AI landscape today compared to the internet in the late 1990...

AI landscape today compared to the internet in the late 1990s:

1. During the internet infrastructure buildout, $Cisco (CSCO.US)$ revs increased ~15.5x over 6 years from the end of 1994, when Netscape Navigator (the first mass internet browser) was introduced, to its peak and Cisco’s stock rose ~4000%.

2. $NVIDIA (NVDA.US)$ , since the end of 2022 when OpenAI’s ChatGPT was introduced, has seen revenues go up “only” 5x over the past 7 quarters through today’s results with the stock up “only” 760%.

So far the six Magnificent 7 names that already reported have declined 4% on average the day after reporting on generally disappointing guidance. Only $Apple (AAPL.US)$ and $Meta Platforms (META.US)$ both beat revs and EPS for Q2 and had forward revenue & EPS estimates go up as well.

NVidia’s stock is also likely to decline tomorrow given it had the smallest rev/EPS beat of 4%/5% over the past 5 quarters with an even smaller increase relative to consensus for the next quarter.

This increases the odds in my mind of a pending AI capex digestion phase at some point over the next six months. While industry hyperscaler capex is expected to be up ~50% in CY24, NVDA revs were up 262% y/y in CQ1:24 and up 122% y/y for CQ2. Customers have probably been ordering more GPUs than they need over the past 7 quarters given supply has severely lagged demand. As the CEOs of Google, Microsoft and Meta have all said recently, they view the greater risk to be underspending on AI versus overspending. What this means to me, is that when they decide to digest their historical spend now that supply is catching up with demand, much like following Covid, this may be a sharp correction. Deceleration to ~10% capex growth or worse next year would not surprise me.

Following CQ2 results, forward revenue estimates went down at the three AI related hyperscalers: $Microsoft (MSFT.US)$ , $Amazon (AMZN.US)$ and $Alphabet-A (GOOGL.US)$ . These are three of the biggest customers for NVidia (in addition to Meta) and in 2022 digestion at these same customers occurred following the build out during Covid. As a result, Nvidia revenues declined 28% in six months while revenues went from up 53% year-over-year in CQ4:2021 to down 21% by CQ4:2022. The stock declined 66% from peak to trough during this slowdown.

In summary, I continue to expect a digestion phase to start over the next six months (and a commensurate fall in the stock price) but then for the AI buildout to continue for the next several years. As a reminder, Cisco’ stock had intra-year declines of 26% in late ‘95, 38% in early ‘97; and 37% in late ‘98 as expectations of growth ebbed and flowed while the stock ultimately climbed ~4000% from the end of 1994 to its peak. I anticipate similar volatility for Nvidia over the next several years while both the revenues and stock double over that time as the AI buildout continues.