Anthropic’s CEO stuns Davos with Nvidia criticism
Introduction The article centers on the U.S. government’s reversal of an export ban to authorize Nvidia’s H200 chips and an AMD chip line for approved Chinese customers, and the ensuing backlash from Anthropic CEO Dario Amodei at the World Economic Forum in Davos. It highlights tensions between AI hardware supply, national security concerns, and the strategic partnerships shaping the AI industry.
Key facts, terms, and numbers
The U.S. administration officially approved the sale of Nvidia’s H200 chips, along with an AMD chip line, to approved Chinese customers after reversing an earlier ban.
The chips in question are high-performance processors used for AI, and their export has been controversial.
Anthropic CEO Dario Amodei criticized the decision at Davos, calling it “crazy” and warning of “incredible national security implications” of AI models, using a nuclear proliferation analogy.
Nvidia is a central partner and investor in Anthropic, providing GPUs that power Anthropic’s AI models; Nvidia announced an investment of up to $10 billion in Anthropic and a “deep technology partnership” two months earlier.
Anthropic runs on servers from Microsoft, Amazon, and Google; its Claude coding assistant has a strong reputation among developers, and the company is valued in the hundreds of billions after raising billions in funding.
Closing The episode underscores how policy decisions on export controls intersect with aggressive private-sector AI development, potentially reshaping partnerships and competitive dynamics in the AI race.
Introduction OpenAI outlines a shift toward practical adoption of AI in 2026, emphasizing closing the gap between what AI can do and how people actually use it. In a blog post by CFO Sarah Friar, the company ties its heavy infrastructure spending to real-world value, outlining how intelligence could transform health, science, and enterprise, and hinting at new hardware ventures with designer Jony Ive.
Key points
2026 focus: practical adoption of AI, aiming to translate capabilities into real-world use, with opportunities highlighted in health, science, and enterprise.
Infrastructure and usage: about $1.4 trillion in infrastructure commitments as of November of last year; weekly active user and daily active user metrics are at all-time highs, driven by a flywheel of compute, frontier research, products, and monetization.
Revenue and product strategy: introduction of ads on the platform and the cheaper ChatGPT Go subscription worldwide; future growth expected to extend beyond current offerings with licensing, IP-based agreements, and outcome-based pricing.
Capital and partnerships approach: commitments to world-class compute require years-long planning; growth may outpace capacity at times, so OpenAI emphasizes a light balance sheet, partnering rather than owning, and flexible contracts across providers and hardware types; capital is deployed in tranches aligned with real demand signals, plus potential AI hardware devices with Jony Ive to be shown later this year.
Closing The emphasis on practical adoption signals a shift toward monetizable, real-world impact and broader societal applications, potentially accelerated by new hardware and innovative pricing models. This direction could reshape OpenAI’s business model and influence AI deployment across health, science, and enterprise.