Fast by Design: The Infrastructure Powering the Agentic Era | Rodrigo Liang | RAISE Summit 2026

Capital Milestones and Valuation:

• SambaNova announced the first close of a $1 billion funding round at an $11 billion valuation.
• JP Morgan has officially selected SambaNova as its preferred inference provider.

Technical Strategy and Performance:

• Focus is on running multi-trillion parameter models in full precision (avoiding quantization) to maintain accuracy while achieving speed.
• The platform delivers a 5x to 10x performance advantage in the "decode" phase of inference on specific models compared to competitors.
• Heterogeneous disaggregated inference combines NVIDIA Blackwell pre-fill capabilities with SambaNova decode, resulting in a 2x to 3x aggregate throughput increase on existing infrastructure.
• The architecture supports "VC2" (Virtual Compute Cloud 2), demonstrated at Computex running side-by-side with B300s.

Deployment and Partnership Ecosystem:

• Partnerships with NVIDIA and Intel provide large-scale capital backstops and infrastructure support.
• SambaNova targets enterprises with a "hybrid" approach: offering air-gapped on-premises appliances for security and cloud-based solutions for scaling.
• The company leverages a network of 94 software subsidiaries with over $35 billion in cumulative revenue to drive enterprise distribution and trust.

Market Positioning vs. Competitors:

• Vs. NVIDIA: SambaNova explicitly states it does not aim to "beat" NVIDIA; instead, it aims to increase productivity and speed within NVIDIA-centric data centers to serve as a complementary solution.
• Vs. Frontier Labs (OpenAI, Anthropic): The strategy prioritizes the enterprise sector's need for data sovereignty, privacy, and the ability to run open-weight models, rather than competing directly for the "general intelligence" market currently dominated by labs.
• Vs. Pure-Play AI Hardware (Cerebras, Grok): SambaNova differentiates itself by supporting full-precision inference on massive models, whereas competitors often rely on quantization which may sacrifice accuracy.

Enterprise Adoption and Use Cases:

• Enterprises are shifting from experimentation to live production of multi-agent systems, where speed is identified as the primary bottleneck for agent-to-agent interaction.
• Workloads are bifurcated by latency requirements: real-time customer service agents require maximum speed, while batch processes like document ingestion can run at standard speeds.
• The company anticipates a market with multiple winners, driven by enterprises' need for choice over a potential duopoly of OpenAI and Anthropic.

Open Source and Geopolitical Outlook:

• SambaNova maintains a bullish long-term thesis on open-source models, citing historical precedent (e.g., Linux vs. Solaris) where open solutions eventually capture significant market share due to cost and flexibility.
• The company acknowledges geopolitical risks, including the dominance of Chinese open-source models and potential restrictions, but prioritizes providing customers with heterogeneous model choice regardless of origin.
• The core argument for open source is preventing vendor lock-in and ensuring price competition, allowing enterprises to select the most cost-effective model for specific use cases.

Future Outlook and Competitive Moat:

• Distribution is identified as a key differentiator, utilizing the sales force of 94+ software companies to penetrate trusted enterprise environments for air-gapped and hybrid deployments.
• SambaNova expects the market to continue expanding with larger, self-learning models, necessitating scalable capacity solutions that support the full range of enterprise security and compliance requirements.