Wafer Raises $40M for AI Inference
Wafer raised $40M to scale AI inference optimisation, pointing to the next cost bottleneck in production AI.
AI infrastructure is moving from model training hype into the harder economics of running models at scale.
What happened
Wafer raised a $40M Series A to scale its AI inference optimisation platform. The company focuses on helping teams run AI workloads more efficiently across different hardware environments, with an emphasis on cost, latency and deployment reliability.
The round backs a more practical layer of AI infrastructure: not building another foundation model, but making existing and future models cheaper and easier to operate in production.
Why it matters
As AI usage grows, inference can become one of the largest recurring costs for companies deploying models. Training gets most of the attention because the upfront numbers are huge, but inference is where usage turns into an ongoing operating expense.
That creates room for infrastructure startups that can help companies squeeze more performance out of GPUs, choose the right hardware mix, reduce latency and avoid overpaying for compute. The value is especially clear for products with high-volume AI calls, real-time user interactions or enterprise workloads that need predictable performance.
The bigger picture
The next phase of AI infrastructure will not be won only by whoever owns the most chips. It will also depend on the software layer that routes workloads, optimises deployment and controls unit economics.
Wafer fits that shift. As companies move from experiments to production AI, they need systems that make model usage financially sustainable. Inference optimisation is becoming a picks-and-shovels category for the AI economy: less flashy than frontier models, but increasingly central to whether AI products can scale profitably.