FundaAI

FundaAI

Research|Interesting Observations During GTC

Coreweave, Manus, Coding Agent, Mobile Agent, Recommendation Algorithms, Glean vs Office Copilot

FundaAI's avatar
FundaAI
Mar 27, 2025
∙ Paid
1
Share

Disclaimer: The content provided in this newsletter is for informational purposes only and does not constitute investment advice. We are not registered investment advisors, and nothing in this newsletter should be construed as a recommendation to buy or sell any securities. Always do your own research and consult with a licensed financial professional before making any investment decisions.

Over the past two weeks, I've met with many companies and friends in the Bay Area. I'd like to first write about AI Hyperscaler, how hot Manus is, how big Coding Agents could become, the latest advances in recommendation algorithms, mobile Agents, Glean, and Office Copilot.

Later, I'll write a more detailed piece on large language models and technological changes.

Let's start with the hottest topic: Coreweave's IPO and the AI Hyperscalers.

Negative Sentiment in AI Hyperscalers

Coreweave is set to go public on Friday, but the IPO pricing has been challenging, with a last-minute 25% reduction dropping the valuation from a planned $35B to $23B.

In yesterday's PDIE guidance to the sell-side, many of Coreweave's numbers were notably lower than what they had previously communicated to the market before the IPO.

Most notably, the CAPEX figures that sell-side analysts who spoke with Coreweave are providing for 2025-2027 are generally $18B/$14-15B/$13-14B USD, indicating consecutive declines in 2026 and 2027. Of course, this should be taken as a reference point, since companies and the sell-side typically leave some buffer before an IPO.

Actually, before the IPO, Coreweave's 2026 CAPEX guidance in the private market already showed a decline, which could become a potential issue for semiconductor stocks. But I didn't expect 2027 CAPEX to be even lower, nearly 30% below what was communicated in the private market.

Beyond CAPEX guidance, revenue expectations were also significantly reduced, by roughly 20-30%.

In fact, our research in early March found that Microsoft had cut about $300M in Coreweave capacity. Although they didn't cancel the contract, delaying capacity is still a very negative signal.

We were among the first in the industry to write about why Microsoft delayed two data centers.

Research|Why Microsoft Has Delayed Two Data Centers

Research|Why Microsoft Has Delayed Two Data Centers

FundamentalBottom
·
Feb 22
Read full story

Beyond the Coreweave IPO, there were many negative signals for AI Hyperscalers in the Bay Area last week.

For example, Coreweave and Nebius billboards are everywhere on Bay Area highways, and you can even see Nebius advertisements in Uber rides, raising more concerns.

A friend who works at a Bay Area AI Hyperscaler startup told me, "When a leading company starts emphasizing that their core differentiator is 24/7 service, it means everyone is struggling to find real differentiation."

This may be exactly what's happening in the industry.

How Hot Manus Is

Keep reading with a 7-day free trial

Subscribe to FundaAI to keep reading this post and get 7 days of free access to the full post archives.

Already a paid subscriber? Sign in
© 2025 FundaAI
Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture