Bernstein maintains CoreWeave at Underperform, saying Google + Blackstone AI cloud will intensify competitive pressure

Overview

This is a Bernstein company research report on CoreWeave, Inc. The key event is Google and Blackstone's plan to form a new AI cloud company. The report argues that this combination brings together Google's TPU, software, and services full-stack capabilities with Blackstone's data center capital, development pipeline, and physical assets, creating direct competition for CoreWeave and other neoclouds, and reinforcing the view that hyperscale cloud providers are the clearest winners in the AI cloud race.

Core views

Bernstein maintains its Underperform rating on CRWV and its USD 67 target price. The report's core view is that CoreWeave may still sign more contracts in the near term, and its share price may remain supported by contract-signing momentum; however, from a long-term fundamental perspective, its business model, ability to keep signing large hyperscale customer contracts, long-term margins, access to power resources, and software moat all face pressure. As data center supply becomes more abundant after 2028, hyperscale cloud providers may compete directly for enterprise AI cloud demand, putting downward pressure on pricing and margins for new CoreWeave deals.

Analysis framework

The report uses an event-driven and competitive landscape analysis, comparing Google's and Blackstone's new AI cloud plan with CoreWeave's key debate points one by one: capital cost, long-term margins, power and data center supply, product-stack differentiation, and enterprise customer penetration. On valuation, Bernstein uses an EV/adjusted EBIT multiple approach to set the target price.

Methodology notes

• Valuation methodsEV/Adjusted EBIT multiple approach

  Enterprise value relative to adjusted EBIT

  The report says the USD 67 target price is derived from 28.4x 2027E adjusted EBIT per share of USD 5.81.

• Competitive analysisHyperscale cloud provider vs. neocloud competition framework

  Comparison of cost, product stack, capital, and data center resources

  The report compares the Google + Blackstone combination with CoreWeave's core capabilities and argues that hyperscale cloud providers have structural advantages in full-stack products, capital cost, data center assets, and power resources.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• CoreWeave, Inc. (CRWV.US)

  Covered company and primary impacted asset

  Strengths

  May continue to sign more AI cloud contracts in the near term, has already contracted about 3.5 GW of power, and the share price may be supported by contract-signing momentum.

  Weaknesses

  Long-term revenue expectations are below market expectations, and its business model, long-term margins, capital costs, power access, and software moat are being questioned.

  Comparison

  Compared with Google + Blackstone and other hyperscale cloud providers, CoreWeave is at a disadvantage in capital, data center assets, full-stack software services, and cost structure.

  Risks

  Intensifying competition could push down new-contract pricing and margins, and weaken its ability to expand into enterprise customers.

• Google + Blackstone AI cloud

  Potential direct competitor

  Strengths

  Google provides TPU, software, and services full-stack capabilities; Blackstone provides data center capital, development pipeline, real estate, and operational experience.

  Weaknesses

  TPU functionality is more constrained by Google's ecosystem than NVIDIA GPUs, and the new AI cloud remains relatively small in the short term versus CRWV's expected 2027 scale.

  Comparison

  The report believes this combination is superior to neoclouds in cost, product, capital, and physical assets.

  Risks

  There is still uncertainty around execution timing, capacity ramp-up, and customer migration pace.

Key data

• RatingUnderperformBernstein maintains its underweight rating on CoreWeave.
• Target price67.00 USDBased on 28.4x 2027E adjusted EBIT per share of USD 5.81.
• Google + Blackstone equity investmentAbout USD 5 billionThe WSJ reported that Blackstone will provide about USD 5 billion in equity.
• New AI cloud target capacityAbout 500 MW, planned for 2027 launchThe report says this capacity is about half of CRWV's current active capacity.
• CRWV contracted powerAbout 3.5 GWThe report says most of this is expected to come online by the end of 2027.
• Blackstone data center assetsAbout USD 150 billionThe report says Blackstone is the world's largest data center investor.
• Blackstone development pipelineAbout USD 160 billionThe report roughly estimates this as about 10 GW of existing capacity and more than 10 GW of additional future capacity.

Impact & implications

For CoreWeave, the Google + Blackstone plan may not have a material impact on 2026 results in the short term, but its strategic significance is greater: hyperscale cloud providers are directly entering AI cloud supply with their own technology stacks, capital, and data center resources. This could absorb part of AI cloud demand, weaken neocloud pricing power, and make CoreWeave's transition toward a more enterprise-heavy customer mix more difficult.

Risks

• Hyperscale cloud providers building or jointly launching AI clouds may absorb demand and push down CoreWeave's new-deal pricing and margins.
• As data center supply becomes more abundant after 2028, CoreWeave's ability to continue signing large hyperscale contracts may decline.
• CoreWeave's existing software moat may struggle to withstand the scale and investment capacity of hyperscale cloud providers such as Google.
• If hyperscaler self-built capacity launches later than expected, CoreWeave's dedicated software platform continues to strengthen, or data center power supply remains tight, these could be upside risks to the report's bearish view.

What to watch

• The actual 2027 launch progress of the Google + Blackstone AI cloud company's 500 MW capacity.
• Changes in the pricing, term, and margin profile of CoreWeave's new large-customer contracts.
• The share of CRWV's roughly 3.5 GW contracted power that actually comes online on schedule.
• Whether enterprise-customer mix increases and can offset hyperscale customer competition.
• The pace of TPU adoption among AI inference customers and whether its per-token cost advantage expands.