GPU Clouds and Neoclouds: CoreWeave, Lambda, Nebius — 2026-10-28
Elon Musk’s warning of a "guppy-level death" for neoclouds triggered a selloff in CoreWeave and Nebius, while SpaceX’s $40 billion GPU debt deal highlights the massive capital requirements facing the sector. Despite financing headwinds, CoreWeave announced new capacity in Denmark, and Nebius CEO Arkady Volozh confirmed that asset-backed lending has tightened, though the company remains funded for its gigawatt expansion.
GPU Clouds and Neoclouds: CoreWeave, Lambda, Nebius — 2026-10-28
Top developments
Musk Warns of Neocloud "Guppy-Level Death"; Stocks Slide On October 26, Elon Musk stated that more than 200 neocloud providers face "guppy-level death and consolidation" as lenders stop financing companies without proven revenue. Following this warning, CoreWeave shares dropped 5%, IREN slid 3%, and Nebius slipped 2% on October 26. Musk noted that SpaceX itself has closed a $40 billion debt deal to purchase GPUs arriving in the first half of 2027, illustrating the sheer scale of capital required to compete.
CoreWeave Expands to Denmark; Stocks Climb On October 27, CoreWeave announced the activation of its first data center in Denmark and a network push across seven countries. This expansion drove AI cloud stocks higher, with CoreWeave and Nebius climbing 4% and Oracle ticking up. The move comes as the company seeks to convert contracted power into active power, though questions remain regarding whether the buildout can outrun the financing costs.
Nebius CEO Admits Financing Has Tightened Nebius CEO Arkady Volozh acknowledged on October 26 that asset-backed financing and debt markets have "temporarily dried up" for neoclouds. Despite this, Volozh stated that Nebius has sufficient capital to build its gigawatt-scale infrastructure. The company reported receiving $9 billion in customer prepayments in 2026 alone, which helps reduce the risk associated with infrastructure build-outs.
General Compute Raises $100 Million Series A In a counter-narrative to the neocloud distress story, General Compute raised $100 million in a Series A round on October 27. The company is betting on inference chips from outside the Nvidia ecosystem, suggesting that investors are still willing to fund specialized compute plays if they offer differentiation from the crowded Nvidia-based rental market.
Local view
Chinese Media Reports on GPU Price Volatility and Utilization Local Chinese tech media outlets, including Huxiu and Sina, are highlighting a stark contrast in the global GPU market. Reports from October 26-27 discuss a "three-year reversal" where H100s have gone from scarce to seeing idle capacity in some domestic smart computing centers, leading to rental price drops. However, other reports note that older A100 GPUs are still renting at higher rates than earlier in the year, with some data suggesting A100 rental prices hitting $1.09/hr. This reflects a bifurcated market where legacy hardware retains value due to specific software compatibility, while newer hardware faces pricing pressure from oversupply in certain regions.
Context & numbers
- CoreWeave Backlog & Debt: CoreWeave’s contracted backlog reached $104.2 billion, with its total debt pile estimated at $51 billion. The company doubled revenue to $2.58 billion but carries significant leverage.
- GPU Rental Prices: The median H100 rental price has fluctuated, with some indices showing it around $3.38–$3.70 per hour. B200 rental prices are median at $8.19 per GPU-hour.
- Nebius Contracted Revenue: Nebius reports more than $40 billion in additional contracted revenue from investment-grade customers such as Microsoft and Meta.
- SpaceX Debt: SpaceX closed $40 billion in mostly callable debt specifically for GPU purchases.
On the radar
- Anthropic IPO Filing: Reports indicate Anthropic has formally filed a registration statement with the US SEC for a Nasdaq IPO, which could signal broader market confidence in AI infrastructure plays.
- Nvidia’s Role as Financier: Investors are scrutinizing Nvidia’s role as an industry financier, with CNBC noting that the chipmaker faces its "first big test" as neoclouds take a market hit.
- Utilization Rates: AIMultiple data shows confirmed availability is lowest for MI300X (7%) and B300 (11%), while H100 availability is at 33%. This scarcity in newer chips may drive further price volatility.
This content was collected, curated, and summarized entirely by AI — including how and what to gather. It may contain inaccuracies. Crew does not guarantee the accuracy of any information presented here. Always verify facts on your own before acting on them. Crew assumes no legal liability for any consequences arising from reliance on this content.
Powered by
