SambaNova Raises $1 Billion at $11 Billion Valuation, Posing a Strong Challenge in Nvidia’s Inference Track

TradingKey - On Wednesday, AI chip startup SambaNova announced the completion of its $1 billion Series F funding round, boosting the company's valuation to $11 billion and making it a new decacorn in the AI chip sector.

This funding round was led by General Atlantic, with participation from Seligman Ventures, T. Rowe Price, and Capital Group. Notably, SambaNova had already raised over $350 million from investors, including Intel ( INTC ), in the first half of this year, and has established a deep partnership with Intel.

Rodrigo Liang, co-founder and CEO of SambaNova, stated at the Raise AI summit in Paris that the explosion of inference technology has unlocked an entirely new growth space for the company. As an independent enterprise, they can quickly respond to market demands and expand their business into a wider range of industries.

The funding will help the company accelerate the deployment of rack systems urgently needed by customers and further scale up its operations. He also revealed that the company is seriously considering an initial public offering (IPO) in the United States in 2027.

This funding event once again highlights capital markets' frenzied pursuit of the AI chip sector, particularly bets on challengers to Nvidia.

As AI technology shifts from the model-training phase to large-scale inference applications, the inference chip market is becoming the industry's new competitive focus.

SambaNova's Differentiated Competition

Unlike giants such as Nvidia ( NVDA) that focus primarily on GPUs for large-scale model training, SambaNova has centered its business on the AI inference chip market.

Inference chips are mainly used to run already-trained AI models, with the core objective of reducing costs while maintaining operational speed. As the deployment of complex AI models accelerates, the importance of the inference stage has become increasingly prominent, driving explosive growth in market demand.

SambaNova's core product is a server unit integrated with its latest SN50 chip, which can be deployed directly in data centers. Its architectural design is fundamentally different from Nvidia's GPU architecture, focusing more on the optimization of inference scenarios.

Beyond the inference chips themselves, SambaNova is also betting heavily on on-premise deployment for enterprises. Customers can install its servers in their own data centers to run AI models internally without relying on third-party cloud service providers. This model offers distinct advantages in data security and processing speed, making it particularly attractive to highly data-sensitive industries such as finance and healthcare.

Recently, JPMorgan Chase ( JPM) announced it would deploy SambaNova's systems to handle demanding enterprise AI workloads and on-premise inference tasks.

Rodrigo Liang stated that for industries like banking, where data is the lifeline, on-premise deployment ensures that AI systems run within the firewall, allowing enterprises to maintain complete control over their models and private data to achieve highly secure and private AI applications. This on-premise deployment strategy has carved out a unique growth path for SambaNova in the highly competitive AI chip market.

Competition Heats Up in the AI Inference Chip Market

The rise of SambaNova is just a microcosm of the intensifying competition in the AI inference chip market. Currently, a large number of startups focusing on inference chips are emerging in the market, attempting to challenge the dominance of giants like Nvidia in this burgeoning sector.

Park Sunghyun, CEO of South Korean AI chip company Rebellions, recently revealed that the company plans to conduct an initial public offering on South Korea's KOSPI market in the first or second quarter of 2027, further demonstrating the heat in the inference chip sector.

In the face of competition in the inference chip market, Nvidia is also actively positioning itself. Last year, Nvidia signed an agreement with inference chip startup Groq to license the latter's relevant technology, in an attempt to consolidate its leading position in the AI chip field.

As AI model applications gradually transition from the training phase to large-scale inference, the scope of competition in the AI chip market is expanding further, with technical architectures, deployment models, and industry solutions all becoming focal points of competition.