Why one analyst just took the brave step of downgrading Amazon and Microsoft

• Rothschild & Co Redburn downgraded Amazon and Microsoft stock this week.

• It said it failed to see a clear bull case due to key issues with the business model for AI firms.

• Confidence is already shaky in the AI trade, with top tech stocks selling off in recent weeks.

One analyst has taken a dimmer view of two AI kingmakers this week.

Alex Haissl, an analyst at Rothschild & Co Redburn, is a lonely dissenting voice among research analysts when it comes to shares of Amazon and Microsoft. In a note to clients this week, Haissl downgraded the firm’s rating on the tech giants despite the ongoing hype for generative AI and the billions both firms have pledged to spend on the new tech in the coming years.

Investors look like they’re valuing Amazon and Microsoft’s capex plans as “as if cloud-1.0 economics still applied,” Haissl wrote, referring to the low-cost structure of cloud-based services that allowed Big Tech firms to boom in the 2010s.

But there are a few problems that suggest the AI boom likely won’t play out in the same way, and it is is probably far more costly than investors realize, he said.

“We downgrade Amazon and Microsoft with a heavy heart,” Haissl wrote, noting that the firm had held a bullish view on the two stocks for the last several years.

“But after turning over every stone, we no longer see a credible path back to cloud 1.0 economics. The market, however, still prices in that outcome, implying returns we believe are no longer achievable,” he added.

There are key differences between the business model for “cloud 1.0” and generative AI, Haissl said.

For one, AI is far more costly. According to Rothchild’s analysis, a GPU costs firms around $40 billion in capex per gigawatt of power.

It’s also believed to generate less revenue. Each GPU generates around $10 billion of revenue per gigawatt, per the firm’s calculations.

Amazon, Microsoft, Meta, Alphabet, and Apple are on track to spend around $349 billion in capex this year, much of which is directed toward AI infrastructure, according to company statements.

“We are not concerned about operating margins; those can be managed. The real constraint is the capex required relative to the growth delivered,” Haissl said, later adding that he believed the market was overestimating the returns related to AI investment.

Second, AI chips are also known to have a short lifespan, another factor that can make projects costly. If a GPU were to be replaced every three years, projects would turn “value destructive,” according to Rothchild’s estimates, implying losses for the owners of the chips.

Third, hyperscalers have limited pricing power, Haissl said, pointing to the expenses associated with AI hardware, which is largely provided by tech startups.

“Passing higher costs on to startups would simply deepen their losses unless those costs could be pushed through to end users; if not, hyperscaler economics deteriorate further,” he said.

Amazon and Microsoft already look like they’re “re-rated meaningfully” in price, Haissl said, pointing to the recent declines in both stocks. Amazon shares are down around 13% from their peak in early November, while Microsoft shares are down 10% from their peak in late October.

“While there may still be some upside to growth, we believe it is limited relative to buy-side expectations and, more importantly, less relevant given the low value of that growth,” Haissl said.

That’s not to say the firm is turning completely negative on Amazon and Microsoft. It’s hard to hash out a bear case for the near-term for these stocks, Haissl wrote, adding that the firm believed both would continue to post strong growth figures over the next year.

“In our view, it is not a bear case — but it is no longer a bull case either,” he said, adding that the firm would need to see high growth and a “meaningful reduction in capex” in order to turn more optimistic.

It’s a fragile time for the AI trade, with tech stocks sinking in recent weeks as investors take stock of high valuations and begin to doubt the billions being sown into AI-related investments. The tech-heavy Nasdaq 100 is down 6% from its high in late October.

The Roundhill Magnificent Seven ETF, which tracks the seven tech titans at the heart of the AI trade, is down 7% from its peak last month.

Read the original article on Business Insider