Nvidia's Double Play: Bankrolling Tomorrow's AI Customers While Uncle Sam Opens the China Door

Nvidia negotiates an anchor stake in Anthropic, which reportedly eyes a $100 billion IPO, as H200 exports to China win approval with a 25% fee.

Nvidia is no longer content to simply sell the processors that power the artificial intelligence boom. The chipmaker is now negotiating to anchor an investment in AI developer Anthropic — a move that would plant its flag squarely in the model-development layer it currently supplies from a distance.

Talks remain preliminary, but the ambition behind them is not. Anthropic is reportedly eyeing an initial public offering that could raise as much as $100 billion and value the developer at roughly $2 trillion. For Nvidia, an anchor stake would be anything but a passive financial bet: it would effectively finance a future customer, lock in long-term hardware offtake, and tether a leading model builder to Nvidia's own infrastructure.

That playbook is becoming a familiar one. On Friday, word emerged that Nvidia-backed British cloud provider Nscale is preparing a US listing, with filings revealing revenue growth of 1,252% in the first half of 2026. Add the company's deepening push into the software layer — where the central hub for open-source developers is steadily migrating under its corporate roof — and the picture is one of a company building a software and platform empire that extends well beyond silicon. Stitch software, infrastructure and model development together that tightly, and rivals face a wall that will not be easy to scale.

Washington Lends a Hand as Beijing Loosens the Reins

Political tailwinds are blowing in Nvidia's favor. The US government has cleared H200 exports to China, attaching a 25% fee to the sales, while announcing a new initiative to promote the technology sector and rejecting attempts to artificially slow the industry's growth. The opening could give Nvidia fresh momentum in what is the world's second-largest market, where strict export curbs had been visibly crimping business while Chinese competitors expanded their local share. With the H200 green light, Nvidia can once again serve Chinese cloud providers' demand more directly.

CEO Jensen Huang is making the most of the moment. In a CBS interview, he dismissed warnings of existential AI risk as unscientific, putting the probability of artificial intelligence destroying the world by 2030 at zero percent. New regulatory constraints, in his view, are unnecessary — existing liability law already covers the ground, and careful engineering can contain potential dangers. He favors maximum development speed instead.

That stance places him in direct opposition to prominent voices elsewhere in the industry. Senior figures at developers including OpenAI and Anthropic have recently floated binding guardrails and a controlled slowdown in model development, spurred in part by earlier safety researchers' warnings about risks that could slip beyond human control. Huang considers such scenarios irresponsible.

Growth Targets Meet a Stretched Valuation

Huang has also voiced confidence in global demand for computing power, targeting a doubling of chip sales next year compared with the current one. The engine behind that dynamic remains the enormous spending by large cloud providers on new data centers.

Shareholders have been rewarded, though not without risk. The stock closed Friday at EUR 193.10, up 1.1% on the day and 20% since the start of the year, leaving it just 4.6% shy of its 52-week high of EUR 202.50. That valuation embeds an expectation that the ecosystem keeps expanding without major disruption.

Near term, the opportunities still look like they carry the upper hand. Nvidia's ability to funnel its own profits into strategically pivotal companies creates a competitive advantage that will be hard to match. The regulatory flank, however, is widening — and investors would be wise not to underestimate it. Antitrust authorities are taking notice as Nvidia pushes deeper into software, platforms and stakes, and the company itself flagged during the Hugging Face announcement that state restrictions on open-source AI models, particularly those from China, could weigh heavily on the platform.

Should competition watchdogs block future acquisitions, or political intervention hit the software business, the very lever driving growth today could swing hard in the opposite direction.