The AI hype train is still running at full tilt — just don’t ask where the profits are

• Plenty of companies have suggested AI will generate big profits.
• The technology popularized by ChatGPT has attracted billions of dollars of investment as a result.
• However, it’s not clear yet if AI can generate enough returns to match the hype.

Like the discovery of gold in the hills of California in the 19th century, artificial intelligence has triggered its own rush in the 21st century — with no shortage of modern-day prospectors promising riches.

Mustafa Suleyman, the cofounder of businesses like DeepMind and Infection AI who’s now in charge of Microsoft’s AI division — cited research last year suggesting generative AI “is set to explode, with revenue expected to exceed $1 trillion” by 2032. The figure was $40 billion in 2022.

Expectations for the AI boom to generate serious money are absurdly high, then, which helps explain why the hype train for the technology is still running at full tilt. Here’s the thing — it’s not remotely clear where the returns on investment will come from.

Fresh questions about when — and even if — tech companies will be able to justify high capital expenditure and heady valuations over AI emerged this week as several gave a fresh look at the returns the technology is offering them.

Returns vs hype

On Tuesday, Google showed signs that its bet on AI had yet to make a meaningful impact as it reported second-quarter earnings.

Although revenue increased to $84.7 billion in the April-June quarter from $74.6 billion in the same period last year, execs such as Sundar Pichai and Philipp Schindler struggled to give investors specifics on how AI was contributing to its finances.

As my colleague Katherine Tangalakis-Lippert noted, an investor question about click-through rates and monetization from the “AI overviews” feature designed to summarize Google Search results (the same feature that recommended adding glue to pizza) got a pretty hazy response.

“Both to more countries, and also, you know, we have taken a conservative start, focused on quality, making sure the metrics are healthy and so on, but you will see us expand the use cases around it,” Pichai said.

That was probably hard for investors to hear given AI has pushed Google to spend more. Its capital expenditure in the quarter almost doubled year-on-year to $13 billion, as money has poured into chips and computing technology needed to power AI.

Stock slide

The Google CEO had another way of addressing concerns. The “risk of underinvesting is dramatically higher than overinvesting,” Pichai said, as Alphabet was “in the early stage of a very transformative era.”

Investors aren’t so sure. Alphabet shares fell 5% on Wednesday as a wider tech sell-off took hold of a market driven by overwhelming enthusiasm for AI this year. Tesla’s earnings this week, which showed a slump in net income, also weighed on the performance of tech stocks.

Similar questions around the gap between returns and hype have shown themselves this week in startup land, too.

Though investors typically give younger companies more time and leniency, it’s clear that many AI startups are notching valuations that represent multiples far beyond the returns they are making — with no guarantees of those returns scaling sufficiently in the future.

On Wednesday, The Information reported that OpenAI could be on track to lose as much as $5 billion this year, based on an analysis of its costs including the expense of running and training large language models behind the company’s AI.

That would be a stark situation for the ChatGPT maker, which has raised billions of dollars from Microsoft and has been valued at about $80 billion.

Toronto-based AI startup Cohere, founded by ex-Googlers in 2019, announced a fresh funding round of $500 million on Monday, putting its valuation at about $5.5 billion. That makes it one of the highest-valued AI startups in the world.

It also makes it a startup with a valuation that far exceeds the money it makes. Cohere, which offers large language model (LLM) applications to enterprises, hit an annualized revenue of $35 million in March, BI understands.

In one respect, this is a positive sign for Cohere, with annualized revenue trebling from the end of 2023 to the end of the first quarter of 2024. It’s also fair to assume revenue has grown further since then.

Still, if you were to assume growth has happened at the same rate since the start of the year, the $5.5 billion valuation it secured this week would still far exceed the money it is generating by a hefty multiple.

The company also laid off about 20 employees after the raise, Fortune reported.

In a statement addressing the layoffs, a Cohere representative told Business Insider: “With our most recent round of financing in place, we have a clear vision for the future of Cohere, which has required some internal realignment.

“We will continue to aggressively hire people as we work to offer companies the most accurate, secure and private multilingual AI solutions in the market.”

On Tuesday, another startup — Harvey — announced a $100 million funding round led by Alphabet’s venture capital arm, Google Ventures. Its valuation now sits at $1.5 billion.

High expectations

In a blog, Winston Weinberg and Gabriel Pereyra — who founded Harvey as a startup that offers generative AI services to the legal sector — said its annualized recurring revenue had also tripled since its last fundraising in December.

Back then, the figure was about $10 million, The Information previously reported. That implies the annualized recurring revenue is now about $30 million, with a valuation 30 times greater, highlighting how high expectations are for Harvey.

It remains to be seen whether these companies can generate returns that warrant the hype surrounding them. Last year, veteran venture capitalist Vinod Khosla suggested most startups were overvalued and that most investments in AI “will lose money.”

Those throwing cash at AI can only hope he’s wrong.