Unlock Editor’s Digest for free
FT editor Roula Khalaf has chosen her favorite stories in this weekly newsletter.
How did Nvidia become the world’s largest publicly traded company without partnering or partnering with Apple? There are two answers. One is that the company is the dominant maker of chips that power the artificial intelligence era. The other is that we have benefited from the magic mathematics of addressable markets.
Nvidia founder Jensen Huang introduced a number of new products and predictions to attendees at CES, the massive technology showcase in Las Vegas, this week. Those in attendance applauded as he brandished nondescript black boxes and devices. But the real drama is saved for his vision of “physical AI,” the next frontier in artificial intelligence, where complex models will produce not just words but robot movements.
Even for a $3.4 trillion company, the numbers thrown around are big. Goldman Sachs suggests that humanoid robots alone could be a $38 billion market. But this pales in comparison to self-driving cars, or legless robots, which Mr. Huang considers a “multi-trillion” industry — and his customer, Tesla boss Elon.・Mr. Musk makes a similar claim. Manufacturing is a $50 trillion market ripe for robotization.
In theory, all of this would require data processing chips of the kind that Nvidia sells in large quantities. That’s plausible. Analysts at Citigroup recently estimated that there will be 1.3 billion AI-powered robots roaming around us by 2035, from vacuum cleaners to drones. By 2050, about 648 million people, roughly the population of Latin America, will resemble humans to some degree. These rely on trained models to navigate the world, device-level computing power, and fine-tuning that learns through experience.
This is all at the heart of an industry obsessed with “Total Addressable Markets” (TAM). This concept, long used in venture capital, is now commonplace among large publicly traded companies. The phrase appeared in 3,743 filings with the Securities and Exchange Commission between 2015 and 2020. In the five years since then, it has appeared in almost 20,000 cases. It’s easy to see why. Identify a new market, envision its future share, and anticipate future revenue numbers that analysts and investors will begin to price in.
Nvidia is the poster child for TAM expansion. Before the AI boom, that market was gaming chips. It has expanded dramatically to data centers. Robots, broadly defined, could make it even bigger again. Bank of America estimates that the company will have an 81% share of the $359 billion “accelerator” (AI) chip market by 2030. However, we also expect TAM to rise and Nvidia’s future value to be estimated accordingly.
Huang goes on to say that physical AI is having a ChatGPT moment. This is a claim worth considering for two reasons. The release of ChatGPT in 2022 was a moment when new possibilities, tools, and TAM suddenly came into view. But it’s also the moment when investors collectively lost their minds about the value of future profits that don’t yet exist and may not exist.
john.foley@ft.com
