A young AI startup is starting to turn heads after raising a big funding round to tackle a problem many in the industry quietly struggle with—how inefficient AI systems can be.

Founded by Zain Asgar, Gimlet Labs has raised $80 million in a Series A round led by Menlo Ventures. The goal is clear: make AI run faster without wasting expensive computing power.

Right now, most AI workloads rely heavily on specific chips, mainly GPUs. But not every step of an AI process needs the same kind of hardware. Some parts need raw compute, others need memory, and some depend more on network speed. That mismatch often leads to resources sitting idle.

Gimlet Labs is trying to fix that by spreading AI workloads across different types of hardware at the same time. Its system can split tasks between CPUs, GPUs, and other specialized machines, depending on what each step actually needs.

In simple terms, it acts like a smart traffic controller for AI workloads—sending each part of the job to the most suitable hardware. The company says this can make performance up to 10 times better without increasing costs.

The idea also taps into a bigger issue. A lot of expensive infrastructure in data centres isn’t used fully. By some estimates, a significant chunk of computing power stays idle. Gimlet believes better coordination, not just more hardware, is the key to solving this.

To make it work, the startup has teamed up with major chip players like NVIDIA, AMD, Intel, Arm, Cerebras Systems, and d-Matrix.

The product isn’t meant for casual developers. It’s built for large AI labs and data centres that deal with complex, high-volume workloads. Companies can use it as software or through Gimlet’s own cloud platform.

Even though it launched publicly just a few months ago, the startup says it’s already seeing strong traction, with revenue in the multi-million-dollar range and a growing list of high-profile customers—though names haven’t been shared yet.

The founding team, including Michelle Nguyen, Omid Azizi, and Natalie Serrino, has worked together before, building infrastructure tools in the past.

With total funding now reaching $92 million, Gimlet Labs is betting that the future of AI won’t just depend on building more powerful chips—but on using the ones we already have much more efficiently.

Also Read: Meta Acquires AI Agent Social Network Moltbook, Founders Join Superintelligence Labs