The current boom in artificial intelligence (AI) spending and the associated market surge is creating a bubble worse than that of the 2000s dotcom era.
This is according to GQG Partners, who in a recent research note argue the tech sector now faces lower growth, higher competition and greater capital intensity.
“In our view, the consequences of the current Al boom could be worse than those of the dotcom era, as its scale-relative to the economy and the market-is far greater,” the note said.
“We also think that some are making an ill-advised bet on the Al boom with their clients’ retirement security on the line.”
“We are not perma-bears on the technology sector, in fact, we were comparatively larger buyers of Nvidia in 2023, and the stock has been among the top performers since the firm’s June 2016 inception,” they said.
“Clients regularly pushed back on our historical overweight position in the technology sector just a few years ago.”
GQG has previously benefitted from holding overweight positions in the likes of Microsoft, Amazon and Nvidia. The latter was the firm’s flagship global equity strategy’s largest position as recently as June 2024.
Their views have since changed, and they now caution investors from allocating to a “cyclical sector where many of the largest players appear to be exhibiting growth deceleration, free cash flow margin deterioration, and increasing competition”.
Market leadership effectively indexed to AI theme
The overvaluation of equities is not confined to just tech stocks, according to GQG, which asserted that the market froth has extended into sectors that include industrials, financials, retail and healthcare.
“In industrials, companies linked to the power theme, such as GE Vernova and Constellation Energy, have seen unsustainable multiple expansions, in our view,” they said.
“This feels reminiscent of the 1990s euphoria over internet-driven electricity demand, which ultimately collapsed.”
When it comes to financials, GQG pointed to Robinhood’s meteoric rise to a $100bn valuation, which they argue is reminiscent of Charles Schwab’s ascent during the 1990s before years of stagnation.
In retail, they said high-quality companies like Walmart and Costco now trade at “massively inflated multiples”, mirroring Walmart’s peak valuation during the dotcom bubble.
Looking at the broader index level, GQG noted that 35% of the S&P 500’s weight now tilts toward high-multiple names, up from 25% in 2000.
“At GQG, we rarely turn cautious purely based on rich valuations. However, the trifecta of rich valuations, increasing macro risk, and-perhaps most importantly-deteriorating company fundamentals is very dangerous,” they said.
“We believe much of today’s market leadership has effectively become indexed to the Al theme, whether it be industrials, independent power producers, semiconductors, or capital market businesses.”
Deteriorating fundamentals
GQG also noted that the recent AI capital expenditure has been funded largely by advertising revenue, which is cyclical in nature.
“In the commodity cycle, investors similarly paid a heavy price for confusing inherently cyclical businesses as secular growers,” they said.
“To be clear, Nvidia was an unquestionably great company in 2018 and 2022, yet the stock still collapsed around 60% in both cases due to the underlying cyclicality.”
The firm drew parallels to telecom bubble where supply outstripped demand for the new technology, and companies “never earned a return on their investment”.
“In our view, this is far worse than the internet bubble, which at least generated meaningful revenue,” they said.
“We believe today’s intensive Al CapEx may structurally reduce returns on capital for the entire sector. Indeed, this is exactly what happened to the telecom sector during the 1990s fiber rollout.”
They warned that today’s surge of datacentre capital expenditure and cash-burning Al startups could end in a “bloodbath”.
They said: “To date, big tech has been able to mitigate the impact of massive CapEx spending on their earnings by repeatedly extending the depreciation periods for their investments.”
They believe that current depreciation numbers for the largest hyperscalers are “grossly understated” since they need to keep buying the latest Nvidia chips annually to stay competitive.
They said: “Once reality sets in, investors may find that earnings are massively inflated due to much higher depreciation.”
