Where do we stand on the AI cycle?

To keep the party for AI suppliers going, AI customers need to see results. Recent earnings releases in the software and hardware distribution space suggest that momentum may point south from here.

Disclaimer: The information contained in this article is not and should not be construed as investment advice. This is my investing journey and I simply share what I do and why I do that for educational and entertainment purposes.

---

TLDR Summary

• ‘AI’ is providing a suitable narrative to drive this bull market. So far, the act of digging for digital gold alone has been sufficient fuel to drive earnings of ‘AI suppliers’ and to build optimism on those going forward. But eventually, those funding the frenzy via equipment orders and consulting engagements will have to see results to justify the spending.

• Deep learning shows tremendous potential to raise productivity, especially in software development and business process automation. But there are limits with respect to the problems it can be applied to and how quickly it can be implemented to solve them. Once those limits are reached, the RoI of corporate AI projects declines which may deter executives from maintaining their investment budgets. First compute shortage, then compute glut.

• The initial impulse of this cycle happened at the supplier level. ChatGPT served as the prototype for what’s possible and then everyone started chasing. The turning point will happen at the user level when a crucial number of AI customers decide to dial down their investments.

• NVIDIA won’t likely be among the first to get that memo. Those will be the hardware and software system integrators that distribute all the new computing power into the market place. I am talking about a huge and heterogenous group of companies, many members of which are subject to company-specific business drivers. But some respectable players in this group have recently sold off following their earnings releases. Some of them did so due to weaker than expected actual performance. Others performed strongly, but issued cautious guidance. This may be noise within a much larger industry cycle that goes far beyond my imagination. But it could be that these are early signs that the AI cycle is peaking with downward momentum ahead.

• Deep learning has not altered the laws of economic physics in information technology spending. IT spending is obviously growing structurally with a steep slope, but it’s also notoriously cyclical. Together, these two characteristics make it extraordinarily difficult to identify where the current midpoint of the cycle is. Markets typically overshoot in both directions. With every day that passes, the odds are rising that we are passed the midpoint of this cycle.