Companies see revenues grow, losses widen
Neoclouds CoreWeave, Nebius, and Cerebras posted their Q2 2026 earnings results this week.
The AI cloud providers all saw revenues accelerate as demand for AI services increased, but also posted growing losses as they continue to quickly build out their data center capacity.
CoreWeave: Revenue accelerating, losses widening
AI cloud firm CoreWeave continued to accelerate both its revenues and losses this quarter.
The company posted revenues of $2.575 billion for Q2 2026, up from $1.212bn in Q2 2025. The company posted a $49 million operating loss, a swing from $19m in operating income the year prior.
Despite adjusted operating income of $128 million, the company posted a net loss of $626 million, increasing from $290m the same quarter last year. Adjusted EBITDA for the quarter was $1.51 billion, up from $752 million in Q2 2025.
“CoreWeave reached an important inflection point this quarter as our scale began to translate into expanding operating leverage. Customer demand is accelerating, as enterprise adoption broadens and we continue to deepen our technology platform,” said CoreWeave CEO and co-founder, Michael Intrator. “CoreWeave is built on the conviction that AI is foundational to every industry and that realizing its full potential requires a purpose-built platform. This quarter reinforced that conviction.”
CoreWeave’s capex in Q2 totaled $9.4 billion. Total capex for the year is expected to reach $35-39bn.
The three-month period saw CoreWeave expand its active power by nearly 500MW – including 300MW in June – to reach 1.5GW across 51 active data centers. Eight facilities have been added in 2026, and the company expects to end the year with 1.85GW in live capacity.
The company has 4.2GW in contracted power, and is targeting 8GW by 2030. Its first self-build project is due to come online later this year.
CoreWeave’s revenue backlog was approximately $104bn as of June 30, not including more than $25bn of net new customer commitments added in early Q3.
Intrator said the company had grown its managed inference platform from $1 million to more than $100m ARR and expects to exit 2026 with at least $250m of managed inference ARR.
During the earnings call, CoreWeave execs noted demand for older generations of Nvidia GPUs remains solid as it offers installed and energized production-grade compute available quickly. The company said it recently signed an A100 contract that extends into 2029 at an “attractive price” despite the A100s coming out in 2020.
On the ongoing spate of data center moratoriums being introduced in the US and whether it will impact the company’s build-out, Intrator said: “They’re not going to impact the demand for this infrastructure. They are going to impact where this infrastructure gets built. And so our approach to how you engage with the stakeholders is that you have to be extremely collaborative with the communities that ultimately host the infrastructure. Ultimately, at the end of the day, it is in our interest, and it is in their interest for us to be good neighbors of their community.”
Nebius: massive growth, profit a challenge
Nebius group posted revenues of $582.3 million for the quarter, a 454 percent increase on Q2 2025. Adjusted EBITDA swung from a $21m loss in Q2 2025 to $236.2m in 2026.
Adjusted net losses dropped from $91.5m to $33.2m, while net income dropped from $584.4m to a $190.4m loss in Q2 2026 – though much of 2025’s figure was gains from a revaluation of investments in equity securities.
Almost all the group’s revenue is derived from its AI cloud business. Nebius AI cloud revenue grew 514 percent Year-on-Year (YoY) to $575m.
Nebius has been on a rapid expansion spree, and costs have risen as a result. Total operating costs and expenses jumped from $216.3m in Q2025 to $758.2m for Q2 2026.
This week saw the company announce deployments with Vantage in Wales, Greenergy in Estonia, and plans for a second self-built facility at its campus in Mäntsälä, Finland.
In a letter to investors, founder and CEO Arkady Volozh said Nebius closed four deals valued at more than $1bn each in the quarter, at a yield of $20-25 million per megawatt.
Volozh also said Nebius has contracted another 1GW in capacity in the quarter, saying the firm anticipates ending 2026 with 5GW of contracted power. The company plans to deploy more than 1GW of compute per year, starting in 2027.
Notably, Volozh also said that production inference workloads more than tripled in Q2.
Cerebras: Cloud revenues up, losses grow
Cerebras, which develops its own wafer-scale chips and offers them as an on-premise or cloud-based solution, announced that its inference cloud business nearly quadrupled YoY. The company, however, posted a large net loss.
“This was an outstanding quarter for Cerebras. Core revenue more than doubled to $210 million, and our cloud business nearly quadrupled year-over-year,” said Andrew Feldman, Cerebras co-founder and CEO. “Speed changes what AI can do. It makes AI more useful, more productive, and opens entirely new markets. As a result, the demand for fast inference is enormous, and Cerebras is scaling to meet it, securing more data center capacity, expanding manufacturing, and growing with customers and partners including OpenAI, AWS, AMD, and CrowdStrike.”
The firm posted GAAP total revenues of $180.1m, up 74 percent YoY. Within that, the firm saw GAAP cloud and other services revenues of $126.0m, up 281 percent YoY.
Cerebras saw gross profits total $25.56m down from $32.1m the year previous. The company, however, posted a net loss of $450.4m, down from $309.5m in net income in Q2 2025. Adjusted EBITDA for the quarter was a $53.1m loss, compared to a $38.3m loss the year prior.
The quarter saw data center capacity live and under contract for delivery by the end of 2027 increase to more than 600MW. The firm says its total pipeline is in the gigawatts.
The firm secured contracts with AI coding companies including Cognition and Lovable during the three-month period.
https://www.datacenterdynamics.com/en/news/neocloud-results-q2-2026-coreweave-nebius-cerebras/
