This analysis was first published in SvD Näringsliv, in Swedish, on March 19th, 2024. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.
A supremely confident Jensen Huang presented Nvidia’s plans to the world. The message was unmistakable: without Nvidia, there is no AI development. But on the horizon, competition is stirring from a troubling direction.
Jensen Huang, CEO of Nvidia, walks onstage to bombastic orchestral music that could have come from a Hollywood blockbuster. “You have come to a developer conference,” he clarifies.
The music, it turns out, was written by AI — but performed by a live symphony orchestra. Huang himself, in his now-familiar leather jacket, seemed to harbour a certain rock star energy. But only so much.
GTC — Nvidia’s own conference — has been held since 2009. Back then it was a small, internal affair. On Monday evening, thousands of participants gathered in San Jose — with many times more following the livestream — to see what the company at the centre of the AI revolution has planned for the future.
Huang is acutely aware of the position he now occupies — and how far they have come since he co-founded the company in 1993. He pauses on 2017, when Nvidia produced a new kind of computer called the DGX-1. “I delivered the first unit by hand to a startup in San Francisco,” he recalls. “They were called OpenAI.”
The message is clear: without Nvidia, no OpenAI, and no AI revolution. Now it is about scaling up further. From the stage, Huang proudly unveils the company’s next major platform — Blackwell — which will make generative AI more powerful than ever. “God, I love Nvidia,” he says spontaneously. It shows. And he has 23 trillion kronor worth of reasons to feel that way — that is Nvidia’s current market capitalisation.
Is there anything that could stop them?
The new-found success has allowed Nvidia to broaden its operations considerably. Not only do they supply the hardware on which AI development runs — they have also expanded into financing the companies that do the developing. Last year, Nvidia invested in over 30 AI-related companies, and a month ago bought a larger position in the listed Arm Holdings. Nvidia had previously attempted to acquire all of Arm, but that deal was blocked in 2022 by competition regulators. At the end of January, the value of these investments stood at over 16 billion kronor.
The investment activity stands out because it creates a kind of financial circular flow. Nvidia invests money in promising AI companies, which then spend a substantial portion of it on buying products and services from Nvidia itself. Access to Nvidia’s chips has been a challenge for the entire industry. Spending newly raised capital on them is entirely rational. What becomes slightly unusual, however, is that the money comes from the same supplier that companies then must buy from. Investments in other companies effectively become direct revenues for Nvidia.
The insights gained into how products are used also help shape the next generation of development — and make inventory management easier, since demand becomes more predictable. Right now it is exploding, but as recently as mid-2022 Nvidia had to write down part of its inventory after misjudging market demand.
Investing in customers also keeps competitors at bay. By financing companies that use Nvidia’s products, Nvidia creates both demand and a head start for those chosen firms.
Competitors do exist, even if the market currently seems to value Nvidia as unbeatable. The share price is up just over 80 percent since the start of 2024, and over 240 percent compared to a year ago. Nvidia leads specifically in what is called the GPU — graphics processing unit — which, broadly speaking, is the type of processor used to train AI models. But established players such as AMD, Intel, and Qualcomm see the explosive growth and are unlikely to let it pass them by forever.
Another competitor that receives less attention is Apple. They do not sell chips to others, but they manufacture their own — which means they can build chips specifically designed to support Apple’s own strategy. Apple is therefore not a competitor to Nvidia in selling chips to external customers, but rather in terms of the AI functionality that is built with them.
Nvidia’s software development platform, CUDA, is today a standard for AI development. But building platforms of this kind is something Apple has been world-class at for years. In the longer run, this could give rise to a new kind of competitive dynamic between giants in the multi-trillion-kronor bracket.
In the short term, Nvidia still stands largely alone at the top. They have outmanoeuvred the existing players in the industry. Who talks about Intel today? And even if players like Apple eventually mount a challenge, it is many years away from becoming reality.
That more competition will come is, however, entirely certain. Large and small — from startups to tech giants — everyone sees the explosive AI development and wants to stake a position. How long can Huang and Nvidia hold on?
