CoreWeave & Intel – Have and Have Not

CoreWeave Q3 2025 – Have.

• CoreWeave was saved from being used as a sign of peak demand when weak guidance was rescued by a huge backlog, clearly showing that the top of the AI boom has yet to arrive.
• Q3 2025 revenue / adj-EPS were $1.36bn / LOSS$0.22 ahead of forecasts of $1.29bn / LOSS$0.57, but guidance was cut.
• FY 2025 revenues will now be $5.05bn – $5.15bn ($5.10bn), down $150m from $5.15bn – $5.35bn ($5.25bn) that was set at the Q2 2025 results.
• This was explained by a temporary delay at a third-party data centre developer whose roll-out is behind schedule.
• This means that CoreWeave will be more capacity-constrained than it expected, resulting in the $150m revenue miss.
• Normally, the doomsayers would immediately jump on this as a sign of the top, but the revenue backlog clearly points to the veracity of CoreWeave’s statements.
• Here, remaining performance obligations (RPOs) jumped to $55.6bn, almost double what they were just three months ago.
• This is a clear signal that demand remains as strong as ever, buoyed by the $14.2bn order that it has received from Meta Platforms.
• Consequently, I do not see this as a peak in demand, but instead further evidence that demand continues to outstrip supply.
• This means that the AI boom remains as strong as ever and is likely to remain so for at least a few more quarters.
• Hence, I expect that Nvidia will report another strong quarter next week and give guidance that is just above expectations.
• This will be enough to help it maintain its current valuation, but is unlikely to result in another leg up.
• This is why I prefer the adjacencies of inference at the edge and nuclear power, where there are plenty of examples of where one can find value.

Intel – Have not.

• Intel is continuing to lose talent as its AI chief has departed to help OpenAI run its infrastructure, which I think gives Intel a great opportunity to shut down business lines where it has no chance of seeing any success.
• Chief AI Officer, Sachin Katti, is moving to OpenAI, and, already long-suffering, CEO Lip-Bu Tan will add this role to his very full slate.
• Mr Tan arguably has the most difficult job in Silicon Valley, but I think that the new relationship with Nvidia gives him a good opportunity to cut areas where he no longer needs to be.
• The fact that there is even a deal with Nvidia basically means that Intel’s AI efforts and its integrated graphics chip activities cannot compete and should be closed down.
• Intel says that AI is a top strategic priority, but much like mobile, AI is passing Intel by, and there appears to be very little the company can do about it.
• Consequently, the best option for Intel is to focus on what it is good at, address the hordes that are invading its territory, and jettison the rest.
• The problem is that Intel seems to have very little focus when it comes to a recovery strategy, which will hurt the numbers once the Nvidia shine wears off.
• There is a bounce in demand for x86 in the data centre as an adjacency to the AI boom, and so I would like to see this period of better performance used to restructure the company and focus it for survival in the long term.
• Currently, I don’t really see any of these things, and so if I were holding Intel, I would sell immediately with no price limit.