Will AI Companies Generate Enough Revenue to Continue the Build-Out? Here's What Nvidia CEO Jensen Huang Says.

Artificial intelligence (AI) stocks have taken a tremendous hit in October, as a toxic combination of higher interest rates and skepticism over the returns on AI spending has led to a steep sell-off.

Higher interest rates make it more costly to build AI data centers, all else being equal. Meanwhile, while leading AI labs Anthropic and OpenAI are still private, recent media leaks about their third-quarter revenue figures have suggested a potential slowdown.

If these two labs' revenues don't grow as much as some people anticipate, investors may cast doubt on the entire AI build-out, since Anthropic and OpenAI wouldn't be able to fund their commitments.

Are the fears real, or overblown? This week, Nvidia (NASDAQ: NVDA) CEO Jensen Huang weighed in on this very issue on a popular tech podcast.

The controversy

On Oct. 8, The Financial Times reported that in an investor presentation, OpenAI revealed it had reached $50 billion in annualized recurring revenue (ARR). That is well short of the $70 billion revenue figure leaked in September. While the difference was later addressed as a discrepancy in accounting methods, the headline still caused anxiety.

Then last week, Anthropic held its own investor presentations, revealing that it had made roughly $16.8 billion in revenue in Q3. Considering Anthropic generated only $4.6 billion in revenue in all of 2025, that's a remarkable achievement. Yet investors may still have been disappointed. In August, it was leaked that Anthropic was already at a $65 billion ARR at the end of July. So, the $67.2 billion Q3 annualized revenue figure would indicate a pretty sharp slowdown from July through September.

The back-to-back "leaks," along with rising inflation, were enough to send AI-related stocks, especially in the semiconductor sector, into a tailspin. After all, if Anthropic and OpenAI are slowing down, or not making as much as they thought, who is going to buy all of these chips and build all these expensive data centers?

Image source: Getty Images.

Jensen: AI is bigger than Anthropic and OpenAI

Last Thursday, Nvidia (NASDAQ: NVDA) CEO Jensen Huang appeared on a podcast with venture capitalist Brad Gerstner. In that interview, Gerstner put the big question to Huang: Would Anthropic and OpenAI generate enough revenue to continue growing the AI build-out?

Huang remained confident for the following reasons.

1. There's more than just two frontier labs

The first point Huang made was that Anthropic and OpenAI aren't the only two frontier labs. During the interview, Huang pointed not only to OpenAI and Anthropic but also Grok, the chatbot and agent from xAI, which is owned by Space Exploration Technologies (NASDAQ: SPCX), Meta Platforms' (NASDAQ: META) new Muse agent, and Alphabet's (NASDAQ: GOOG) (NASDAQ: GOOGL) Gemini, as all "doing great" and growing.

Of note, Grok Bot, xAI's new personal and business agent, launched in August. Meta Platform's much-lauded Muse personal agent debuted in September. An impressive open-weight model from China's Moonshot AI, called Kimi, debuted in July. Not to be outdone, OpenAI released its latest Codex model in July as well, meant to compete more strongly with Anthropic's leading coding capabilities.

That's a deluge of competitive agents substantively entering the fray during Q3. As such, it's perhaps inevitable that Anthropic's slowdown was due to competition and market-share loss rather than an overall slowdown in AI usage or spending.

While we don't have exact figures, Axios reported that OpenAI's revenue run rate grew by roughly 70% in Q3, well above Anthropic's 46% quarter-over-quarter growth. That seems to price in the share-loss story, not a general AI slowdown story. We also don't know how much SpaceX, Alphabet, or Meta's AI-related revenue grew, as they haven't reported earnings yet. None of these companies breaks down AI agent revenue specifically, either. So, it may be difficult to even understand how fast AI agentic revenue is growing inside these companies when they do report.

2. Frontier Labs are only 25% to 33% of Nvidia revenue right now

Another point Huang made was that frontier labs account for only a quarter to one-third of Nvidia's chip demand. Another third, according to Huang, is general computing, while the final third involves new and emerging AI-related companies. These companies, for instance, may use an open-weights model and then add proprietary data and industry knowledge to create new AI products.

On general computing growth, one point Huang has stressed over the past couple of years is that Moore's Law is "over." Moore's Law says that the amount of computing power roughly doubles every year at the same price, as chips pack more and more transistors onto a piece of silicon. However, Huang claims that semiconductor manufacturing is running up against the laws of physics and that, as a result, compute efficiency gains are slowing. Therefore, if someone wants to double their compute next year, the cost will increase rather than remain the same.

Given that there are more than just two labs, and that frontier labs account for only 25% to 35% of demand, Huang thinks people are putting way too much emphasis on OpenAI and Anthropic. Meanwhile, those labs still appear to be doing quite well even as they face more competition.

There should still be a shortage

Almost all commentary from leading AI chip companies says there is still a huge shortage of compute relative to demand. So even if leading frontier labs are under competitive pressure, semiconductors seem likely to find a buyer somewhere.

It appears from early Q3 earnings reports, especially from Taiwan Semiconductor Manufacturing, that chip demand remains strong. However, AI investors would do well to listen to other companies downstream from the chip giants to see whether they are generating AI-related revenue growth and efficiency gains.

Ultimately, AI will have to generate revenue and/or cost efficiencies for the end customer to justify this spending. So, commentary from AI users -- not necessarily the chipmakers themselves -- may be more important to chip stock performance than the chip companies' earnings during this upcoming earnings season.

Billy Duberstein and/or his clients have positions in Alphabet, Meta Platforms, and Taiwan Semiconductor Manufacturing. The Motley Fool has positions in and recommends Alphabet, Meta Platforms, Nvidia, and Taiwan Semiconductor Manufacturing. The Motley Fool has a disclosure policy.