SambaNova Systems raised $1 billion in financing at a valuation of $11 billion, marking one of the largest funding rounds for an AI chip startup challenging Nvidia's dominance.

The Palo Alto-based company sells inference chips designed specifically for AI workloads. Its latest processors are packaged as complete server units that can be deployed directly in data centers.

SambaNova's inference chip business is scaling rapidly, according to the company. The fresh capital will accelerate deployment of server racks that customers are demanding.

Major model releases drive competition

Meanwhile, OpenAI introduced GPT-Live, a full-duplex voice model that can listen and speak simultaneously while handling natural conversational cues. The system can delegate complex tasks to GPT-5.5 while maintaining dialogue flow.

SpaceX's AI division launched Grok 4.5 as its strongest model for coding and agentic tasks. The company noted the model was trained alongside Cursor, the AI code editor.

Cognition released SWE-1.7, which the company claims reaches frontier-level intelligence at significantly lower cost. The model is now available in Devin across web, desktop, and CLI platforms.

Meta deployed Muse Image across its applications, including Instagram Stories and WhatsApp. The image-generation model can create and edit images, perform multi-reference composition, and draw from public Instagram photos when users tag accounts.

Infrastructure and tooling advances

ByteDance debuted Seedream 5.0 Pro, a multimodal image-creation model designed for production work rather than single-shot outputs. The system offers precise editing capabilities and supports over 10 languages.

Mistral AI released Robostral Navigate, an 8-billion parameter model enabling robots to navigate environments using only a single RGB camera.

OpenAI acquired Northslope, an applied-AI consultancy, adding hundreds of forward-deployed engineers to help integrate AI systems into enterprise operations.

SambaNova is strongly considering an IPO in 2027, positioning itself as a key alternative to Nvidia's AI chip ecosystem.