The Nvidia CEO may have just revealed the next winners of the AI cycle

Key Takeaways

• During Computex 2026, Jensen Huang spent most of his keynote arguing that the era of Agentic AI has arrived and that demand for AI infrastructure is only getting started.
• Nvidia unveiled several major milestones in its roadmap, including Vera Rubin, Vera CPU and RTX Spark, which are expected to become foundational building blocks for the next generation of AI applications.
• Yet the biggest market reaction did not come from Nvidia's main keynote.
• After describing Marvell as the “next trillion-dollar company,” MRVL surged more than 25% within hours on Hyperliquid.

Computex 2026 Confirms Nvidia’s AI Thesis Remains Intact

If you watched Jensen Huang’s full Computex keynote, you probably noticed that he spent nearly an hour and a half repeating the same concepts over and over again: Agentic AI, AI Factories, token generation, compute infrastructure, and more.

Nvidia’s core message is that LLMs are not the final product of AI. Instead, they are the engines that will power future agents capable of reasoning, using tools, querying databases, writing code, and making decisions. According to Jensen Huang, we are moving from a world where AI mainly answers questions to one where autonomous agents can execute complete tasks.

In other words, compute demand per user could become significantly higher than what we see today. To support this transition, Nvidia unveiled several important announcements.

• Vera Rubin is now in production, validating the continuation of massive hyperscaler investments in AI infrastructure. Jensen Huang repeatedly emphasized that demand is no longer limited to model training but increasingly driven by inference and large-scale agent deployment. According to Nvidia, future data centers will need to generate billions of tokens continuously, requiring an entirely new generation of infrastructure.
• Vera CPU was officially introduced, a processor specifically designed for AI and agent workloads. Jensen Huang explained that traditional CPUs were built for human users, whereas Vera was designed to orchestrate systems composed of thousands (or even millions) of AI agents operating simultaneously. Through this launch, Nvidia is also expanding its influence beyond GPUs into a market historically dominated by Intel and AMD.
• RTX Spark and the new generation of AI PCs developed alongside Microsoft and MediaTek were also showcased during the conference. Nvidia describes these machines as the first computers truly designed to run AI agents locally without relying entirely on the cloud. According to Jensen Huang, this could represent the biggest change to the PC industry in more than 40 years, with the ambition of turning every computer into a genuine AI assistant capable of executing complex tasks directly for its user.

The key takeaway from this conference is that Nvidia continues to see very strong growth in demand for AI infrastructure. Whether through Vera Rubin, Vera CPU or RTX Spark, every announcement was built around the same assumption: companies will continue investing heavily in compute power in order to deploy more AI models and agents.

For investors, this suggests that the AI cycle remains largely driven by infrastructure spending. Nvidia’s GPUs are obviously among the primary beneficiaries, but the entire value chain involved in building, connecting and operating these systems could continue benefiting from this trend over the coming months and years.

When Jensen Huang Speaks, the Market Listens

However, the most impactful announcement for investors did not come from Nvidia’s main keynote. During an appearance alongside Marvell CEO Matt Murphy, Jensen Huang pointed directly at him and declared: “the next trillion-dollar company, ladies and gentlemen.”

A simple sentence, but one coming from the individual who likely has the best visibility in the world across the entire AI value chain. The market reacted immediately: Marvell (MRVL) jumped from roughly $219 at the previous close to more than $272 on Hyperliquid outside regular market hours, representing a gain of more than 24% within just a few hours.

This reaction was not driven solely by Jensen Huang’s reputation. Marvell has become one of the most strategic companies in the AI ecosystem. In its latest quarterly results, the company reported a record $1.9 billion in revenue, with nearly three quarters of total sales coming from its Data Center segment. That business grew 76% year-over-year, largely driven by hyperscaler demand for AI infrastructure.

Unlike Nvidia, which directly sells compute power, Marvell operates further down the infrastructure stack that enables these systems to function at scale. The company develops networking components, optical interconnects and custom silicon used inside modern data centers. As AI clusters evolve from thousands to hundreds of thousands of GPUs, efficiently moving data across those systems becomes almost as important as the compute itself.

In other words, Nvidia provides the engines of this revolution, while Marvell helps build the highways that allow those engines to work together. When Jensen Huang publicly describes Marvell as a future trillion-dollar company, the market understands one thing: Nvidia likely views Marvell as one of the primary beneficiaries of the next expansion phase of AI Factories.

Beyond Marvell, however, there is a broader lesson here. Investing in AI may no longer be limited to buying Nvidia itself. It may also be worth paying attention to where Nvidia allocates capital, which companies it partners with, and which businesses Jensen Huang chooses to highlight publicly.

Start Trading on Hyperliquid

Trade 100+ perps with up to 40x leverage on a fully decentralized exchange.

---

Premium Members Only

In our latest Premium Alpha, we compiled the key companies mentioned by Jensen Huang during Computex 2026, Nvidia’s public equity holdings, and several businesses that could directly benefit from the next expansion phase of the AI industry alongside the technology giant.

Because today, following Nvidia may no longer be only about buying Nvidia stock. It is also about understanding where Nvidia invests, who Nvidia works with, and which companies Jensen Huang appears to view as the future winners of the AI cycle. Want to discover the full list? Head over to Premium.