Newsletter Archive

AWS SageMaker now supports bidirectional streaming for real-time inference, enabling sub-100ms speech-to-text by eliminating request-response latency. This post shows how to deploy Mistral's Voxtral-Mini-4B speech model using vLLM containers for live captioning, voice agents, and contact center analytics.

Amazon AWS ML · 13 min read Tools

Anthropic will pay xAI $1.25 billion per month for compute access under a new deal, cementing xAI as a major compute provider and giving Anthropic a critical alternative to Nvidia-dependent infrastructure amid ongoing GPU scarcity.

TechCrunch · 3 min read Industry

Anthropic commits nearly $45 billion to SpaceX over three years for computing resources to power Claude, marking a major infrastructure partnership in the race for AI compute capacity.

Bloomberg Tech · 1 min read Industry

SpaceX's S-1 filing reveals a $1.25 billion-per-month cloud services agreement with Anthropic through May 2029, giving the AI lab access to COLOSSUS and COLOSSUS II compute clusters while SpaceX trains Grok 5 on the latter. The deal underscores the scale of modern foundation model training infrastructure and Anthropic's compute hunger.

Simon Willison · 2 min read Industry

Nvidia posted record $81B revenue, crushing analyst expectations on explosive demand for its AI chips. The result signals sustained market dominance and capital availability for scaling AI deployments.

NY Post · 1 min read Market

Anthropic is paying SpaceX $1.25 billion monthly through May 2029—$15 billion annually—for GPU compute capacity, according to SpaceX's IPO filing. The deal underscores AI labs' acute compute scarcity and represents 83% of SpaceX's current annual revenue, signaling how critical hardware access has become to frontier model development.

Axios · 2 min read Industry

Nvidia hit an $82B quarterly revenue record as agentic AI systems move from labs into production. The timing reflects booming enterprise demand for inference infrastructure and the shift from single-model deployments to multi-agent workflows requiring heavier GPU utilization.

PYMNTS · 5 min read Market

AMD pledges over $10 billion in Taiwan investments to expand AI chip partnerships and packaging capacity, signaling an aggressive push against Nvidia's market dominance.

Bloomberg Tech · 1 min read Hardware