Chinese AI Cloud Provider Infinigence Files for Hong Kong IPO

OpenAI has released GPT-6 Astra Ultrafast, a new model variant running on NVIDIA Blackwell GPUs that delivers up to 8x faster token generation than Astra Standard mode. The model is now available through the OpenAI API and to eligible ChatGPT Work and Codex users. OpenAI optimized the inference software using its own models to take advantage of Blackwell's architecture, with the performance gains particularly beneficial for coding agents and interactive applications that require rapid response cycles.

Nvidia and SoftBank have each completed their final $10 billion investments in OpenAI's funding round, fulfilling their respective $30 billion pledges. The investments represent the conclusion of commitments made as part of OpenAI's March announcement of $122 billion in total funding commitments. The completion of these major capital injections underscores continued institutional confidence in OpenAI's trajectory and AI infrastructure development.

Sharon AI, an Australian neocloud provider, has secured $356 million through a senior secured loan backed by its GPU assets and customer contract cash flows. The financing marks the latest instance of cloud providers leveraging their hardware and revenue streams to fund expansion. The deal reflects growing investor confidence in GPU-backed lending as a financing mechanism for infrastructure-heavy AI companies.

SpaceX's AI unit has shifted into selling computing capacity to major AI labs, generating billions of dollars in monthly commitments from clients including Anthropic and Google. The unit held talks with Microsoft over the summer about leasing compute resources, marking a significant pivot for Elon Musk from initial skepticism toward the cloud computing business. This move positions SpaceX as a major player in the compute-constrained AI infrastructure market.