Nvidia's Two-Front Strategy: Courting Anthropic While Arguing Against Its Regulation Playbook

Nvidia CEO Jensen Huang pushed back on slowing frontier AI, calling safety a technical issue, as his firm weighs a stake of up to $10 billion in Anthropic's IPO.

Jensen Huang used a San Francisco stage this week to draw a line in the sand — and then, according to media reports, moved to write a very large check across it. The Nvidia chief executive pushed back hard against calls to slow down frontier AI development, even as his company weighs a stake of up to $10 billion in the IPO of Anthropic, the very firm whose leader is urging caution.

Speaking at Salesforce's Dreamforce conference, Huang dismissed the idea that safety requires new legislation. In his view, it is a purely technical problem, and market forces already provide sufficient incentive to build secure systems — making fresh rules or regulatory intervention unnecessary. He went further in an interview with CNBC, calling proposed exemptions from US antitrust law completely unnecessary. That position directly challenges Anthropic CEO Dario Amodei, who argued in an essay over the weekend that the pace of leading model development should be slowed, floating both government mediation and antitrust carve-outs as tools to get there. OpenAI chief Sam Altman has similarly campaigned for greater caution and external testing. Huang found an ally in Meta's Mark Zuckerberg, who pointed to the liability exposure of model operators and the discipline of market mechanisms.

A $10 Billion Bet Inside the Storm

The regulatory clash has done little to cool the commercial ties between the two camps. Anthropic ranks among Nvidia's customers and is reportedly targeting a valuation of roughly $2 trillion in its listing. A stake of as much as $10 billion would fit a familiar pattern for Nvidia, which has steadily locked in relationships with the sector's leading developers to keep them anchored to its systems over the long haul.

That same week brought other partnership news. Salesforce unveiled "Koa," an enterprise reasoning model built on Nvidia's open-source Nemotron 3 Super architecture and trained on CRM data. The companies say it delivers a markedly lower error rate than existing solutions, with the first customer pilots beginning in October. Separately, Nvidia struck a collaboration with Palantir Technologies to build software-driven control for critical supply chains, starting inside Nvidia's own manufacturing operations and drawing on both Nemotron and Palantir's Foundry platform.

Analysts Chase the Demand Curve Higher

Surging appetite for accelerator chips keeps pulling earnings estimates upward. Erste Group Bank raised its fiscal 2028 earnings-per-share forecast to $15.78 from $12.70. Raymond James analyst Simon Leopold continues to see a multi-trillion-dollar opportunity in global AI infrastructure and carries a $515 price target on the stock. The bullish case rests on scarcity: Nvidia has guided to roughly $108.0 billion in revenue for the third quarter of fiscal 2027, and Huang says the existing supply chain covers only 70 percent of actual demand. That shortfall hands the company considerable pricing leverage, even as rivals bring alternative chips to market — Nvidia's deep entrenchment in data centers preserves a meaningful lead.

Risks remain operational rather than existential. Delayed supply chains or big customers tightening data-center budgets could trigger short-term swings, observers note.

The Tape Tells a Calmer Story

In trading, the shares added 0.6 percent to EUR 184.86, extending gains after a prior close of EUR 183.80 — a modest 0.7 percent advance that left the stock 9.2 percent below its 52-week high. For investors, the picture is one of a company betting that momentum in the enterprise sector will outrun regulatory noise, with strategic billions and persistent shortages underpinning its dominant position.