pollar.news
text-only edition
← Back to top stories
Business · · 6 sources

Nvidia considers $10B anchor investment in Anthropic $100B public offering

Anthropic is preparing an initial public offering in New York to raise up to $100 billion at a $2 trillion valuation, with chipmaker Nvidia discussing an anchor stake of up to $10 billion ahead of the November US elections.

IPO structure and valuation targets

Anthropic is negotiating to secure Nvidia as an anchor investor for its planned initial public offering in New York, where the artificial intelligence startup seeks to raise up to $100 billion (86.2 billion euros). The share sale aims for a company valuation of approximately $2 trillion, with Nvidia considering a direct equity commitment of up to $10 billion. Anthropic submitted preliminary listing filings earlier in 2026, with the formal prospectus scheduled for release in September and the listing expected prior to the United States midterm elections in November. Neither Anthropic nor Nvidia provided official comment on the discussions. Anchor investors commit capital before public marketing begins, following models used when Nvidia and Amazon backed Arm, and when Saudi Arabia's Public Investment Fund backed SpaceX.

Selected mega IPO capital raise sizes
$B
SpaceX (June 2026)75
Anthropic (target)100

Market scale and revenue growth

The planned raise would exceed the listing completed by Elon Musk's SpaceX in June 2026, which raised $75 billion at a valuation of $1.77 trillion. It would also surpass Europe's largest recent public offering, the 9.4 billion euro listing completed by Porsche in 2022, by roughly tenfold. Dealogic figures indicate that total capital raised through US public listings reached $137 billion by the end of August 2026. Anthropic's financial expansion has accelerated alongside compute demand, with quarterly revenue crossing $11.5 billion in August 2026, a fourteen-fold increase compared to the same period a year earlier.

Compute commitments and hardware suppliers

The listing plans build on extensive operational agreements between the two companies. In November 2025, Nvidia announced plans to invest up to $10 billion in Anthropic, coinciding with Anthropic's commitment to purchase $30 billion in Microsoft Azure computing capacity powered by Nvidia GPUs. While Anthropic relies heavily on Nvidia graphics processing units to develop and operate its Claude chatbot, surging user demand has strained available computing infrastructure and prompted efforts to diversify component suppliers. In April 2026, Anthropic agreed to commit more than $100 billion over a decade to Amazon Web Services while deploying more than 1 million Amazon Trainium2 chips, alongside ongoing compute partnerships with Alphabet's Google.

Timeline of Anthropic and AI compute financing milestones
2025-11Nvidia announces plan to invest up to $10B alongside Anthropic $30B Azure compute deal
2026-04Anthropic commits over $100B over a decade to Amazon Web Services
2026-06SpaceX completes $75B listing at a $1.77T valuation
2026-08Anthropic quarterly revenue surpasses $11.5B
2026-09-08Mistral closes 3B euro financing round backed by Nvidia
2026-09-11Nvidia enters talks to commit up to $10B as anchor investor in Anthropic IPO

Circular financing and European comparison

Nvidia's discussions with Anthropic match investment structures observed across the artificial intelligence sector, where chipmakers invest in software developers that contractually purchase their hardware or lease processing capacity. The Bank for International Settlements has raised concerns regarding the transparency and disclosure standards of these circular arrangements. Three days before the Anthropic discussions emerged, Nvidia participated in a 3 billion euro funding round for French startup Mistral on 8 September 2026, which valued the company above 21 billion euros. Mistral, which previously received backing led by ASML, borrowed $830 million from seven commercial banks to purchase 13,800 Nvidia processors for a data center located south of Paris, aiming to supply one gigawatt of compute power by 2030.

Read the full version on pollar.news →

Sources