Nvidia (NASDAQ: NVDA) stock has been on a tear over the past three years, fuelled by the AI boom. In late October, it became the first publicly traded company to surpass a $5trn market capitalisation. Since hitting that milestone, the stock has dropped sharply. But even today, it trades at an eye-watering 33 times sales, not earnings. So how likely is it to repeat the same performance in 2026?
Hyperscalers driving AI demand
There’s no doubt that the boom in AI infrastructure is being driven by hyperscalers. Companies such as Google, Microsoft, and Meta are spending hundreds of billions to build data centres and AI labs, with high-end GPU chips forming the hardware backbone.
On paper, it looks incredible – the GPU maker effectively dominates the market for AI chips, and demand appears endless.
From cash flow to debt
But here’s the catch: the current AI build-out is very different from the tech expansion of the 2010s. Back then, hyperscalers grew rapidly using their enormous cash flow reserves, leveraging infrastructure that was already in place.
Netflix, for example, could piggyback on fibre laid down years earlier during the dotcom bubble, while Microsoft and Google required relatively little new capital. Growth came almost for free.
Today, the story is reversed. AI expansion is being funded not through cash flow but through debt. Oracle’s $300bn data centre plan and SoftBank’s $20bn injection into OpenAI are stark examples.
Investors initially cheered these mammoth deals. But the market soon questioned how they would be financed. After surging on the announcement, Oracle’s share price has since plunged 40%.
The circular AI boom
I’ve been warning for some time that the AI ecosystem is becoming increasingly circular. The parallels with the final days of the dotcom bubble are hard to ignore.
Here’s how it works. Investors pour vast sums of cash into AI startups. Those startups then spend heavily on high-end GPUs from companies like Nvidia. The GPU makers’ shares surge, which encourages even more investment into AI startups. And the cycle repeats, faster and bigger than before.
Take OpenAI, for example. It burns $12bn-$15bn per quarter just to keep the lights on. Some upcoming IPOs could value OpenAI at $700bn-$900bn, before it has ever made a cent of profit. The money is chasing money: hype, expectations, and valuations are driving the market, not actual earnings.
It’s exactly the kind of pattern we saw in the late 1990s. Back then, companies built infrastructure for websites, fibre networks, and servers, assuming revenues would follow. When it didn’t, valuations collapsed and investors lost fortunes.
Bottom line
Nvidia remains the poster child of the AI boom, delivering extraordinary gains over the past few years. But where is the margin of safety in the stock today?
In the dotcom bubble post-mortem, Sun Microsystems’ CEO explained it perfectly: trading at 13 times sales, it would have taken 13 years for an investor to get their money back – assuming no costs, taxes, or wages! Nvidia now trades at 33 times sales.
That’s why I’m steering clear of Nvidia and the other Magnificent 7 stocks for now. For UK investors, I think the FTSE 100 offers far more attractive opportunities for steady growth and long-term returns. Here’s one I have my eye on.
