Evercore ISI reiterates CoreWeave stock rating on AI demand outlook By Investing.com

Evercore ISI reiterated an Outperform rating and $150 price target on CoreWeave after the Blackstone-Alphabet joint venture, arguing the new TPU cloud provider competes on TPU capacity rather than CoreWeave's GPU offerings. CoreWeave fell 5% on the announcement, closing at $99.30, but the firm pointed to recent Meta and Anthropic expansion deals as evidence demand remains intact. The article also highlights CoreWeave's $3.1 billion AI infrastructure loan facility and several recent analyst views, reinforcing the stock's central role in AI cloud infrastructure.

Analysis

The market is treating the new TPU venture as if it were a direct substitute for CoreWeave’s economic moat, but the more relevant read is that it expands the overall AI compute TAM while fragmenting pricing power at the margin. GPU supply remains the scarce, latency-sensitive layer for frontier training and inference workflows that need flexibility across models and frameworks; TPUs are more likely to siphon off standardized workloads than displace the highest-value demand. That means the near-term read-through is not “CoreWeave demand breaks,” but “incremental competition caps multiple expansion,” especially if hyperscalers start using bespoke silicon to negotiate better GPU terms.

The second-order beneficiary is META: it now has optionality to arbitrage compute across vendors, which should improve procurement leverage and reduce exposure to any single infrastructure provider’s capacity bottlenecks. GOOGL also benefits strategically because the JV externalizes TPU adoption without fully cannibalizing Cloud; it turns a captive hardware stack into a broader distribution channel, which matters if AI capex intensity stays elevated for several years. BX is a quieter winner as the capital partner to an infrastructure theme with long-duration cash flows, but the key question is whether the JV returns exceed the cost of funding an asset-heavy, depreciation-prone stack in a market that is already moving down the stack toward commoditized compute.

The consensus is probably overestimating how quickly TPU availability will compress CRWV’s growth, while underestimating the risk that financing markets tighten if investor sentiment shifts from scarcity to overbuild. CoreWeave’s biggest vulnerability is not one announcement; it is a 6-12 month sequence where hyperscalers, model labs, and private capital all add capacity simultaneously, causing utilization to normalize before backlog fully monetizes. Conversely, if Nvidia prints strong data-center demand and AI capex guidance stays aggressive, CRWV can rip higher on the same “more compute needed” logic that currently supports the stock.

Key Decisions for Investors

• Long CRWV only on pullbacks into the low-$90s with a 6-9 month horizon; use a tight risk budget because upside likely depends on continued scarcity pricing, while downside accelerates if TPU supply broadens faster than backlog converts.
• Buy GOOGL vs. short CRWV as a relative-value pair for 1-3 months: GOOGL gains from optionality and ecosystem control, while CRWV is more exposed to multiple compression if the market concludes AI infra is becoming more substitutable.
• Sell out-of-the-money CRWV calls or structure a call spread into the next 1-2 earnings windows; implied volatility should stay elevated, but the catalyst path is asymmetric to the downside if competitive concerns resurface.
• Overweight META on a 3-6 month basis versus other AI infrastructure names: it improves bargaining power over compute vendors and can shift marginal workloads to the cheapest acceptable stack, supporting margins if AI spend keeps rising.
• Avoid chasing BX as a pure AI infrastructure proxy; treat it as a funding beneficiary only if the JV proves it can earn above-cost returns on a durable 2-3 year deployment cycle.