TLDR
- Amazon Web Services is integrating Cerebras’s Wafer-Scale Engine processors into its infrastructure for AI inference workloads.
- The partnership spans multiple years, though specific financial details remain undisclosed.
- According to Cerebras, its technology delivers inference processing up to 25 times quicker than Nvidia’s GPU offerings.
- In January 2026, OpenAI entered into a separate agreement with Cerebras valued at more than $10 billion.
- A $1 billion funding round in February 2026 pushed Cerebras’s valuation to approximately $23 billion.
In a significant move to enhance its artificial intelligence capabilities, Amazon Web Services has forged a multiyear collaboration with chip innovator Cerebras Systems. The agreement centers on integrating Cerebras’s Wafer-Scale Engine processors throughout AWS’s global data center network, specifically targeting AI inference operations—the critical moment when an AI system generates responses to user prompts.
As the world’s dominant cloud computing platform, AWS has traditionally relied on proprietary silicon developed internally. The company’s Trainium chips, engineered by its semiconductor division Annapurna Labs, have been the backbone of its AI infrastructure. This latest partnership represents a strategic shift, with AWS planning to deploy Cerebras processors alongside Trainium to create a superior inference platform.
According to Cerebras, its Wafer-Scale Engine technology excels particularly during the “decode” phase of inference—the stage where AI models actually construct their responses—delivering performance up to 25 times superior to Nvidia’s graphic processing units.
AWS plans to market this as a high-performance tier. “If you want slow inference, there will be cheaper ways to go,” stated Cerebras CEO Andrew Feldman. Amazon has clarified that budget-conscious inference options utilizing Trainium exclusively will remain available.
A Big Week for Cerebras
This AWS announcement follows closely on the heels of another major milestone for the chip startup. OpenAI signed its own Cerebras deal in January 2026, a contract reportedly exceeding $10 billion in value. The OpenAI partnership aims to power ChatGPT infrastructure using Cerebras chips, with plans to implement up to 750 megawatts of computational power.
Cerebras’s momentum continued with a substantial $1 billion capital raise in February 2026. The funding round elevated the company’s total raised capital to $2.6 billion and established a corporate valuation near $23 billion. The investor roster includes prominent names such as Fidelity Management, Benchmark, Tiger Global, and Coatue.
The startup had previously attempted to go public, filing for an IPO in September 2024, but ultimately pulled the application approximately twelve months later.
Pressure on Nvidia
The AWS-Cerebras collaboration represents another headwind for Nvidia as competition intensifies in the inference segment. The artificial intelligence sector is experiencing a notable transition from training-focused workloads, where Nvidia maintains market leadership, toward inference applications that prioritize rapid response times.
Nvidia has responded aggressively to these competitive pressures. The company secured a $20 billion licensing arrangement with chip manufacturer Groq in December 2025. Additionally, Nvidia has announced plans to introduce a novel processing architecture incorporating Groq’s innovations in the coming months.
Nafea Bshara, Annapurna Labs co-founder and AWS executive, emphasized that the Cerebras collaboration aims to maximize performance while minimizing costs. “Our job is to push the speed and lower the price,” he explained.
Shares of Amazon’s AMZN stock declined 0.44% following the announcement.
