Artificial intelligence chipmaker Cerebras Systems Inc. said today it raised $1.1 billion in a late-stage funding, bringing its post-money valuation to $8.1 billion, as demand for AI inference services continues to surge.

The Series G round was led by Fidelity Management & Research Co. and Atreides Management, joined by well-known venture firms including Tiger Global, Valor Equity Partners and 1789 Capital. Existing backers Altimeter, Alpha Wave and Benchmark also participated.

Cerebras is best known for its wafer-scale processors, dinner-plate-sized chips purpose-built to accelerate AI training and inference. The company delivers ultra-fast inference services through its silicon infrastructure. In a recent demonstration, Cerebras said its system ran OpenAI’s newest open-source model, gpt-oss-120B, at about 3,000 tokens per second, a breakthrough in responsiveness for extremely large AI models.

Tokens are how an AI model breaks down text into manageable pieces for analysis and output generation. They can be words, parts of words or punctuation. How fast a model can read and generate tokens speaks to how quickly the model responds and completes generating answers.

Cerebras claims to be the fastest inference provider in the world, citing benchmark results showing it outperforms Nvidia Corp.’s graphics processing units by more than 20 times on both open-source and closed-source models.

“Since our founding, we have tested every AI inference provider across hundreds of models. Cerebras is consistently the fastest,” said Micah Hill-Smith, chief executive of leading benchmarking firm Artificial Analysis.

The company’s customer base spans the AI industry, from cloud providers and model developers to enterprises and government agencies. This year, AI leaders have partnered with the company, including Amazon Web Services Inc., Meta Platforms Inc., IBM Corp. and Mistral AI SAS. Other customers include enterprise companies and governments, such as GSK plc, Mayo Clinic, the U.S. Department of Energy and the U.S. Department of Defense.

The company said it serves trillions of tokens per month across its own cloud services, on-premise and across partner platforms. The company’s ultra-high speed inference capabilities have delivered over 5 million monthly requests on Hugging Face Inc., the leading AI hub for developers, making it the top provider on the site.

To meet rising demand, Cerebras announced in March that it began deploying its wafer-scale chips across six new cloud data centers in North America and France.

“Cerebras is turbocharging the future of U.S. AI leadership with unmatched performance, scale and efficiency — these new global data centers will serve as the backbone for the next wave of AI innovation,” said Chief Operating Officer Dhiraj Mallick.

With its new $1.1 billion infusion, Cerebras plans to expand processor design, packaging and system innovations for AI supercomputers. The company also said funds will go toward boosting manufacturing and data center capacity in the United States.

Images: Cerebras Systems