The talk of the town, stock market-wise, has no doubt been artificial intelligence chip leader Nvidia (NVDA 4.43%). But after soaring 532% over the past three years and with its market cap exceeding $3 trillion, is there any more room for this AI darling to run higher?
One technology investor thinks so. In fact, he thinks there's a lot of room to run.
NASDAQ: NVDA
Key Data Points
Nvidia to $50 trillion?
Last Sunday, investor James Anderson of Lingotto Investment Management in London told the Financial Times he thinks there's a chance Nvidia reaches a $50 trillion market cap over the next decade. That would mark a 16 times increase from today's levels, which is already the result of blockbuster performance over the past two years.
That seems pretty implausible. After all, a $50 trillion valuation would be nearly twice the current gross domestic product (GDP) of the U.S., which reached $27.4 trillion in 2023, and it's almost three times the GDP of China at $17.8 trillion.
Of course, U.S. GDP, which is the sum total of all personal consumption, business investment, government spending, and net exports, may not be a good comparison to market cap. GDP accounts for revenue, while a market cap is a multiple of a company's revenue or earnings. Yet even in a scenario in which U.S. GDP rises 50% over the next decade, the $50 trillion market cap scenario suggests Nvidia's revenue would make up a rather large portion of U.S. GDP, and also for it to trade at a high price-to-sales ratio, suggesting it maintains sky-high profitability, too.
But Anderson was a former partner at Baillie Gifford, which is one of the biggest growth-oriented mutual funds in the world, and was an early investor in Amazon and Tesla. So his opinion shouldn't necessarily be discounted. Let's dive into his assumptions.
How Anderson gets there
Anderson, obviously, sees the current artificial intelligence boom not as a short-term blip but rather as transformative for society. Anderson estimates today's demand growth rate for AI data centers is about 60%, and he merely extrapolates that out 10 years. Furthermore, he also sees no reason Nvidia's margins should contract, even though Nvidia's adjusted net margins have surged to a sky-high 58.5% last quarter, which is really, really high for a chipmaker.
If that scenario plays out, Anderson sees Nvidia earnings growing to $1,350 per share and free cash flow of $1,000 per share in 10 years. At a future 20 times free cash flow multiple, that would put the stock at $20,000 per share, or a $49 trillion market cap at today's share count.
While Anderson admits this scenario is not his base case, in the interview, he gives this $49 trillion scenario a 10% to 15% probability.
Image source: Getty Images.
Are Anderson's projections reasonable?
Anderson's projections could seem reasonable to some. After all, other industry participants have projected huge growth rates for AI chips in the immediate future. Nvidia's manufacturing partner, Taiwan Semiconductor Manufacturing (TSM +0.53%), has previously guided that AI chips will grow at a 50% average growth rate for the next five years. And on its last earnings conference call with analysts, TSMC management also noted it would grow its CoWoS advanced packaging capacity for AI applications by 60% for the next four or five years. Currently, that capacity is in shortage right now and needs to catch up to the market.
But to grow at that rate for another five years after that, on a base of revenue that will be considerably larger, is a much, much bigger ask. Usually, the law of large numbers states that as a company grows, it gets harder to maintain the same rate of growth merely because a company would get so big in relation to its industry or the economy at large.
These numbers seem unlikely
And this is why Anderson's numbers seem implausible to this investor. Because while a 60% growth rate in AI chips, from very minimal numbers last year, would get one to about $400 billion in AI chip revenue in 2027, another 60% annual growth for another five years after that would bring overall AI chip revenue to $4.1 trillion by 2033 -- more than 10 times the already-large AI chip market forecast by 2027.
For reference, many people who are optimistic on semiconductors think the entire semiconductor industry will reach only about $1 trillion by 2030.
Moreover, there is more competition coming soon for Nvidia, not only from its merchant chipmaker competitors ramping up their own accelerators this year for the first time, but also from cloud giants increasingly investing in their own in-house AI chips.
These chips will probably have much lower prices. With customers paying currently through the nose for Nvidia chips and probably side-eyeing Nvidia's sky-high margins, they will no doubt be a tremendous push to look for cheaper alternatives running on open-source software. Given Nvidia's super-high margin, there's lots of room for other tech giants to undercut the AI giant on pricing and still do well.
No doubt, these competitors are well behind Nvidia at the moment, but with the whole of the technology industry looking for more cost-efficient ways to run AI, I find it unlikely Nvidia will be able to maintain these gigantic margins forever.
Given its current first-mover status in AI, Nvidia may continue to be a solid stock. But $50 trillion? Never say never, but that appears fantastical.
