CoreWeave, an NYC startup that began life as an Ethereum mining venture has secured a large amount of funding in order to continue its transition into a general-purpose platform for cloud computing.
CoreWeave has announced that it has raised $221 million as part of a Series B round of funding led by Magnetar Capital. Other participants include Nvidia, the former GitHub chief Nat Friedman, and ex-Apple executive Daniel Gross. Magnetar contributed an initial $111 million. The remainder of the investment is split between Nvidia and Friedman. Nvidia’s spokesperson said the investment was a “deepening of their partnership with CoreWeave”.
CEO Mike Intrator stated that the tranche, which values CoreWeave as $2 billion pre-money, and brings the total raised by the company to $371 millions, will be used to support CoreWeave’s expansion of data centers in the United States, with the opening two new centers this coming year. CoreWeave operates five data centers in North America.
CoreWeave, founded in 2017, was created by Intrator, Brian Venturo, and Brannin McCBee to fill what they perceived as a “void” in the cloud computing market. Venturo is a hobbyist Ethereum miner who bought GPUs at a discount from insolvent cryptocurrency mines. He chose Nvidia for its increased memory.
CoreWeave focused its initial efforts exclusively on cryptocurrency-related applications. In the last few decades, CoreWeave has focused on general-purpose computing technologies and generative AI technology, such as text-generating AI models.
CoreWeave offers over a dozen Nvidia SKUs in the cloud including H100s (for AI and machine learning), A100s (for visual effects and rendering), A40s (for batch processing) and RTX A6000s.
Intrator told TechCrunch via email that “our clients include generative AI businesses, such as Tarteel AI or Anlatan – the creators NovelAI – and we’ve also supported a range open source AI and Machine Learning projects, like EleutherAI or Stability AI’s Stable Diffusion.” “We also work with a number of notable VFX and animation studios such as Spire Animation, and partner closely with 3D streaming and ‘metaverse’ companies such as PureWeb.”
It’s tough for any cloud provider to compete with the incumbents in the space — i.e., Google, Amazon and Microsoft. AWS generated $80.1 billion in revenue, while Google Cloud and Azure brought in $75.3 billion and $25.28 billion respectively.
Obviously, these figures are multiples of CoreWeave’s value, let alone the war chest.
Statista’s analysis is a good way to drive home the point. report AWS held a market share of 32% in the fourth quarter 2022. Azure held a share of 23% and Google Cloud held a share 10%.
This is not to say that smaller players can’t succeed. There’s success stories like Paperspace, Scaleway and DigitalOcean (despite its ups and downs), as well as newer entrants like Clever Cloud and Vultr.
CoreWeave also seems to be a good example of this. The startup secured funding despite the market’s rough quarter. As my colleague Ron Miller wrote, companies looked for ways to cut back on spending in an uncertain economy, slowing the market to 21% growth — a precipitous drop from the 36% growth in the year prior.
“We have over 1,000 customers across our four key verticals — machine learning and AI, batch processing, pixel streaming and visual effects and rendering,” Intrator said.
CoreWeave makes the case that the dominant cloud providers — Google Cloud, Azure and AWS — have failed to meet the demand for generative AI in particular with their “legacy cloud infrastructure.” Them’s fighting words, to be sure, especially as AWS launches a dedicated service for serving text-generating models. But in Intrator’s eyes, the incumbents aren’t set up to meet the demand of thousands of new AI companies clamoring for GPUs — at least not at CoreWeave’s (ostensibly lower) prices.
CoreWeave claims its hardware for inference — i.e., serving AI models — is industry leading, able to “autoscale” within three seconds. It also touts newer products, including Nvidia’s H100 server. platform.
“For a while now, technology decision-makers have faced the increasingly complex — and costly — task of deploying their highly specialized compute tasks supporting modern AI and machine learning applications to more generalized cloud computer providers,” Intrator said. CoreWeave understands that this demand will require a deep investment in scalable, attainable capacity for next-generation AI firms.
CoreWeave is trying to differentiate itself from other companies by offering services like its accelerator program which was launched in late October. (Introducer says it has more than 30 members.) The accelerator — which operates on an open-ended basis, with no deadlines — provides companies compute credits in addition to discounts and other hardware resources on the CoreWeave cloud.
Intrator says this new tranche will see more efforts like these.
He said that with CoreWeave’s emergence and this new funding, the company can now provide more companies with customized solutions which can outperform legacy providers. “While deep-learning image generation and large-scale language models have existed for a long time, their prominence is driving an intense race to secure processing power in order to develop ever more powerful apps. CoreWeave understands that this demand will require a deep investment in scalable, attainable capacity for next-generation innovative AI firms.”
The money will also be used to expand CoreWeave’s team. The company employs “just over” 115 people now — up 150% in the last 12 months — thanks in part to its acquisition of the cloud rendering platform Conductor Technologies in January, and Intrator says that the plan is to keep hiring “throughout the year.”
The question is, of course, whether CoreWeave can maintain its impressive momentum — particularly if the generative AI bubble bursts anytime soon. Friedman, Gross and others seem to be convinced of the strategy. They sent an email with this statement:
CoreWeave builds the grid of the new economy. AI is the new electric power. We’ve worked for Apple and Microsoft, invested in companies like Stripe and Figma, and Airtable, and can confidently state that CoreWeave is moving at an unprecedented pace. It’s a daily sprint to victory, which is evident in the quality and number of CoreWeave customers. CoreWeave has been preparing for years to scale up the infrastructure and culture in order to meet the demand for AI inference.
There are some reasons to be optimistic. According to a recent survey by ESG, 59% of companies plan to spend more on public cloud apps in 2023 while 56% expect that their public cloud infrastructure services spending will increase.