What are the top AI data center startups by revenue today?

The ranking of top startups in the AI Data Center Market by revenue

Below is a table ranking the startups of this market by their current revenue scale. 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 on the AI data center market.

Which AI data center startups make the most money today?

CoreWeave currently ranks first among AI data center startups by revenue, followed by IREN, Lambda, Applied Digital, and DayOne Data Centers. The gap is substantial: CoreWeave reported $2.58B in quarterly revenue for Q2 2026, while IREN's second-place figure is a $1B operating revenue run-rate as of August 2026.

The ranking covers 20 startups with credible financial figures, from large GPU cloud operators to companies developing and leasing high-density data centers. Core Scientific, Cerebras Systems, Runpod, ST Telemedia Global Data Centres, and Nscale complete the top ten.

The companies make money in different ways. CoreWeave sells AI computing services, DayOne develops hyperscale facilities, and Core Scientific leases computing infrastructure. We also encounter different financial measures: quarterly sales, annual revenue, half-year revenue, ARR, and revenue run-rates. Each remains labeled according to what the company actually reported.

CoreWeave generated more revenue in a single quarter than many independent data center companies reported over an entire year. Yet several developers with multigigawatt portfolios remain unranked because their disclosed capacity cannot establish how much revenue they earn.

Rank	Startup	Revenue figure	Metric and period
1	CoreWeave	$2.58B	Quarterly Revenue, Q2 2026
2	IREN	$1B	Revenue Run-Rate, Aug 2026
3	Lambda	~$760M	Revenue Run-Rate, end of 2025
4	Applied Digital	$342M	Quarterly Revenue, Q1 FY2027
5	DayOne Data Centers	$512M	Half-Year Revenue, H1 2026
6	Core Scientific	$635M	Revenue Run-Rate, July 2026
7	Cerebras Systems	$126M	Quarterly Cloud & Other Services Revenue, Q2 2026
8	Runpod	>$325M	ARR, Oct 2026
9	ST Telemedia Global Data Centres	INR 23.5B	India subsidiary FY2025 revenue
10	Nscale	$141M	Half-Year Revenue, H1 2026
11	Crusoe	~$276M	Estimated Annual Revenue, 2024
12	CtrlS Datacenters	INR 15.6B	Fiscal-Year Operating Revenue, FY2025
13	Scala Data Centers	BRL 916M	Fiscal-Year Net Operating Revenue, 2025
14	Ionic Digital	$48.6M	Quarterly Revenue, Q2 2026
15	atNorth	$72.2M	Icelandic Entity Fiscal-Year Revenue, FY2025
16	EcoDataCenter	SEK 685M	Fiscal-Year Net Sales, FY2025
17	ESDS Software Solution	INR 4.73B	Fiscal-Year Operating Revenue, FY2026
18	TeraWulf	$31.9M	Quarterly HPC Lease Revenue, Q2 2026
19	Sesterce	€11.9M	Historical Annual Revenue, 2024
20	Sharon AI	$1.9M	Quarterly Revenue, Q2 2026

Why are AI data center startups making so much money now?

AI data center startups are making more money because companies are buying far more computing power to train and run AI models, and the infrastructure needed to deliver that power has become increasingly expensive.

The International Energy Agency estimates that data centers consumed 485 TWh of electricity worldwide in 2025 and projects approximately 950 TWh by 2030. Electricity use at AI-focused facilities is expected to triple over that period. The agency also found that AI server power density increased elevenfold between 2020 and 2025.

That demand is already showing up in sales. CoreWeave's second-quarter revenue more than doubled year over year, Runpod's ARR climbed from $120M in January 2026 to more than $325M in October, and Nscale's first-half revenue increased more than thirteenfold compared with the previous year.

There's another reason companies are spending so much: access to AI infrastructure can be difficult to secure. Building new capacity requires grid connections, transformers, cooling equipment, GPUs, and often years of construction. Customers with demanding AI workloads increasingly want infrastructure that is ready to use rather than capacity promised for a future date.

The financial winners so far have been companies that can turn that demand into operating computing services or billable facilities. Large construction pipelines may eventually produce substantial revenue, but the conversion often takes time.

How much revenue does CoreWeave make today?

CoreWeave generated $2.58B in Q2 2026 revenue, making it the first-ranked AI data center startup and the clearest example of how large a dedicated AI cloud business has become.

According to its August 11 financial results, revenue increased from $1.21B in Q2 2025 to $2.58B in Q2 2026, approximately 112% growth. CoreWeave added roughly $1.36B in quarterly revenue in twelve months.

Its first-half revenue reached approximately $4.65B, compared with $2.19B a year earlier. Few independent infrastructure companies have managed that kind of absolute growth in such a capital-intensive business.

CoreWeave rents GPU computing capacity to AI companies and enterprises, alongside infrastructure management, networking, storage, and software services. That allows it to earn revenue across much more of the computing stack than an ordinary data center landlord.

There is an important financial catch. CoreWeave reported an operating loss of approximately $49M in Q2 2026 and a net loss of around $626M, partly reflecting the substantial financing costs of its infrastructure expansion.

Still, the sales figures leave little doubt about its commercial scale. CoreWeave has already built a multibillion-dollar quarterly AI cloud business.

Is IREN catching up with CoreWeave?

IREN is growing quickly, but CoreWeave remains well ahead on recognized AI revenue. IREN ranks second with a $1B operating revenue run-rate, while its FY2026 results recorded $128.8M in actual AI cloud revenue.

IREN's August 27 financial announcement is especially revealing because it separates three measures that can easily be confused.

The company's AI cloud services produced $128.8M in recognized FY2026 revenue, up from $16.4M in FY2025. Its annualized operating revenue run-rate reached $1B as of August 26. Meanwhile, IREN disclosed $4B in contracted ARR associated with planned 2026 capacity.

The recognized AI cloud revenue increased almost eightfold in one fiscal year. However, the operating run-rate was approximately 7.8 times the completed-year AI revenue, illustrating how quickly the business was commissioning new capacity.

The $4B contracted amount is even further ahead of the historical accounts. It depends on planned equipment becoming operational and passing customer acceptance.

IREN's customer announcements include Microsoft, Cohere, Perplexity, and other AI companies. It has also been converting infrastructure previously used for bitcoin mining into AI computing facilities.

For now, IREN has a large contracted growth opportunity and a much smaller history of earned AI cloud revenue. We would need several quarters of substantial recognized sales before describing it as a close revenue rival to CoreWeave.

How big is Lambda's AI cloud business?

Lambda ranks third among AI data center startups, with approximately $760M in estimated annualized revenue at the end of 2025. That puts it among the largest independent GPU cloud providers, although the figure is less certain than the financial results published by publicly listed competitors.

The estimate comes from Sacra's company estimates rather than a complete audited income statement. Lambda sells dedicated GPU infrastructure for AI training and inference, serving customers that need computing capacity without building their own facilities.

A business approaching $800M in annualized revenue is already operating far beyond the experimental stage of AI infrastructure. However, we cannot establish Lambda's recognized annual sales, margins, or capital expenditure with the same confidence as CoreWeave's and IREN's.

Lambda ranks behind IREN, whose selected operating run-rate is more recent and directly disclosed.

Is Applied Digital making more money from AI data centers?

Applied Digital has become a substantial AI infrastructure business, reporting $342M in revenue for Q1 FY2027, ended August 2026. It ranks fourth among AI data center startups.

The company's October 7 results showed quarterly revenue climbing from approximately $80.9M a year earlier to $341.9M, an increase of roughly 323%.

Applied Digital reported approximately $262.8M in services revenue and $79.1M from data center rental and other revenue. Its results include fit-out and service activity associated with developing large computing facilities.

A portion of its sales therefore comes from building and preparing infrastructure rather than receiving recurring rental payments from completed facilities.

Applied Digital also reported adjusted revenue of approximately $300.4M after excluding the contribution of its ChronoScale business, showing how corporate scope affects comparisons.

The company's rapid growth is clear, although recurring rental income remains a more useful indicator of its long-term hosting business than construction-related services.

Is DayOne Data Centers already a big AI infrastructure company?

DayOne Data Centers is already a major commercial operator, generating $512M in consolidated revenue during H1 2026. That puts the company fifth among AI data center startups.

The figure comes from its securities prospectus, which provides a much clearer financial picture than the capacity announcements typically released by private data center developers.

DayOne's business focuses on building and operating hyperscale data centers, particularly across Asian markets. Its customers include large cloud and technology companies seeking infrastructure that can handle increasingly power-intensive workloads.

The company's October 2026 prospectus described approximately 4.6 GW of secured resources across ten markets and around 2.3 GW of bookings, primarily linked to seven large technology and hyperscale customers.

Those figures show the size of DayOne's commercial pipeline, while its $512M of recognized first-half revenue establishes the scale of the existing business.

DayOne ranks just below Applied Digital despite reporting a larger amount over six months than Applied Digital did in one quarter. Applied Digital's faster recent expansion and different revenue mix help explain the ordering.

How much money does Core Scientific make from AI data centers?

Core Scientific ranks sixth with a $635M annualized high-density colocation revenue run-rate as of mid-July 2026. Its transition into AI hosting is already producing substantial sales.

The company's July 28 results reported $136.7M in Q2 2026 colocation revenue, compared with $77.5M in Q1 and just $10.6M in Q2 2025.

That represents an increase of approximately 1,190% year over year. Core Scientific's quarterly colocation revenue has grown almost thirteenfold as infrastructure has entered commercial service.

The company also reported approximately 1.1 GW of leased customer power capacity, associated with more than $24B in potential contracted revenue.

Yet only 437 MW was billing as of mid-July, equivalent to roughly 40% of its leased power capacity. Those operational facilities generated the reported $635M annualized revenue measure.

Quite a gap. Core Scientific has already signed contracts for substantially more power capacity than it was billing at that point.

The opportunity is large, but the infrastructure still has to be delivered and activated before that contracted capacity becomes recurring revenue.

Is Cerebras Systems making serious money from AI cloud services?

Cerebras Systems is already making substantial revenue from AI computing services, with $126M in cloud and other services revenue during Q2 2026. It ranks seventh in our AI data center startup ranking.

Cerebras is best known for developing specialized AI processors, but it also operates an inference cloud built around its own hardware.

Its August 12 financial results showed cloud and other services revenue increasing approximately 281% from the previous year. Total reported quarterly revenue reached $180.1M, but our ranking uses the narrower cloud and services figure because it better reflects commercial AI compute delivery.

The company has a different approach from GPU cloud providers that buy processors from Nvidia or AMD. Cerebras designs its own computing hardware and uses it to deliver AI inference services.

Its cloud business expanded much faster than the company's total GAAP revenue, which increased 74% year over year. Demand for Cerebras's computing services is growing rapidly, although its specialized architecture still faces competition from increasingly powerful GPU-based inference.

How quickly is Runpod growing right now?

Runpod is growing extremely fast, with annual recurring revenue rising from $120M in January 2026 to more than $325M in October. The developer-focused GPU cloud provider ranks eighth among AI data center startups.

Its October 13 announcement implies more than 170% ARR growth in approximately nine months. That means the business added more than $205M in annualized recurring revenue over that period.

Runpod also reported more than 1.2 million developers using its platform, up from approximately 495,000 a year earlier.

The developer population increased by roughly 142%, while the number of new signups reached more than 95,000 during September alone.

These figures point to a business benefiting from growing demand for smaller-scale AI infrastructure as well as enterprise workloads.

Runpod offers GPU computing for model development, training, and inference without requiring customers to operate their own servers. That makes it accessible to developers and smaller teams alongside larger organizations.

Its >$325M ARR is a strong commercial figure, although ARR should not be confused with completed annual sales. Runpod's ranking reflects the current scale of its recurring business rather than an audited full-year revenue statement.

Which AI data center startups are growing the fastest?

Nscale, Core Scientific, IREN, Cerebras Systems, Runpod, and CoreWeave have all reported exceptionally rapid growth in their selected business lines. Nscale recorded one of the largest percentage increases, while CoreWeave added far more revenue in absolute dollars.

Nscale's September 2026 IPO filing showed first-half revenue increasing from $10.4M to approximately $141M, a reported 1,252% gain. The percentage is enormous, although it partly reflects a low starting base.

Core Scientific's quarterly colocation revenue rose from $10.6M to $136.7M, an increase of roughly 1,190%.

IREN's FY2026 AI cloud revenue increased approximately 685%, while Cerebras Systems reported 281% growth in quarterly cloud and other services revenue.

Runpod's ARR climbed more than 170% between January and October 2026.

CoreWeave's growth rate was lower at approximately 112%, but its additional $1.36B of quarterly revenue represents a much larger absolute increase than the others.

We should also distinguish established operating growth from the first revenue recognized on new facilities. A data center coming online can create a spectacular year-over-year percentage increase that becomes difficult to repeat as the business matures.

Startup	Revenue comparison	Calculated growth	Measurement
Nscale	$10.4M to $141M	~1,256%	H1 revenue, YoY
Core Scientific	$10.6M to $136.7M	~1,190%	Q2 colocation revenue, YoY
IREN	$16.4M to $128.8M	~685%	Annual AI cloud revenue
Cerebras Systems	~$33M to $126M	~281%	Q2 cloud and services revenue, YoY
Runpod	$120M to >$325M	>170%	ARR, Jan–Oct 2026
CoreWeave	~$1.21B to $2.58B	~112%	Q2 consolidated revenue, YoY
Applied Digital	~$80.9M to $342M	~323%	Q1 fiscal revenue, YoY

Is Crusoe bigger than its reported revenue suggests?

Crusoe could be significantly larger today than its historical revenue estimate indicates, but we cannot verify that from its current public financial disclosures. It ranks eleventh with approximately $276M in estimated annual revenue for 2024.

The figure comes from Sacra's estimates and is less recent than the financial results available for several companies ranked above Crusoe.

Crusoe has expanded into large-scale AI computing infrastructure, developing major data center projects and offering cloud services. Its commercial plans now extend far beyond the energy-focused computing operations associated with its earlier business.

The complication is that most of the largest numbers associated with Crusoe concern financing, infrastructure investment, or contracted business rather than recognized annual sales.

A multiyear infrastructure agreement may eventually generate billions of dollars, but the revenue arrives as services are delivered over time.

Crusoe's business could now be much larger than the 2024 estimate suggests. Until stronger current financial disclosures become available, we cannot responsibly translate its development pipeline into annual revenue.

Which AI data center startups have more than $100 million in revenue?

Thirteen ranked AI data center startups have selected financial figures above $100M, although their disclosures cover different periods and financial measures.

They are CoreWeave, IREN, Lambda, Applied Digital, DayOne Data Centers, Core Scientific, Cerebras Systems, Runpod, ST Telemedia Global Data Centres, Nscale, Crusoe, CtrlS Datacenters, and Scala Data Centers.

That is 13 of the 20 numbered companies, or 65% of the ranking.

The distinction between completed annual sales and other revenue measures remains important. Cerebras Systems reported $126M for one quarter, DayOne Data Centers recorded $512M over six months, and Runpod disclosed more than $325M in ARR.

Crusoe's ~$276M figure is also historical and estimated rather than a recent audited result.

Still, the aggregate tells us something useful. The AI data center market now contains a broad group of startups with financial activity measured in hundreds of millions, well beyond the handful of best-known GPU cloud operators.

How much money are Indian AI data center startups making?

Indian AI data center startups already generate substantial revenue from hyperscale facilities, cloud services, and managed infrastructure. ST Telemedia Global Data Centres, CtrlS Datacenters, and ESDS Software Solution are the ranked companies offering the clearest examples.

ST Telemedia Global Data Centres ranks ninth using INR 23.5B in FY2025 revenue from its Indian subsidiary, equivalent to approximately $272M. ICRA reported that the subsidiary's operating income increased 17.9% that year, supported by better utilization and new data centers.

CtrlS Datacenters ranks twelfth with INR 15.6B in FY2025 operating revenue, roughly $180M. ESDS Software Solution ranks seventeenth with INR 4.73B in FY2026 operating revenue, about $53.8M.

These businesses serve somewhat different customers. ST Telemedia Global Data Centres focuses on hyperscale and colocation services, CtrlS combines infrastructure with managed services, and ESDS operates cloud and managed data center services.

Their revenue also includes workloads beyond AI. Enterprise hosting, banking systems, conventional cloud infrastructure, and other digital services remain part of the business, so we cannot isolate their AI-specific revenue from the reported figures.

Are European AI data center startups making real money yet?

European AI data center startups are already generating meaningful revenue, led by Nscale, with atNorth, EcoDataCenter, and Sesterce offering additional evidence of commercial operations.

Nscale ranks tenth with $141M in H1 2026 revenue. Its September prospectus disclosed approximately $33M in full-year 2025 revenue, showing how quickly the company expanded during the following six months.

atNorth ranks fifteenth with $72.2M in FY2025 revenue from its Icelandic entity. Icelandic company accounts show that this business was growing rapidly, although the figure does not cover atNorth's entire international operation.

EcoDataCenter ranks sixteenth with SEK 685M in FY2025 net sales, approximately $69.9M. The Swedish operator focuses on high-density and energy-efficient computing facilities.

Sesterce ranks nineteenth with €11.9M in historical 2024 annual revenue. Its more recent AI data center plans may be substantial, but the available figure does not establish how much revenue those newer projects currently produce.

The Nordic companies are particularly interesting because electricity access and cooling conditions can support large computing operations. Their commercial advantage will depend on how much new capacity they can bring online and fill with paying customers.

Which major AI data center startups still have no disclosed revenue?

Some of the most ambitious AI data center startups remain unranked because their public figures describe infrastructure capacity, contracts, or development plans rather than comparable revenue.

Aligned Data Centers, Vantage Data Centers, AirTrunk, Princeton Digital Group, EdgeCore Digital Infrastructure, Firmus, Fluidstack, Lancium, DataVolt, Volta, Global Technical Realty, Yondr Group, DC BLOX, Metrobloks, Empyrion Digital, Echelon Data Centres, Fermi America, TensorWave, Starcloud, PowerBridge, STACK Infrastructure, Digital Edge, T5 Data Centers, and Éclairion all fall into this group.

Aligned Data Centers is particularly notable. A consortium including BlackRock's Global Infrastructure Partners and MGX completed its acquisition in July 2026 at an enterprise value of approximately $40B. The company reported 51 campuses and more than 6.4 GW of operating and planned capacity.

That acquisition occurred recently enough for Aligned to remain eligible under our startup rule, but the transaction price tells us nothing precise about annual revenue.

Similarly, Lancium reported 4 GW of leased capacity, EdgeCore described more than 1.8 GW delivered or under development, and Princeton Digital Group disclosed a portfolio exceeding 1.1 GW.

These companies may eventually generate very large sales. For now, assigning them numerical revenue positions would require assumptions about utilization, pricing, or future operations that the evidence cannot support.

Unranked startup	Reported commercial scale	Main limitation
Aligned Data Centers	6.4 GW operating and planned	Mixed current and future capacity
Vantage Data Centers	AI deployment at CWL1 campus	Revenue not disclosed
AirTrunk	>150 MW contracted at Johor JHB1	Single-facility figure
Princeton Digital Group	>1.1 GW portfolio capacity	Different development stages
EdgeCore Digital Infrastructure	>1.8 GW delivered or developing	Includes unfinished projects
Firmus	>900 MW contracted capacity	Future AI factory deployments
Fluidstack	300 MW contracted capacity	Future-oriented leasing commitment
Lancium	4 GW leased capacity	Power-secured campuses
DataVolt	100 MW under construction	Commercial availability pending

Do AI data center landlords make as much money as GPU cloud startups, even with more capacity?

AI data center landlords can generate hundreds of millions in revenue, but integrated GPU cloud startups currently occupy the highest positions in our ranking. Having more physical capacity does not guarantee higher sales.

CoreWeave, IREN, and Lambda sell computing services built around GPUs and related infrastructure. A landlord or colocation provider earns money from space, electricity, cooling, connectivity, and long-term hosting.

DayOne Data Centers reported $512M in half-year revenue, while Core Scientific disclosed a $635M annualized colocation run-rate. CtrlS Datacenters and Scala Data Centers also have substantial revenue from physical infrastructure and related services.

The business models carry different costs. GPU cloud providers can charge for the computing hardware and services inside a facility, but must finance expensive processors that depreciate quickly. Data center landlords generally rely more on occupancy, power supply, and long-term leases.

Even among landlords, announced capacity can be far ahead of revenue-producing infrastructure. As seen above, Core Scientific was billing for 437 MW out of approximately 1.1 GW of leased customer power capacity in mid-July 2026, leaving roughly 60% of that leased capacity not yet billing.

Compare that with Lancium's 4 GW of leased capacity or Aligned Data Centers' 6.4 GW operating and planned portfolio. Neither disclosure establishes annual sales without knowing how much capacity is operational, what customers pay, and when billing begins.

The largest announced data center portfolios may eventually produce enormous revenue. Today, integrated GPU cloud providers have the stronger disclosed financial scale.

Are AI data center startups profitable yet?

AI data center startups can generate enormous revenue while still losing money, and several of the largest companies have yet to demonstrate consistent profitability.

CoreWeave reported approximately $2.58B in Q2 2026 revenue alongside a $49M operating loss and a net loss of roughly $626M.

Its interest expense, net, was approximately $640M for the quarter. That financing burden helps explain why rapid sales growth does not automatically translate into net profits.

IREN reported a FY2026 net loss of approximately $702.6M against $707M in consolidated revenue. Its accounts included $638.8M of noncash impairments, primarily related to retiring bitcoin mining equipment as the company expanded its AI cloud business.

Cerebras Systems also reported substantial GAAP operating losses during Q2 2026, reflecting its investment in specialized processors, infrastructure, and commercial expansion.

These losses have different causes, so comparing net profit margins alone would give an incomplete picture.

For GPU cloud providers, the financial outcome depends heavily on hardware utilization, depreciation, customer pricing, and borrowing costs. Physical infrastructure operators face major construction expenditure but can sometimes secure longer-term contractual income.

The sector has already demonstrated that customers will spend heavily on AI computing. The harder test now is whether that revenue can support the cost of building and financing the infrastructure.

Could electricity shortages stop AI data center startups from growing?

Electricity shortages and slow grid connections could seriously delay AI data center revenue growth, particularly for startups developing large new campuses.

The IEA's updated energy outlook estimates that global data center electricity consumption could nearly double between 2025 and 2030. The agency also warns that grid connections, transformers, semiconductors, and other equipment shortages may slow construction.

AI data centers are especially demanding because high-density computing racks require considerable electrical power and cooling.

An interesting detail from the IEA is that AI server power density increased elevenfold between 2020 and 2025, with another fourfold increase projected by 2027. Infrastructure that was suitable for conventional cloud computing may need substantial upgrades to accommodate newer accelerator systems.

A company can secure a customer contract and acquire computing equipment yet still be unable to earn the expected revenue until power and cooling systems are operational. For developers planning gigawatt-scale campuses, delays of several months can postpone substantial sales.

Can smaller AI data center startups compete with CoreWeave?

Smaller AI data center startups can compete successfully with CoreWeave by serving customers who want different hardware, more flexible access, or specialized infrastructure.

Runpod is a good example. Its developer-focused services have attracted more than 1.2 million developers and generated >$325M ARR, showing that smaller AI teams and enterprises represent a substantial market.

Cerebras Systems has chosen a different approach, selling inference services powered by its own processors. That gives customers an alternative to conventional GPU-based computing for certain workloads.

Regional infrastructure companies also have opportunities. atNorth and EcoDataCenter can compete for customers that prioritize local electricity availability, hosting requirements, and infrastructure services over a globally distributed GPU cloud.

CoreWeave's revenue lead is clear, but smaller operators can build substantial businesses around customers that need more specialized services. Their challenge will be keeping costs competitive as larger providers expand and computing hardware changes.

What could change the AI data center revenue ranking next?

The AI data center startup ranking could change quickly as contracted capacity becomes operational, particularly at IREN, Core Scientific, Nscale, and other companies expanding their computing infrastructure.

IREN's disclosed $4B contracted ARR for planned 2026 capacity represents a substantial opportunity beyond its $1B operating run-rate. How much of that future capacity begins generating recognized revenue will be critical.

Nscale offers another important case. Its H1 2026 revenue of $141M was already more than four times its $33M full-year revenue for 2025. Continued growth would make its annual financial results increasingly comparable with those of larger operators.

Developers such as Aligned Data Centers, AirTrunk, Vantage Data Centers, and Princeton Digital Group could also become easier to assess if they publish consolidated revenue figures.

The most revealing future disclosures will be completed-quarter revenue, audited annual sales, and recurring income from commissioned facilities. Those numbers will show how much of the industry's enormous contracted pipeline is actually turning into business.

We would pay less attention to another record funding round or a new gigawatt campus announcement unless it changes operating capacity or customer revenue.

So, what are the top AI data center startups by revenue today?

CoreWeave is currently the leading AI data center startup by revenue, followed by IREN, Lambda, Applied Digital, and DayOne Data Centers. CoreWeave's $2.58B in Q2 2026 sales puts it firmly ahead in recognized commercial scale.

Core Scientific, Cerebras Systems, Runpod, ST Telemedia Global Data Centres, and Nscale complete the top ten.

Crusoe, CtrlS Datacenters, Scala Data Centers, Ionic Digital, atNorth, EcoDataCenter, ESDS Software Solution, TeraWulf, Sesterce, and Sharon AI occupy the remaining ranked positions.

Our analysis also shows that several major AI infrastructure developers are still missing comparable revenue disclosures despite operating or planning enormous facilities.

The market has already produced a handful of very large businesses, but much of its future revenue remains tied to capacity that has yet to be commissioned.

For now, CoreWeave has the strongest claim to revenue leadership. IREN and other rapidly growing operators have substantial room to expand, while many physical data center developers are still working to convert their capacity pipelines into recurring sales.