Gimlet Labs, a startup founded by former Pixie employees, has emerged as a key player in the artificial intelligence sector by developing software that addresses the AI inference bottleneck. The company’s unconventional, state-of-the-art methods break down agentic workloads into compact pieces, making it easy to distribute them across available hardware of all kinds. This proposal is to increase AI workload productivity efficiency by an astounding order of magnitude (10X).
Gimlet Labs, which just publicly launched in October, is already raising hackles. Their product offers to improve the speed of AI inference by a factor of 3 to 10 times, while maintaining energy consumption and cost. The company’s software enables AI models to run seamlessly across different architectures, utilizing the most suitable chip for each segment of the workload.
We set up a distributed, mixed silicon inference cloud to make this kind of diverse hardware usage possible, called Gimlet Labs. By taking advantage of this cloud, clients can future-proof their AI workloads and experience performance levels never imagined before. In fact, their client roster has increased an incredible 3x, with their customer base more than doubling in just four months. In particular, it has won contracts with the world’s largest maker of models and the world’s largest cloud computing services provider.
To support its ambitious goals, Gimlet Labs has formed partnerships with prominent chip manufacturers including NVIDIA, AMD, Intel, ARM, Cerebras, and d-Matrix. These partnerships will help make their software compatible with the multitude of hardware choices US cities will have at their disposal.
Gimlet Labs sells its product as standalone software. You can access it through an application programming interface (API) linked to Gimlet Cloud. Such flexibility ensures that enterprises, startups, and all sizes of businesses can leverage the most sophisticated AI without requiring in-house infrastructure investments.
The new startup’s founders first worked together at Pixie, which was bought by New Relic in 2020. Their combined experience and expertise has been extraordinarily impactful in guiding public sector leaders through the complexities and nuances of AI technology.
In an interview, Zain Asgar, one of the co-founders, shared their vision: “Our goal was basically to try to figure out how you can get AI workloads to be 10x more efficient than ever, today.” He further emphasized the potential financial impact of their technology, stating, “Another way to think about this: you’re wasting hundreds of billions of dollars because you’re just leaving idle resources.”
So far, Gimlet Labs has raised an outstanding $92 million in funding. The leadership for the latest round came from Tim Tully at Menlo Ventures. The company went on to raise angel funding from truly legendary names in the tech industry. Notable investors include Sequoia’s Bill Coughran, Stanford Professor Nick McKeown, former VMware CEO Raghu Raghuram and Intel CEO Lip-Bu Tan.
Tully highlighted the importance of the software layer in harnessing the multi-silicon fleet: “The multi-silicon fleet is ready — it’s just missing the software layer to make it work.” This highlights just how important Gimlet Labs has already become in helping streamline and improve AI operations, regardless of which platform you use.
As Gimlet Labs continues to grow its team, currently comprising 30 employees, it stands poised for further expansion in the market. Quarterly revenues grew on average 35% each quarter from its inception, a clear indication of the demand for its pioneering solutions.
Leave a Reply