SambaNova Reaches $74 Billion Valuation as AI Chip Race Accelerates
The competition to build next-generation AI infrastructure continues to intensify, and SambaNova has emerged as one of the industry’s fastest-rising players. The U.S.-based AI chip startup has completed a new funding round that values the company at $74 billion, underscoring growing investor confidence in the enterprise AI inference market.
The fresh capital will support manufacturing expansion, accelerate product development, and strengthen SambaNova’s end-to-end AI platform, spanning custom silicon, integrated systems, software, and enterprise AI infrastructure. The company is also preparing to begin shipments of its fifth-generation SN50 AI accelerator in the second half of 2026, with SoftBank expected to become its first customer.
💰 New Funding Fuels Enterprise AI Expansion #
The latest financing round was led by General Atlantic, with participation from several prominent institutional investors, including:
- BlackRock
- Qatar Investment Authority
- Intel Capital
- Additional strategic investment partners
According to the company, the proceeds will be used to:
- Expand chip production capacity
- Accelerate next-generation AI hardware innovation
- Scale global enterprise deployments
- Support sovereign AI initiatives
- Grow partnerships with cloud service providers
- Continue investment across chips, software, and full-stack AI infrastructure
Rather than focusing solely on AI processors, SambaNova continues to position itself as a provider of complete AI computing platforms designed for enterprise-scale deployment.
🏦 JPMorgan Selects SambaNova for Private AI Infrastructure #
One of the most notable announcements accompanying the funding round is SambaNova’s partnership with JPMorgan Chase.
The financial institution plans to deploy SambaNova’s SN40L and SN50 accelerators to build an on-premises AI inference platform capable of supporting sensitive internal large language model (LLM) workloads.
Unlike public cloud deployments, financial institutions often require AI systems that offer:
- Strict data privacy
- Regulatory compliance
- Dedicated infrastructure
- Low-latency inference
- High operational reliability
SambaNova CEO and co-founder Rodrigo Liang believes this deployment represents a broader shift across highly regulated industries.
Rather than relying exclusively on hyperscale cloud providers, enterprises are increasingly investing in secure private AI infrastructure to run mission-critical generative AI applications.
As AI adoption expands, Liang expects similar demand from sectors including healthcare, government, manufacturing, and telecommunications.
🚀 From Stanford Startup to AI Infrastructure Unicorn #
Founded in 2017, SambaNova was established by a team of Stanford University researchers focused on developing AI hardware optimized for modern machine learning workloads.
The company’s leadership includes several well-known figures in the semiconductor industry.
Notably:
- Rodrigo Liang serves as co-founder and CEO.
- Intel CEO Lip-Bu Tan has served as Chairman of the Board.
- Intel Capital has participated in multiple investment rounds.
SambaNova has steadily expanded from designing custom AI processors into delivering complete AI infrastructure solutions that combine hardware, networking, software, and runtime optimization.
Earlier in 2026, the company raised $350 million in a Series E funding round. At that time, industry reports estimated its post-money valuation at approximately $2 billion.
Its latest funding represents a dramatic increase in valuation, reflecting heightened investor enthusiasm surrounding enterprise AI infrastructure and inference computing.
⚙️ SN50: SambaNova’s Fifth-Generation AI Accelerator #
A major driver behind the company’s momentum is the upcoming SN50, its fifth-generation AI accelerator introduced earlier this year.
SambaNova describes the SN50 as an inference processor specifically engineered for the rapidly emerging era of agentic AI, where AI systems execute multi-step reasoning, planning, and autonomous workflows.
Key architectural improvements include:
- Up to 5× higher compute performance than the previous SN40L
- Approximately 4× greater networking bandwidth
- Improved performance per watt
- Enhanced scalability for extremely large AI models
The processor is built around SambaNova’s Dataflow Architecture, which is designed to maximize utilization during inference workloads while reducing memory bottlenecks common in conventional GPU architectures.
📈 Designed for Massive AI Models #
Beyond raw compute performance, the SN50 emphasizes large-scale deployment.
Through multi-chip scaling, the platform is designed to support:
- AI models with up to 10 trillion parameters
- Context windows reaching 10 million tokens
- High-throughput enterprise inference workloads
These capabilities target organizations deploying increasingly sophisticated foundation models that require enormous memory capacity and sustained inference performance.
As context windows continue expanding and agent-based AI systems become more complex, scalable inference infrastructure is becoming a key competitive differentiator.
⚡ Performance Claims Against NVIDIA Blackwell #
SambaNova has also released benchmark data comparing the SN50 with NVIDIA’s Blackwell B200 GPU.
According to the company’s published results, when running Llama 3.3 70B, the SN50 delivers:
- More than 5× higher peak processing speed
- Over 3× greater throughput
While vendor-provided benchmarks should always be interpreted carefully until verified by independent testing, the results illustrate SambaNova’s strategy of competing primarily in high-performance AI inference rather than traditional AI training workloads.
The company’s emphasis is on maximizing utilization and throughput for production AI deployments where latency, efficiency, and operating cost are often more important than raw floating-point performance.
📦 Deliveries Begin in the Second Half of 2026 #
Commercial deployment of the SN50 is scheduled to begin during the second half of 2026.
According to Rodrigo Liang, SoftBank will become the first customer to deploy the new accelerator.
The partnership highlights growing demand among major technology companies for alternatives to conventional GPU-based AI infrastructure, particularly for large-scale inference environments.
As enterprises seek to reduce infrastructure costs while increasing AI deployment capacity, specialized inference accelerators are becoming an increasingly important segment of the semiconductor industry.
🤝 Deepening Collaboration with Intel #
SambaNova’s relationship with Intel has continued to strengthen throughout 2026.
Earlier this year, the two companies announced a long-term strategic partnership aimed at developing next-generation heterogeneous AI data centers.
The collaboration combines:
- Intel Xeon processors
- Intel GPUs
- Intel networking technologies
- Intel storage platforms
- SambaNova AI inference systems
Together, the companies aim to deliver cost-efficient enterprise AI infrastructure capable of addressing the rapidly expanding AI inference market.
The partnership also reflects an industry-wide trend toward heterogeneous computing, where CPUs, GPUs, custom accelerators, networking, and software stacks are tightly integrated to optimize performance for diverse AI workloads.
📊 Strategic Outlook #
SambaNova’s rapid valuation growth highlights a broader transition within the AI industry.
While much of the attention surrounding artificial intelligence has focused on training frontier foundation models, commercial demand is increasingly shifting toward AI inference, where models are deployed at production scale to serve enterprise applications.
This market requires infrastructure that balances performance, efficiency, scalability, security, and total cost of ownership.
With new capital, expanding enterprise partnerships, and the upcoming launch of the SN50 accelerator, SambaNova is positioning itself as a major competitor in AI inference infrastructure.
Whether the company can maintain its momentum against established semiconductor leaders such as NVIDIA, AMD, Intel, and emerging AI accelerator vendors will depend on the successful execution of its hardware roadmap, software ecosystem, and enterprise deployment strategy over the coming years.