
Anthropic Files SEC S-1 for Planned IPO
Anthropic filed its S-1 on October 26, revealing its intended Nasdaq listing while leaving IPO pricing and timing open.

Anthropic filed its S-1 on October 26, revealing its intended Nasdaq listing while leaving IPO pricing and timing open.

NVIDIA is adding new models and agent workflows to its RTX Spark PCs, putting more AI work on laptops as its November earnings call approaches.

Four AI companies have backed a shared publisher preference mechanism, but the controls remain voluntary and implementation is not yet uniform.

Google’s Project Suncatcher prototype successfully ran an AI model on a TPU in orbit, advancing an ambitious but unproven vision for space-based computing.

Apple plans to begin testing Apple Intelligence in China on November 4, bringing selected AI features to iPhone users through a partnership with Alibaba.

OpenAI has reportedly lowered its fundraising goal to about $20 billion at a $1.25 trillion valuation. The terms remain unconfirmed by the company.

As companies deploy software agents that can perform tasks rather than merely answer questions, identity providers are trying to control who authorizes each action and how those permissions can be withdrawn.

IT consultant Capgemini lifted its 2026 sales-growth target. Acquisitions help the increase, while unchanged cash and margin goals temper the headline.

Toyota will build Waymo-equipped bZ4X vehicles in Japan. Testing starts in 2027, while fleet size, launch locations and commercial timing remain open.

The CMA is examining whether Meta’s business terms unfairly limit publishers’ control over their content. The proposed changes remain on hold.

Tokyo Electron’s new full-year forecast targets ¥1.169 trillion in operating profit, even as investors question returns on AI spending.

Reddit will test a primarily logged-in experience outside the US. Sales rose 53%, but declining search referrals threaten its daily audience.
Reporting and explainers from the full archive.