financings happen at a median age of one year, Series A at two years, Series B at four, Series C at five and Series D+ at nine.

Round sizes rise far faster than company age. The median check moves from $8.75 million at Seed to $38.5 million at Series A, $68 million at Series B, $210 million at Series C and $640 million at Series D+.

Many of the funded companies are genuinely young. Businesses founded in 2023 account for 50 financings, while the 2024 cohort accounts for 42. Investors are backing new AI infrastructure companies quickly, then reserving the truly huge capital pools for businesses that survive several years of technical and commercial validation.

Stage	Deals	Median round	Median company age
Seed	81	$8.75M	1 year
Series A	68	$38.5M	2 years
Series B	33	$68M	4 years
Series C	27	$210M	5 years
Series D+	23	$640M	9 years
Growth Equity	13	$505M	6.5 years
Unknown	21	$44.5M	-

Where does the AI infrastructure funding ladder really break?

The clearest squeeze in AI infrastructure funding comes between Series A and Series B, where the number of financings drops from 68 to 33.

After that, the ladder narrows much more gradually. We counted 27 Series C financings and 23 Series D+ financings.

These numbers should not be read as a literal survival rate because our research covers one continuous funding window. Plenty of recent Series A companies have simply had no time to raise their next round.

Still, the drop from 68 to 33 is hard to ignore. Investors appear quite willing to fund teams through initial product development and early commercialization. The next check asks for much stronger evidence that a company can become an important piece of the AI infrastructure stack.

Once a business clears that stage, access to larger pools of capital becomes visibly easier for the strongest companies.

Is new AI infrastructure money going to new startups or mostly to companies investors already know?

Most AI infrastructure capital is currently going back into companies that have already raised before.

We classified 200 deals as follow-on financings and 64 as identifiable first financings, with two remaining uncertain. Follow-ons captured roughly $48.12 billion, compared with $4.12 billion for first financings. Follow-ons capturing most of the money favor companies that won early rounds; we look at the likely future winners in our AI infrastructure market report.

The typical check is also very different. First financings have a $12 million median, while follow-ons reach $70 million.

That fits the company concentration we saw earlier. New startups continue to enter the market, but the giant capital requirements show up once investors have already seen a company build technology, win customers and prove that it can deploy the next round productively.

Is AI infrastructure funding basically a US market?

AI infrastructure funding is heavily US-centered today: North America accounts for 168 deals and $39.02 billion of capital in our research.

The United States alone represents 166 financings and roughly $38.8 billion. North America therefore captures 74.3% of all quantifiable capital while accounting for around 63% of deals.

Europe has 44 deals and $5.37 billion. Asia-Pacific is close in activity with 47 deals, yet reaches $7.92 billion because a few very large financings, particularly Firmus, lift the regional total.

Europe has produced its own giant exception through Nscale, whose $2 billion Series C is now one of the market’s largest rounds. Still, these individual successes have yet to erase the huge financing gap with the US.

The Middle East is much thinner in this private-company dataset. We found seven deals worth $232 million, with only one above $50 million. Large government, sovereign and corporate AI infrastructure programs can sit outside the venture market we are measuring here, so that figure should be read specifically as startup financing activity.

Which investors keep coming back to AI infrastructure?

AI infrastructure lead investing is surprisingly dispersed, although a small group of funds keeps appearing around major rounds.

We could identify leads or co-leads for 81.6% of financings. After separating syndicates, roughly 250 different investors appear in those lead positions, while only 52 show up more than once.

Atreides Management and Spark Capital lead the count with seven appearances each. Valor Equity Partners, Lightspeed Venture Partners and Andreessen Horowitz each appear five times.

The combined round values below refer to the entire financings those investors led or co-led. They should never be read as the amount personally invested by each fund.

Lead or co-lead	Deals	Combined value of rounds
Atreides Management	7	~$7.91B
Spark Capital	7	~$2.10B
Valor Equity Partners	5	~$6.48B
Lightspeed Venture Partners	5	~$1.47B
Andreessen Horowitz	5	~$1.24B
Tiger Global	4	~$1.37B
CapitalG	4	~$733M
Jane Street	3	~$2.70B
Disruptive	3	~$1.75B
Sequoia Capital	3	~$492M
Seligman Ventures	3	~$455M
IVP	3	~$415M

How much do NVIDIA and other strategic investors actually matter in AI infrastructure?

Strategic investors matter a lot in AI infrastructure: we found them in 116 financings, equal to 43.6% of all deals.

Those rounds are also much larger. The median financing with an identifiable strategic investor reaches $120 million, compared with $19 million for deals without one. Strategic backers lifting round sizes this much shape who gets funded, and by whom, which we track in our AI infrastructure report.

The pattern is strongest around physical infrastructure. Accelerators account for 36 strategic-backed rounds, Network Fabric for 30, Cluster Cloud for 24 and Compute Platforms for 19.

NVIDIA appears in 26 strategic financings in our research, more than any other strategic investor. Intel Capital appears in 11, Samsung Catalyst Fund in eight, Cisco Investments in seven, and AMD Ventures and Arm in six each.

The corporate involvement often has a practical logic. NVIDIA sells the accelerators used by AI clouds, while also investing around networking, optics and infrastructure companies. AMD, Arm, Cisco, Intel and Samsung sit in similar positions elsewhere in the stack.

That makes strategic capital especially useful to watch in this market. A corporate investor can bring manufacturing relationships, technical integration, supply access or customers alongside the money, and those things become more valuable as a company moves closer to large-scale deployment.

What is AI infrastructure funding still missing?

AI infrastructure funding still has surprisingly thin areas, especially once we move away from compute, accelerators and networking.

Power Semiconductors, Sensor Semiconductors and Memory Semiconductors each appear only once as standalone categories in our classification. With samples that small, we cannot say much about those markets from this dataset.

Storage and Server Systems have more capital behind them, yet very few companies. Storage has just three companies and is heavily influenced by VAST Data. Server Systems has seven.

Compute Platforms show a different kind of gap. They produce a large number of early-stage financings, while relatively few reach the later-stage, capital-heavy end of the ladder. That may partly reflect how young many of these companies are.

The same unevenness appears geographically. We can find AI infrastructure startups across North America, Europe, Asia-Pacific and the Middle East, while the deepest private scale-up financing remains concentrated in only a few countries and companies.

The market already has extraordinary financial depth around its most expensive bottlenecks. Breadth across the rest of the infrastructure stack is much less developed.

Can we actually trust the $52.5 billion AI infrastructure funding total?

We are confident in the broad $52.54 billion AI infrastructure picture, while several large transactions required careful treatment before we were comfortable using them.

Fluidstack is a good example. Its $830 million Series A appeared twice in the underlying deal research because the company publicly announced it later while explicitly saying the round had been raised in January. We consolidated that into one financing.

VAST Data required a different correction. Its Series F transaction included both primary and secondary equity and represented approximately $1 billion in total transaction value. We use that figure rather than an earlier $2 billion database value.

For Nscale, we separated the company’s $2 billion Series C from a previously announced $433 million SAFE instead of combining them. Crusoe moved in the opposite direction: an earlier provisional figure was replaced when the company formally announced an initial $3.9 billion Series F closing before our research cutoff.

Other uncertainty remains visible. Six financings have undisclosed amounts, 16 disclosed figures are lower bounds, 28 involve approximations or currency conversions, and several deals involve extensions, tranches or secondary shares.

A funding database this large will never make every dollar economically identical. Our goal is to keep the obvious distortions out, preserve unusual deal structures where they matter and avoid pretending that an approximation is a precise cash receipt.

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

AI infrastructure has become a huge funding market, although most startups are still raising ordinary venture-sized rounds. The billions come from a narrow upper layer of companies building the physical capacity behind AI.

Capital is concentrating faster than company count. We found 193 funded businesses, while a small recurring group of clouds, accelerator companies and infrastructure platforms absorbs a very large share of the money. Clouds and accelerator makers absorbing so much money is why we cover who the top companies are today in our AI infrastructure report.

The most expensive bottlenecks attract the deepest funding. AI Cluster Cloud dominates dollars, accelerators lead in deal count, and networking has become unusually attractive to strategic investors.

The funding curve gets steep fast. Companies can enter the market with single-digit or tens-of-millions rounds, while the strongest later-stage businesses move into $200 million, $500 million and billion-dollar financings within only a few years.

Series A to Series B is currently the clearest point where the field tightens. Getting an AI infrastructure product funded is one hurdle; convincing investors that the company can become a major production platform is a much harder one.

Corporate investors are deeply embedded in the market. NVIDIA, Intel, AMD, Arm, Cisco and other industrial players are already financing companies whose technology can affect their own supply chains, customers and platforms.

And the latest commercial developments make the mega-rounds easier to understand. AI infrastructure companies are signing contracts measured in tens of thousands of GPUs, hundreds of megawatts and, in Crusoe’s case, more than $140 billion of stated contracted value. At that scale, the financing needs quickly move beyond what most people would recognize as normal startup funding.