Who are the top investors in AI infrastructure?
In our AI infrastructure market deck, you will find everything you need to understand the market
SUMMARY
Nvidia is the top investor in AI infrastructure today, with Altimeter, General Catalyst, Spark Capital, Index Ventures, Redpoint, Lux, Coatue and Fidelity forming the strongest financial-investor group behind it.
The ranking changes a lot depending on what counts as infrastructure. We focus on independent companies selling AI compute, clouds, inference, model-serving and execution infrastructure, rather than chipmakers, frontier-model labs or ordinary data-center operators.
Dollar totals alone are a poor way to rank these investors. An early $20 million bet in Runpod says more about company selection than a multibillion-dollar strategic investment made after an AI cloud is already a major customer and partner.
The market is splitting into two financing races. Inference and developer infrastructure are attracting huge equity rounds, while AI clouds and data-center platforms increasingly depend on secured debt, project finance and infrastructure capital.
Nvidia stands apart because it can do more than provide equity. Its investments can come with GPUs, engineering support, reference architectures, financing relationships and help securing the physical capacity needed to bring AI factories online.
Among conventional financial investors, Altimeter has the broadest current portfolio across the stack through CoreWeave, Baseten and Crusoe. General Catalyst, Spark, Index and Redpoint look stronger when the lens narrows to developer infrastructure and inference.
The best early-stage records are not the same as the best current portfolios. Greylock, Redpoint, Kleiner Perkins, Sequoia, Amplify Partners, Intel Capital and Dell Technologies Capital deserve more credit for entering before several of today’s winners were obvious.
Repeated follow-on investment is one of the more useful filters in this market. Index in Fireworks AI, IVP in Baseten, Redpoint in Modal and General Catalyst across Together AI and Modal all kept buying after they had far more private information and at much higher prices.
Physical AI infrastructure is pulling a different class of investor into the market. Aker, Founders Fund, Mubadala Capital, Valor Equity Partners, TWG Global, Magnetar, Blackstone, Blue Owl and Brookfield matter because GPUs, power, land and data-center construction now require capital structures that look more like infrastructure finance than traditional venture capital.
The practical conclusion is that there is no single “best” investor archetype anymore. Nvidia is the clearest overall No. 1, but the strongest financial investors depend on whether the company needs early software conviction, growth equity, strategic hardware access or billions of dollars to build physical capacity.
This market map, featured in our AI infrastructure market deck, highlights top companies and startups in the AI infrastructure market
What are we actually calling AI infrastructure?
For this ranking, AI infrastructure means independent companies that sell compute, AI clouds, inference, model-serving or execution infrastructure, rather than every company that happens to buy GPUs.
We include GPU-cloud businesses such as CoreWeave, Lambda, Nscale, Crusoe and Nebius, along with software-heavy infrastructure companies such as Fireworks AI, Baseten, Together AI, Modal and Runpod. These companies sit at different depths of the stack, but customers pay them to train, serve, optimize or run AI workloads.
We leave frontier-model labs such as OpenAI and Anthropic outside the core startup sample because their primary product is the model. We also leave Nvidia, AMD and other chipmakers outside the portfolio-company set, even though strategic investments made by those companies absolutely count when we rank investors. Generic colocation operators are excluded unless the AI compute platform itself is central to the business.
That boundary keeps the ranking useful. Otherwise, a fund with scattered bets across semiconductors, foundation models and ordinary data centers could look more active than an investor that repeatedly picked the companies actually selling AI infrastructure.
Why can’t we just rank AI infrastructure investors by dollars invested?
Ranking AI infrastructure investors by dollars alone would give us the wrong answer, because a $20 million early bet and a $2 billion strategic investment test very different kinds of judgment.
Runpod shows the problem clearly. Intel Capital and Dell Technologies Capital co-led roughly $20 million of early funding before Runpod crossed one million developers and $120 million in annual recurring revenue. Summit Partners later led a $100 million round once the business was already much easier to understand. Both are meaningful investments, but the first one tells us much more about early company selection.
The opposite happens with GPU clouds. Nvidia recently put $2 billion into CoreWeave and another $2 billion into Nebius. Those checks dwarf most venture rounds, yet Nvidia already knows these companies as major customers and partners. Meanwhile, lenders can put even more money into the same businesses without making a venture-style bet at all.
We reviewed disclosed financing across CoreWeave, Lambda, Nscale, Crusoe, Nebius, Together AI, Baseten, Fireworks AI, Modal and Runpod, then separated four things: breadth across strong companies, timing, repeated follow-on conviction and the investor’s actual role in getting infrastructure built. That produces a much more useful ranking than simply adding up checks.
If you want more recent data on this point, please see our latest AI infrastructure market report.
As this chart shows, and as featured in our AI infrastructure market deck, search interest in AI infrastructure has risen sharply
Where is AI infrastructure money going right now?
Right now, the hottest equity money is going into inference and developer compute, while the heaviest balance-sheet money is moving into AI clouds and data-center capacity.
Four recent equity rounds make the inference shift unusually visible. Fireworks AI raised $1.505 billion at a $17.5 billion valuation after passing $1 billion in annualized revenue. Baseten raised $1.5 billion at $13 billion after reporting 20-fold revenue growth and 40-fold growth in inference volume over the previous year. Together AI raised $800 million at an $8.3 billion valuation after annual bookings passed $1.15 billion. Modal raised $355 million at $4.65 billion after saying annualized revenue had passed $300 million and grown fivefold since its previous round.
Together, those four companies raised about $4.16 billion at combined headline valuations of roughly $43.45 billion. Four companies pulling in that much equity tells us the category has moved well beyond a handful of experimental venture bets. Investors are now putting multibillion-dollar capital behind the layer that runs models in production, manages inference economics and gives agents somewhere to execute.
At the physical end of the market, the financing looks completely different. Nscale, Lambda and Nebius have lately added large secured facilities on top of their equity funding, while Nvidia has started organizing institutional capital around AI compute itself. The market now has two funding races happening at once: one around software and inference economics, and another around power, GPUs and data-center construction.
| Company | Recent equity round | Headline valuation | Lead or main investors |
|---|---|---|---|
| Fireworks AI | $1.505B | $17.5B | Atreides, Index Ventures, TCV |
| Baseten | $1.5B | $13B | Altimeter, Conviction, Spark |
| Together AI | $800M | $8.3B | Aramco Ventures |
| Modal | $355M | $4.65B | General Catalyst, Redpoint |
Is Nvidia clearly the top investor in AI infrastructure today?
Yes. Nvidia is currently the most important AI infrastructure investor, and nobody else is close once we combine portfolio breadth, check size and strategic leverage.
We can verify Nvidia on the cap tables of CoreWeave, Lambda, Together AI, Fireworks AI, Baseten, Crusoe and Nscale, while Nebius received a separate $2 billion strategic investment. Nvidia then went further with a recent $1.5 billion investment in SB Energy, tied to an Ohio campus where Nvidia is providing credit support for an initial 4.25 gigawatts of AI capacity and OpenAI is expected to be the customer.
The scale has changed sharply. Nvidia’s older startup investments were often minority venture checks. These days it can invest billions, help secure power and land, provide access to new GPU generations, shape the system architecture and bring financing partners into the same project. CoreWeave’s latest collaboration targets more than five gigawatts of Nvidia-based AI factories by 2030; the Nebius partnership also targets more than five gigawatts of Nvidia systems.
We should read Nvidia’s record differently from a normal VC’s because many portfolio companies also buy large amounts of Nvidia hardware. The commercial incentive is obvious. Still, that relationship can make Nvidia more valuable to an infrastructure company than a purely financial investor, especially when access to chips, reference architectures, engineering support and financing can all affect how quickly capacity comes online.
If you want more recent data on this point, please see our latest AI infrastructure market report.
This chart, included in our AI infrastructure market deck, shows annual VC investment in AI infrastructure startups
Which VC firms have backed the most serious AI infrastructure companies?
Among VC and growth firms, Altimeter currently has the strongest spread across AI infrastructure, while General Catalyst, Spark, Lux and Coatue have narrower but very strong portfolios.
Altimeter appears in CoreWeave’s $1.1 billion private round, Crusoe’s $1.375 billion Series E and Baseten’s financing history, then moved into the lead group for Baseten’s latest $1.5 billion round. Those three companies give Altimeter exposure to GPU cloud, physical AI factories and inference software rather than three versions of the same bet.
General Catalyst has a different shape. It led Together AI’s $305 million Series B at a $3.3 billion valuation and later built a second major position in Modal. Spark built a long-running position in Baseten and also invested in Crusoe. Lux backed Together AI early, stayed involved as it scaled, then led Modal’s Series B. Coatue led CoreWeave’s $1.1 billion Series C and also invested in Together AI.
Fidelity deserves its own category because it behaves more like a crossover institution than a classic VC. Its presence in CoreWeave, Crusoe and Nscale gives it one of the broadest portfolios on the capital-intensive side of AI infrastructure.
| Investor | Strongest disclosed positions | What we think the portfolio says |
|---|---|---|
| Altimeter | CoreWeave, Baseten, Crusoe | Best cross-layer growth portfolio |
| General Catalyst | Together AI, Modal | Strong in open-model and developer infrastructure |
| Spark Capital | Baseten, Crusoe | Deep inference conviction plus physical infrastructure |
| Lux Capital | Together AI, Modal | Early developer-infrastructure exposure |
| Coatue | CoreWeave, Together AI | Strong neocloud exposure |
| Fidelity | CoreWeave, Crusoe, Nscale | Broad crossover exposure to capital-heavy infrastructure |
| Atreides | Fireworks AI, Crusoe | Concentrated late-stage infrastructure bets |
Who is winning the AI inference investing race right now?
Index Ventures and Spark Capital have the strongest current records in AI inference infrastructure, although they got there in very different ways.
Index’s Fireworks AI position is a textbook follow-on bet. Sequoia led Fireworks’ $52 million Series B at a $552 million valuation. Lightspeed, Index and Evantic then led the $250 million Series C at $4 billion. Index stayed in the lead group when Fireworks raised $1.505 billion at $17.5 billion, alongside Atreides and TCV. The valuation moved more than fourfold between the last two rounds while Fireworks reported that annualized revenue had crossed $1 billion.
Spark’s Baseten record is longer. IVP and Spark led Baseten’s $40 million Series B, Spark remained involved through later rounds, and Spark joined the lead group again in the recent Series F. Baseten reported 20-fold revenue growth in the preceding year, so Spark was not simply following a rising valuation; it had been underwriting the inference thesis for years.
General Catalyst is the other investor we would put close to this group. It led Together AI’s Series B before the company more than doubled its valuation to $8.3 billion, and it now also sits on Modal’s cap table as a Series C lead. If production AI keeps shifting toward open models, custom models and agent runtimes, General Catalyst has exposure to more than one way that market can develop.
If you want more recent data on this point, please see our latest AI infrastructure market report.
This chart, included in our AI infrastructure market deck, shows why CoreWeave is winning in AI infrastructure
Who found the best AI infrastructure companies before they were obvious?
Greylock, Redpoint, Kleiner Perkins, Sequoia, Amplify Partners, Intel Capital and Dell Technologies Capital are the investors we would credit most for spotting major AI infrastructure companies early.
Baseten is the clearest correction to the usual narrative. Greylock, through Sarah Guo at the time, led Baseten’s Series A after also co-leading the seed financing. Years later, Baseten is valued at $13 billion. Redpoint led Modal’s $16 million Series A in 2023, while Amplify Partners had led the seed round; Modal is now valued at $4.65 billion and says annualized revenue has passed $300 million.
Kleiner Perkins led Together AI’s $102.5 million Series A when open-model infrastructure was still a much less settled category. Sequoia led Fireworks AI’s $52 million Series B at a $552 million valuation, before production inference had become one of the hottest parts of AI infrastructure.
Runpod gives us a less famous but useful example. Intel Capital and Dell Technologies Capital co-led roughly $20 million when the company was still proving that an AI-specific developer cloud could become a large business. Runpod later passed one million developers and $120 million in annual recurring revenue before raising $100 million from Summit Partners.
Which investors kept doubling down after their first AI infrastructure bet?
Index Ventures, IVP, Redpoint and General Catalyst stand out because they kept reinvesting after they had much more information and a much higher price.
Index stayed with Fireworks AI across consecutive late-stage rounds and remained a lead investor as the company scaled. IVP provides another useful example through Baseten: it co-led the Series B, stayed involved in later financings and then co-led the $300 million Series E with CapitalG. These investors chose to buy more after they had already seen the companies from the inside.
Redpoint’s Modal history may be even cleaner. Redpoint led the $16 million Series A, stayed through the company’s growth and then co-led the $355 million Series C alongside General Catalyst. Modal said it had grown fivefold since the previous round and passed $300 million in annualized revenue, so Redpoint had operating evidence when it chose to increase exposure.
General Catalyst has doubled down across two companies. It led Together AI’s Series B and participated again in the Series C, while separately co-leading Modal’s latest round. We give that behavior more weight than a one-time logo on a large cap table because the investor gets to reassess the company with far better information each time.
This chart, included in our AI infrastructure market deck, shows annual funding in AI infrastructure startups
Who is leading the AI cloud and AI factory funding race?
Aker, Founders Fund, Mubadala Capital, Valor Equity Partners and TWG Global are the strongest names when AI infrastructure turns into a multibillion-dollar physical buildout.
Aker has become especially important through Nscale. It led Nscale’s $1.1 billion Series B with a $285 million commitment that gave Aker 9.3% ownership on a fully diluted basis. Aker then co-led Nscale’s $2 billion Series C with 8090 Industries at a $14.6 billion valuation. The original Aker-Nscale Norway joint venture was also rolled into Nscale as part of that later transaction, deepening the relationship beyond a normal portfolio investment.
Crusoe shows a similar progression with different investors. Founders Fund led a $600 million Series D when Crusoe was valued around $2.8 billion. The next major round was $1.375 billion at a valuation above $10 billion, co-led by Mubadala Capital and Valor Equity Partners, with Founders Fund returning alongside Altimeter, Fidelity, Nvidia, Spark and others.
Lambda has moved into the same capital class. Its Series E raised more than $1.5 billion and was led by TWG Global, after an earlier $480 million round had included Nvidia and several hardware partners. Once these companies start building gigawatt-scale AI factories, the relevant investor pool naturally shifts toward firms that can keep supporting very large physical deployments.
If you want more recent data on this point, please see our latest AI infrastructure market report.
Who actually finances the GPUs and data centers behind AI infrastructure?
These days, private credit and project finance are almost as important as venture equity in AI infrastructure, and the recent numbers make that hard to argue with.
Nscale signed a $1.4 billion GPU-backed delayed-draw term loan led by PIMCO, Blue Owl and LuminArx, then secured another $790 million for its Norway buildout. Lambda closed a $1 billion senior secured credit facility to expand its Nvidia accelerator fleet and data-center capacity. Nebius added a roughly $775 million senior secured facility backed by deployed GPUs and contracted cash flows.
Those four disclosed financings add up to about $3.97 billion. For comparison, the four recent inference equity rounds we examined earlier totaled about $4.16 billion. A handful of secured infrastructure financings have therefore reached almost the same order of magnitude as four of the biggest venture rounds in inference.
The newest development pushes this much further. Nvidia recently announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to create independent AI-compute financing platforms intended to mobilize more than $500 billion of third-party capital over time. The target is enormous, but the direction is already clear: AI compute is being financed more like infrastructure, with long-duration capital and assets that can support credit, rather than relying only on startup equity.
This chart, included in our AI infrastructure market deck, compares the main business model options for AI cloud infrastructure providers
Are strategic investors mainly financing their own AI infrastructure customers?
Yes. Strategic AI infrastructure investors often finance companies that also buy their hardware, and Nvidia is the clearest example.
Nvidia’s investments sit inside a wider commercial loop. CoreWeave, Lambda, Nscale, Nebius, Together AI, Fireworks AI and other infrastructure companies all use Nvidia hardware or software. When Nvidia invests, it can earn an equity return while also creating demand for GPUs, networking, CPUs and its software stack. The recent SB Energy transaction makes that logic unusually explicit because Nvidia is investing in the developer, supporting the land and power buildout, and securing the site for Nvidia compute.
Dell follows a smaller version of the same playbook. Dell Technologies Capital backed Runpod early, while Dell itself invested in Nscale. AMD Ventures led Vultr’s first outside financing alongside LuminArx, and AMD also appeared as a strategic backer of Fireworks AI.
We therefore do not treat a strategic check as the same kind of independent validation as an Index or Redpoint investment. For a founder, though, the strategic investor may bring something the financial investor cannot: hardware allocation, engineering support, supply-chain relationships, enterprise distribution or a path to cheaper financing. In infrastructure, those advantages can directly change how fast the company grows.
Who are the top investors in AI infrastructure today?
Nvidia is the clear No. 1 AI infrastructure investor today, while Altimeter, General Catalyst, Spark, Index, Redpoint, Lux, Coatue and Fidelity form the strongest financial-investor group behind it.
Altimeter gets our highest mark among conventional growth investors because CoreWeave, Baseten and Crusoe give it credible exposure across three different layers. General Catalyst has one of the best current software-infrastructure portfolios through Together AI and Modal. Spark’s multi-round Baseten position, plus Crusoe, gives it a strong mix of inference and physical infrastructure. Index has a narrower portfolio, but its repeated leadership in Fireworks AI is one of the best concentrated bets in the market.
Redpoint and Lux rank highly because Modal gives both firms a strong developer-infrastructure winner, with Redpoint entering especially early and returning as a Series C lead. Coatue’s CoreWeave position still makes it one of the defining neocloud investors, while Fidelity’s CoreWeave, Crusoe and Nscale exposure gives it unusually broad coverage of capital-heavy infrastructure.
We would rank Aker separately rather than squeeze it into a Silicon Valley VC list. Its repeated Nscale commitments, ownership position and industrial relationship make it one of the most important investors in European AI infrastructure. Magnetar deserves the same special treatment for CoreWeave: its early equity exposure and later role in financing the company helped prove that GPU clouds could use structured debt at enormous scale.
Finally, Blackstone, Blue Owl and Brookfield belong in the answer even though they are not venture firms. The newest financings show that the next stage of AI infrastructure is going to be constrained by access to power, GPUs, land and low-cost capital. The investors that can solve those problems are becoming just as important as the funds that picked the startup in the first place.
| Investor | Our current judgment | Best evidence |
|---|---|---|
| Nvidia | Clear No. 1 overall | Broad strategic portfolio plus multibillion-dollar direct investments and financing partnerships |
| Altimeter | Best broad conventional growth portfolio | CoreWeave, Baseten, Crusoe |
| General Catalyst | Top developer and open-model infrastructure investor | Together AI, Modal |
| Spark Capital | Top long-duration inference investor | Baseten, Crusoe |
| Index Ventures | Best concentrated inference bet | Repeated lead investor in Fireworks AI |
| Redpoint | Best early-to-growth developer-infrastructure record | Modal Series A and Series C lead |
| Lux Capital | Strong early developer-infrastructure investor | Together AI, Modal |
| Coatue | Defining neocloud growth investor | CoreWeave lead investor, plus Together AI |
| Fidelity | Top crossover institution | CoreWeave, Crusoe, Nscale |
| Aker | Top industrial AI-infrastructure investor in Europe | Repeated Nscale leadership and ownership |
| Magnetar | Key structured-capital pioneer | Early CoreWeave equity and debt financing |
| Blackstone / Blue Owl / Brookfield | Top capital-stack investors | Large AI compute, GPU and data-center financing |
If you want more recent data on this point, please see our latest AI infrastructure market report.
This chart, featured in our AI infrastructure market deck, shows the share of revenue generated by each customer segment in the AI infrastructure market
OUR METHODOLOGY
There is no clean league table for AI infrastructure investors. The companies sit across different parts of the stack, the investors range from venture firms to strategic technology companies and infrastructure-capital providers, and the largest check is not necessarily the strongest investment signal. We therefore built the ranking around the evidence that reveals investor strength more directly.
Our core AI-infrastructure universe includes independent companies selling compute, AI clouds, inference, model-serving or execution infrastructure. We include companies such as CoreWeave, Lambda, Nscale, Crusoe, Nebius, Fireworks AI, Baseten, Together AI, Modal and Runpod, while keeping frontier-model labs, chipmakers and generic colocation operators outside the portfolio-company set unless the AI compute platform itself is central to the business.
We assessed investors across portfolio breadth, entry timing, follow-on conviction, exposure to the strongest parts of the market, and the investor’s role in helping infrastructure actually scale. We prioritized recent disclosed financings and operating milestones, while using earlier rounds when they were needed to see who entered before a company became obvious or who kept investing over time.
We treated different forms of capital separately. An early venture investment, a multibillion-dollar strategic check and a GPU-backed credit facility tell us different things, so we did not collapse them into a single total. We also gave more weight to repeated investment when an investor had already seen the company from the inside and still chose to buy more at a higher price.
The final ranking is a structured synthesis rather than a mechanical score. Nvidia ranks first because its record combines unusual portfolio breadth, very large direct commitments and strategic leverage over hardware, engineering and financing. Conventional financial investors are judged more heavily on selection, timing, cross-company breadth and follow-on behavior, while infrastructure-capital providers are judged on their ability to finance GPUs, power and data-center construction at scale.
Key sources used for this analysis include: CoreWeave on Nvidia’s $2 billion investment and 5+ GW expansion, Nebius on Nvidia’s $2 billion strategic investment and deployment target, Fireworks AI on its $1.505 billion Series D, Baseten on its $1.5 billion Series F, Together AI on its $800 million Series C, and Modal on its $355 million Series C.
For the capital-intensive side of the market, key sources include: Crusoe on its $1.375 billion Series E, Lambda on its $1.5 billion-plus Series E, Nscale on its $1.1 billion Series B, Nscale on its $2 billion Series C, CoreWeave on its $7.5 billion debt facility, and Nscale on its $1.4 billion GPU-backed term loan.
We also used Intel Capital on Runpod’s early $20 million funding, Runpod on passing one million developers, Lambda on its $1 billion secured credit facility, Nebius on its roughly $775 million secured financing, Nvidia on AI-compute financing platforms targeting more than $500 billion of third-party capital, and Nvidia on its SB Energy investment and Ohio capacity commitment.
This chart, included in our AI infrastructure market deck, shows how GPU cloud infrastructure technology has evolved over time
Related blog posts
- What's worth building in AI infrastructure?
