English

FundingGimlet

Gimlet Completes $300 Million Series B to Accelerate inference via Heterogeneous Hardware

This article is a translation. Read the Japanese original

Gimlet announced that it has completed a $300 million Series B funding round. The round was led by Andreessen Horowitz, with participation from several other firms including Sapphire Ventures and Samsung Ventures.

The company is building a "multi-silicon cloud" that leverages different types of accelerators, such as GPUs, near-memory computing, dataflow architectures, and CPUs.

By assigning each phase of inference workloads to the most suitable silicon architecture, the company claims it can achieve a 5x to 10x speedup while maintaining power consumption levels.

Gimlet is focusing on the fact that industry workloads are shifting from training to inference. The company pointed out that with traditional uniform infrastructure, it is difficult to achieve both high throughput and low latency.

As power shortages in data centers become increasingly severe, the company stated that maximizing throughput per kilowatt will be the key to supporting the scaling of AI.


Source: Gimlet's Series B (Hacker News Frontpage, 2026-09-05)