Nvidia Rides Record AI Boom Into August Earnings, but Washington Tightens the Screws Over China

Nvidia continues to sit at the very heart of the artificial intelligence revolution, cementing its status as the most important supplier of the hardware that powers the technology. The company recently delivered another set of record breaking results, yet its path forward is not entirely smooth. A growing wave of pressure from the United States government threatens to complicate its access to one of the world's largest markets.

A record breaking quarter

In its most recent quarter, covering the first period of its 2027 fiscal year, Nvidia reported total revenue of roughly 82 billion dollars, an increase of 85 percent compared with the same period a year earlier. The sequential growth was equally striking, with revenue climbing 20 percent, a jump of 13.5 billion dollars that marked the largest quarter on quarter increase in the company's history.

The engine behind these numbers remains the data center division, which generated an astonishing 75 billion dollars in revenue, up 92 percent year over year. Within that total, computing revenue reached 60 billion dollars while networking revenue came in at 15 billion dollars, a figure that nearly tripled compared with the previous year. These results underline just how dominant Nvidia has become in the infrastructure of modern AI.

Soaring profits and shareholder rewards

The profitability figures were just as impressive as the top line growth. Nvidia posted GAAP earnings per share of 1.87 dollars, comfortably ahead of the 1.77 dollars that analysts had expected. Free cash flow surged to 49 billion dollars, a substantial rise from the 35 billion dollars recorded in the prior quarter, demonstrating the company's remarkable ability to convert demand into cash at scale.

Flush with cash, the company moved to reward its shareholders generously. Nvidia announced a fresh share repurchase authorisation worth a colossal 80 billion dollars, signalling deep confidence in its own prospects. It also lifted its quarterly dividend by one cent to 25 cents per share. For the current quarter, management is guiding towards gross margins of around 75 percent, a level most firms can only dream of achieving.

Blackwell drives the boom

The star of the show remains the Blackwell architecture, which has been adopted by every major hyperscaler and cloud provider. The latest iteration, known as Blackwell Ultra or GB300, delivers a 2.7 times improvement in throughput alongside a 60 percent reduction in cost per token compared with the previous generation. Hundreds of thousands of these systems have already been deployed across the industry's leading players.

The roster of customers reads like a who's who of the AI world, with Anthropic listed as a strategic partner alongside OpenAI, xAI, Meta and Google. Demand is so intense that rental prices for older hardware have climbed, with H100 chips renting for 20 percent more since the start of the year. Meanwhile, the sovereign AI business, serving national governments, grew more than 80 percent across nearly 40 countries.

The China question looms large

Despite the euphoria, a significant cloud hangs over the company in the shape of China. Notably, Nvidia's guidance for the current quarter, which points to revenue of approximately 91 billion dollars, explicitly excludes any contribution from China data center compute. This cautious stance reflects deep uncertainty surrounding the licensing required to export its powerful H200 chips to the country.

That uncertainty has only intensified in recent weeks. The United States Commerce Department has launched an investigation into how Chinese artificial intelligence firms are gaining access to Nvidia processors through data centers located outside their borders. The scrutiny focuses not on physical smuggling of chips, but on computing capacity rented from facilities in countries such as Singapore, Malaysia and Japan.

Washington tightens the screws

In response to this mounting pressure, Nvidia has taken decisive action of its own. The company has reportedly cut the number of authorised buyers of its AI chips across Asia by more than half. It has also implemented tougher screening requirements, increased inspections and stepped up due diligence in the regions under the spotlight, seeking to demonstrate compliance and head off harsher regulation.

The looming policy threat is that Washington could expand its restrictions to cover rented computing access, not just the physical shipment of chips. Such a move would increase compliance costs for Nvidia and its data center customers, while closing off yet another channel through which Chinese developers currently reach its coveted hardware. It represents a delicate balancing act for a company caught between commerce and geopolitics.

All eyes now turn to the 26th of August, when Nvidia is scheduled to report its next set of results and shed further light on these tensions. The stock closed a recent session at 218.78 dollars, up around 17 percent for the year, as investors weigh spectacular growth against regulatory risk. As chief executive Jensen Huang has bluntly put it, compute is revenues and compute is profit, a mantra that captures the extraordinary stakes at play.