Nvidia Hits $5.26 Trillion, Becomes Most Valuable Company Ever

The most valuable company in the history of public markets is a chip company. Not a technology conglomerate that touches everything. Not a bank sitting on centuries of capital. A company that designs graphics processors and that fifteen years ago was best known for making video game hardware.

On April 24, 2026, Nvidia's stock closed at $208.27, up 4.3% on the day, and that single session pushed the company's market capitalization past $5 trillion for the first time since October. That was the milestone most headlines focused on. But the stock kept moving. By April 28, 2026, Nvidia's market cap reached a record $5.26 trillion, making it the most valuable public company in history. No company has ever been worth more. Not ExxonMobil at its height. Not Microsoft in 1999. Not Apple at its peak. Nvidia.

To appreciate the scale, you need to hold it next to the other giants. Apple, which spent years as the default answer to "what's the most valuable company in the world," sits at $3.97 trillion. Microsoft is at $3.13 trillion. Amazon is at $2.82 trillion. Nvidia is now leading its nearest rival, Alphabet, by approximately $1 trillion. A full trillion dollars of separation at the top of the market.

The trajectory behind that number is almost surreal. Nvidia's stock has risen more than 14 times since late 2022. In roughly three and a half years, the company went from a well-regarded chip designer known primarily to gaming and research communities to the most valuable enterprise ever measured. The reason is not complicated: the artificial intelligence era runs on computing power, computing power runs on GPUs, and Nvidia makes the GPUs that the world's most important AI systems depend on. Not some of them. The overwhelming majority. No chip company in history had ever crossed $5 trillion before this month.

What made the April 24 session especially notable was its backdrop. The semiconductor sector had been navigating trade tensions with China, tariff uncertainty, and persistent questions about whether the AI infrastructure spending cycle was built to last. Then Intel dropped its Q1 earnings: $13.58 billion in revenue, beating expectations badly enough that the stock surged 24% in a single day. That was Intel's best single-day performance since 1987. AMD rose 14% the same session. The Philadelphia Semiconductor Index was already on an 18-day winning streak heading into that day. This was not one company having a moment. The entire chip sector was signaling at once: the AI buildout is real, it is accelerating, and it is NOT done running.

Nvidia's most recent quarter, Q4 of fiscal year 2026, showed revenue of $68.1 billion. Year-over-year growth of 73%. Gross margin of 75.2%. That is software-company margin territory on hardware. Adjusted earnings per share came in at $1.62, which was 82% growth year over year.

The guidance for Q1 of fiscal year 2027 was $78 billion in revenue. That implies roughly 77% growth. The company is adding more revenue in the current period than it used to generate in an entire year, and the growth rate is still accelerating.

CEO Jensen Huang has been blunt about what comes next. He has projected at least $1 trillion in cumulative revenue from Blackwell and Vera Rubin chip sales by end of 2027. When pressed on whether production could meet demand, his answer was direct: "In fact, we are going to be short."

What makes that projection credible is the spending on the other side of the equation. Amazon has announced approximately $200 billion in 2026 AI and cloud infrastructure capital expenditures. Microsoft is committing approximately $146 billion. Meta is planning in the range of $115 to $135 billion for AI infrastructure this year. Add those three together and you get approximately $700 billion in combined capex from a single group of customers in a single year. That money does not go into abstract cloud budgets. It goes into physical data centers. Those data centers run on GPUs. And for the most demanding training and inference workloads, those GPUs are almost always Nvidia's.

The competitive picture reinforces that dominance. Nvidia controls approximately 92% of the data center GPU market. AMD is real competition for cost-sensitive or secondary workloads, and its 14% gain on April 24 shows the sector-wide momentum is broad. But AMD is not challenging Nvidia's hold on frontier AI infrastructure. Meanwhile, Amazon, Meta, and Microsoft are all building custom AI accelerators. Trainium, MTIA, and Maia are real products in production deployment. But all three companies are still buying Nvidia hardware for their highest-priority work, because custom silicon takes years to mature into production-grade reliability, and Nvidia has years of accumulated technical lead.

Alphabet is in a particularly interesting position. Google unveiled new AI accelerator chips for H2 2026 cloud deployment, a long-term hedge against its own Nvidia dependence. Alphabet is simultaneously one of Nvidia's largest customers and a company actively developing the technology to eventually compete with it. That dynamic is not unique to Google. It is the posture every hyperscaler is quietly taking: buy Nvidia today, build the alternative for tomorrow. The difference is that today's procurement commitments are in the hundreds of billions, and tomorrow's alternative is still years away from matching what Nvidia can ship right now.

There is one near-term complication worth naming. Production targets for the Rubin platform have been cut. GPU unit targets came down from 2 million to 1.5 million for 2026 due to HBM4 memory certification delays. Vera Rubin server rack forecasts dropped from the 12,000 to 14,000 unit range down to approximately 6,000. These are real reductions. They do not change the demand story. But they do mean the supply shortage Jensen described is not solely a market overflow problem. Nvidia is also managing constraints on its own production side as it ramps the next generation.

The biggest variable hanging over all of this is China, and it is not a future risk. It is costing Nvidia money right now.

U.S. export restrictions on Nvidia's H20 AI chips to China triggered a $4.5 billion charge in Q1 of fiscal year 2027. Cumulative export controls have stripped approximately $17.1 billion in annual China revenue from the company. That is a country-sized hole in the model, and the political pressure is pushing toward making it larger.

Jensen Huang went to Washington in April 2026 and testified before Congress, arguing that the restrictions were counterproductive. His case was framed in national security terms, not just commercial ones. He said: "The day that DeepSeek comes out on Huawei first, that is a horrible outcome for our nation." The argument is that by cutting off American chip sales to China, the United States is accelerating the development of a competing Chinese technology ecosystem built around Huawei hardware. Once that ecosystem matures, the platform influence that comes with being the global standard goes with it.

He laid out the systemic risk directly: "It would be extremely foolish to create two ecosystems: the open source ecosystem, and it only runs on a foreign tech stack, and a closed ecosystem that runs on the American tech stack." When one lawmaker suggested the company was losing the long game, Huang pushed back hard: "You're not talking to somebody who woke up a loser. That loser attitude, that loser premise makes no sense to me."

Congress has not been swayed. The House Foreign Affairs Committee voted 42-2 to advance legislation that would reimpose Blackwell-class chip bans on China. Forty-two to two is bipartisan by any standard. The current administration has been running on a case-by-case review framework for export licenses, but that framework is now under direct legislative challenge.

The DOJ broke up a chip smuggling ring that had tried to export $160 million in Nvidia H100 and H200 chips to China under the fake "Sandkyan" brand label, specifically to evade the controls.

Jensen has estimated that compliant chip sales to China could add approximately $50 billion annually to Nvidia's revenue if restrictions were eased. That single figure is larger than the total annual revenue of most technology companies. Analysts watching the stock are still projecting higher. KeyBanc Capital Markets has a 12-month price target of $275 on NVDA, implying meaningful upside even from record levels.

There is also a low-level legal overhang. A class-action lawsuit alleging Nvidia concealed its crypto market exposure in 2018 is still pending in the San Francisco federal appeals court. It does not threaten the company's current position, but it is a reminder that Nvidia's relationship with market transparency has had friction before.

So here is where this stands. Nvidia is the most valuable public company in history at $5.26 trillion. Revenue is growing at 73% to 77% year over year at a scale that was unimaginable five years ago. The world's biggest technology companies are collectively committing approximately $700 billion in AI infrastructure spending this year, and most of it runs through Nvidia hardware. Jensen Huang has projected at least $1 trillion in product revenue from next-generation chips in the next two years and told you plainly that supply will not keep up with demand.

The tension in the story is real. Approximately $17.1 billion in annual China revenue has already been cut by export restrictions, more legislation is moving to extend those cuts further, and even the next-generation Rubin platform is running behind on supply. The market has priced all of that in and still landed at $5.26 trillion.

That is either the most rational market verdict in history, or the most expensive bet ever placed on a single outcome. The chips are placed. Whether the AI infrastructure buildout continues at its current rate is the only question that matters now, and every major hyperscaler with a checkbook has already voted yes.