Short story:

Nvidia had another objectively great quarter, but investors were not impressed. Shares were down 4.5% as of 1pm. However, other AI-linked tech shares are slightly up today, implying that investors have not soured on AI as a concept. Rather, they were just hoping for more fireworks from NVDA. Nvidia shares have been in a period of ‘consolidation’ for a couple of months. My speculation is that it could continue to outpace the S&P over the next 1-2 years. But barring a big change in outlook I think we’re likely to see returns more in the range of 10-30% than the 150% we’ve seen over the last 12 months. Also, the risks are larger than appreciated by many.

 

Longer story:

Nvidia reported 30b of revenue and 0.68 EPS, beating expectations handily on both the top and bottom line. YoY revenue is up 122%. Pretty fantastic. The fact that Nvidia shares are down today reminds us of two realities: 1) it’s not how well a company performs, but how well they perform relative to expectations and 2) hyper-growth becomes progressively difficult to sustain as time passes for any firm, in any industry. Nvidia is already a 3 trillion dollar company and #2 in the S&P 500 by market cap. It’s worth the next 6 largest chip makers combined.

 

As to the read-across for the broader market. Nvidia is definitely the standard bearer for the AI trade, and as a large component of the index it’s just going to continue to be very influential. However, I think people are now differentiating between the sellers (aka enablers) of AI and the customers (aka implementors) of AI. That implies that tech as a whole may be less dependent on AI sentiment now than it has been recently. Put another way, people getting rich selling picks and shovels doesn’t mean all the gold miners are hitting it big. And a slow down in the picks and shovels business doesn’t necessarily mean no one is finding gold.

 

For now, the big issues for Nvidia are (still) just how long their client’s appetite will last and how fast they can ship the chips. I think the bull and bear cases are pretty clear. The good stuff is all likely, but the forward multiple of 46x means most of it is (still) priced in. Low probability/high-impact risks exist.

 

Bull case

  • Nvidia is entering into a new product cycle with Blackwell chips which should incentivize continued spending from customers, most of whom are engaged in a capacity arms-race.
  • Management has stated that forward demand is still outstripping supply.
  • Nvidia still has a large moat. Competitors are working hard, but no one else is close (yet).
  • Biggest customers have used Q2 to reiterate that they’re all in on AI, even though it’s starting to dent their results.
  • Nvidia’s announced buybacks show they’re focused on achieving higher share prices.

Bear case

  • Nearly half the spending on advanced chips comes from a handful of large, cash-rich US tech firms. This spending is defensive in nature. CEOs admit that they’re over-building and over-spending. The objective isn’t to make money. No one has any idea what the ROI on buying these chips is going to be. Instead, their focus is simply to “not be left behind” at some point in the future. They believe they may make a good return on it in 10-15 years from now. That’s fine, but only insofar as their core businesses are growing nicely, because that’s what’s funding this CAPEX and that’s what’s keeping their shareholders sweet. If core businesses for big tech (advertising, software, hardware) slow, then you’ll see an about-face on chip spending worthy of Mark Zuckerberg ditching the metaverse.
  • Extreme reliance on TSMC in Taiwan; about 43% of their cost of goods sold.
  • Nothing this good lasts forever. They’re well behind, but competition is coming, both from start-ups creating new chips from scratch and Nvidia’s own customers who don’t want to be held hostage by a single supplier. Nvidia won’t monopolize the high-end forever.
  • Just math: The $ value of sales has continued to grow exponentially, but as the denominator of the YoY growth calc rises, the resulting % growth figures tend to decline.  We can see that happening already.

 

 

Separately, Super Micro (SMCI), a company sometimes mentioned in the same breath as Nvidia has come under tremendous pressure recently. Shares are down 34% over the last 30 days. But this appears to be mostly a stock-specific issue, rather than something that implicates the AI-trade. Super Micro has severe governance issues and persistent accounting irregularities. So much so they were temporarily delisted from NASDAQ a few years back for failing to file financials. Shares are down now because a well known short-seller called Hindenburg has published a report accusing them of being up to their old tricks again. Specifically: inflating their sales volumes through accounting gimmicks. Super Micro responded by essentially confirming the allegations when they delayed their annual report. But again, no major import for the AI trade in general.