#anthropic Tech News.

Apollo and Blackstone have finalized a $35 billion package to lease TPUs from Anthropic. Broadcom is backstopping payments on the debt's largest senior portions. The deal involves leasing TPUs, a type of specialized computer chip. This move is significant for Anthropic, a leading AI research company. The package is a major investment in Anthropic's technology.

Apollo completed a $35 billion debt financing to purchase AI chips for Anthropic. The transaction supports Anthropic's AI chip development efforts. The funds are dedicated to hardware infrastructure for AI training. The move reflects Apollo's strategic focus on AI infrastructure investments.

Claude Design is a feature of the AI app Claude, launched by Anthropic to cater to average users. It allows users to create designs with AI assistance. However, users are reporting difficulties in using the feature, making it less impressive than expected. The feature's limitations have been exposed by users who were initially excited about Claude's capabilities. Claude Design's usability issues are a setback for users who were expecting a seamless design experience.

Anthropic's Mythos model boosted shares of CrowdStrike and Palo Alto Networks more than 70% between April and May. However, recent earnings showed shares lost 8% and 3% respectively. Analysts say AI benefits take time and are a multi-year process. Cybersecurity stocks sold off early in the year due to concerns about AI tools upending business models. Mythos, deemed too powerful to release, renewed enthusiasm for the sector.

SpaceX secured a Google AI compute deal following an Anthropic partnership prior to Google's IPO. The agreement enables Google to deploy AI workloads on SpaceX infrastructure. Google's AI operations will use SpaceX's orbital computing platforms. The deal supports scalable AI training at high altitude data centers.

Anthropic has enhanced Claude with chemical knowledge to assist in molecular research. The update enables Claude to interpret chemical structures and predict reaction outcomes. Users can now query reaction pathways and compound stability with detailed explanations. This capability supports scientific discovery in materials and pharmaceutical development.

Anthropic embedded engineers at the NSA for offensive cyber operations. The company published a report warning AI could build itself without human oversight. This capability could emerge within the next few years. The report urges global pause on AI development to prevent autonomous systems from acting without human control.

Amazon Bedrock releases a new console experience optimized for Anthropic and OpenAI-compatible APIs. The console supports GPT, Claude, and open-weight models via bedrock-mantle endpoint with OpenAI Responses, Chat Completions, and Anthropic Messages APIs. Users can compare up to three models side by side and view token usage, inference requests, and pricing per model. The project dashboard shows real-time inference metrics and model distribution by date and usage.