NewsSep 14, 2026, 20:51 UTC

Anthropic signs $13.7B compute deal with Rumble as AI safety fears spike

What's the deal? Anthropic has signed a six-year, $13.7B compute agreement with Rumble (RUM) Group, deepening its infrastructure investment even as public debate over AI safety intensifies.

Why now? The deal landed Monday, days after a viral social media post from a former OpenAI and Anthropic researcher warned that developers are "racing straight to self-improving superintelligence and gambling with our lives." The post drew 170 million impressions in five days.

The context: Jacob Coxon, who spent three years on pretraining research at both companies, said neither is "acting responsibly." Senator Bernie Sanders responded by calling for a ban on superintelligence, while OpenAI chief executive Sam Altman delayed the company's initial public offering to 2027, citing safety concerns.

What's the endgame? The compute agreement gives Anthropic more capacity to train and run its frontier models. The scale of the commitment signals that developers are prioritising expansion despite the warnings.

Actions over words: Former Trump administration "AI Czar" David Sacks pushed back on calls to slow down, telling frontier developers to "go ahead." He argued that Anthropic and OpenAI hold "a duopoly on frontier intelligence" by market share, revenue growth, and model capability.

The market backdrop: AI capital expenditure and related tech investment account for roughly 50% of US gross domestic product, according to the source. S&P 500 earnings are expected to grow 34% in 2026, more than double the 15% forecast at the start of the year. On Monday, cybersecurity stocks including Palo Alto Networks, Fortinet, Okta, and CrowdStrike rose on the safety concerns.

The signal: The gap between what AI leaders say and what they do is widening. Warnings about existential risk are multiplying, but multibillion-dollar compute deals keep getting signed — suggesting the build-out is far from slowing.