NVIDIA And The Diversification Paradox

How diversification is the market’s end-game

Anyone who knows anything about investment will tell you that diversification is a sine qua non of portfolio construction. Richard Barclay spoke about diversification as the “only free lunch” in finance at our European Investment Symposium last week. There is, though, a stark divide between the theoretical benefit of diversification and the brutal facts of the last few years of heavily concentrated market leadership. As I posted the other day, NVIDIA
NVIDIA

SPDR Dow Jones Industrial Average ETF Trust
has been responsible for fully a quarter of the S&P’s c. 11.5% YTD performance.

I write this in the aftermath of another NVIDIA earnings beat. It feels like the company did just enough – and not much more – to push the prospect of what must surely be an eventual disappointment three months down the line. This morning saw the stock up 8% or so, moving through $1000, but there was a lack of the wow factor that has surrounded previous releases. That “just enough” felt like it wasn’t enough to give bulls a reason to push the stock dramatically higher (remember it jumped 25% post the Q2 results last year), but nor is there much for bears to build a case around. If nothing else, the 10-to-1 stock split should be good for retail flow…

My colleague Sumant Wahi has an impressive pedigree when it comes to speaking about this market. Prior to his investment career, he worked at the cutting edge of the semiconductor industry. I sat down with him soon after NVIDIA released its numbers to chew over the details. Sumant pointed me towards an interesting observation that goes beyond the numbers.

Jensen Huang is developing an increasingly rigorous – not to say desperate – focus on monetization. He spoke about the fact that NVIDIA is providing clients with complete systems solutions, building out AI factories. He spoke about the wave of Gen AI startups that are being unleashed – 15-20,000, he said – all of whom would have voracious demand for chips. It all felt like an attempt to pre-empt questions (which definitely came once the sell side were let loose on the earnings call) about the possibility of waning demand, over-ordering, chips sitting idle in data centres. Decoding the dynamic between analysts and management, the message was clear: how would this look different if it was a bubble? What is the AI killer app (something I’ve written about before)? How does the company continue its remorseless raising of the bar?

You might have seen this graphic before – it’s out of date already (the market cap is actually more than $2.5T post-results). But Sumant used it to make a different point. He sees significantly greater value in the companies on the right-hand side of the chart, and indeed in companies too small even to feature on the scale. The competition is coming for NVIDIA.

All this brings us back to the question of diversification.

Markets are self-regulating entities. While NVIDIA will do everything it can to ensure its position of leadership, it is inevitable that others will begin to chip away at its hegemony. It’s how markets work, it’s the central dynamic of competition, it’s one of the reasons that diversification is such a powerful tool. Nimble specialists will find ways of improving or undercutting or disrupting NVIDIA’s position. People have been holding up Cisco as a warning for NVIDIA to heed; IBM
IBM
is perhaps more relevant just now. IBM totally dominated the personal computing market in the early years. It grew cumbersome and bloated, and Microsoft
Microsoft
and Apple
Apple
profited.

It’s a powerful, practical example of the sometimes abstract theory of diversification, further evidence that you shouldn’t put all your eggs in one basket.