Artificial Intelligence – Confidence Debate

Jensen lends CoreWeave his credibility.

• The crazy thing about financial markets is that it doesn’t matter whether Jensen’s comments about CoreWeave are correct or not, as just by making them he may have restored enough confidence to allow them to come true.
• The latest wobble in confidence in the AI trade has come at the hands of both Anthropic and the US government.
  • First, Anthropic: which held a pre-IPO investor meeting at which it appears that Q3 26 revenues have been lighter than expected.
  • Going into the anticipated IPO, expectations have been that Anthropic would increase revenues by 10x in 2026 such that the annual run rate by the end of the year would be over $100bn.
  • However, it looks like revenue in Q3 26 was around $16.8bn, which, taking into account the growth so far in 2026, would lead one to estimate that the ARR for Q3 26 is closer to $73bn.
  • This implies that growth is slowing more quickly than expected, meaning that everyone is starting to wonder about demand.
  • This is something that RFM Research (see here) highlighted in its recent research, as the price of compute is now so high (due to scarcity) that prices for AI services have to rise to prevent the model makers from going bankrupt.
  • This, in turn, leads the market to think that growth is slowing down, meaning that those that have borrowed money will find it less easy to pay them back, and so yields on bonds rise tom compensate for the added risk.
  • Second, US government: whose unwillingness to cut the fiscal deficit is causing the yield on government bonds to rise.
  • When governments spend more than they earn, it tends to be inflationary, which means that savers become more concerned with the declining purchasing power of fiat currency from inflation and demand a higher return to compensate.
  • Government bonds are the foundation from which all other costs of capital are based, meaning that the cost to issue bonds and the return that investors demand from equity all rise.
• This has hit the neo-clouds the hardest because while they all have massive order books, their valuations remain predicted on data centres that they have not yet built or paid for.
• Hence, they remain hostage to the confidence that the market has in their ability to pay back the loans, resulting in the sharp decline we have seen in their share prices and CoreWeave in particular.
• The problem with confidence is that it can become a vicious cycle, with falling confidence increasing financing costs, which dents confidence further and so on.
• The end result can be a collapse much like Lehman Brothers in 2008.
• Into this maelstrom enters Jensen Huang, whose company owns 10% of CoreWeave and has agreed to purchase any unsold capacity, with the statement “CoreWeave’s strength is improving every day,” and suddenly some confidence is restored.
• This is what was responsible for the extreme volatility yesterday when most of the AI trade swung from sharply down to up somewhat.
• Hence, Jensen has given CoreWeave (and everyone else) some breathing space, but whether it is enough remains to be seen, and fairly shortly, the public Neo Clouds are going to start reporting Q3 26 results, and they need to be good if they wish to prevent the meltdown from resuming.
• The economic environment for the sale of AI compute capacity is currently superb, with returns of 50% to 85% per year on offer, but you can only earn this if you have capacity already built to sell.
• However, customer concentration is still very high, with OpenAI and Anthropic being responsible for most of the deals being struck, meaning that if something goes wrong with one of them, it could trigger a widespread collapse.
• Demand needs to diversify, and Nvidia has shown some signs of this, but everyone else needs to as well.  
• This will be a key feature of the Q3 26 reporting season that is almost upon us.
• For the moment, anyone with capacity (xAI) is currently printing cash like there is no tomorrow, which is why I hope that Meta will also start announcing deals to sell compute capacity to the market through its newly formed enterprise business unit.
• This, combined with its relatively low valuation, is why I have a position in Meta Platforms, where I am looking for around $770 before deciding what to do next.