Gimlet Labs has another $300 million to put behind its AI cloud expansion.
Andreessen Horowitz, the Silicon Valley venture capital firm also known as a16z, led the Series B, which values the AI infrastructure company at $3 billion.
Capital is arriving as the provider lays out a far larger data center buildout than it had just months ago.
Data center pipeline reaches gigawatt scale
In its announcement, Gimlet Labs says it is scaling toward hundreds of megawatts of managed AI infrastructure and has a gigawatt-scale data center pipeline. Contracted revenue has also reached billions since its$80 million Series A in March.
Company disclosures do not detail how much of the announced pipeline is already under development or how much contracted revenue has been recognized.
Series B funding brings total capital raised to $392 million. Sapphire Ventures joined as a major investor, while M12 and Arm came in as new backers. Proceeds are expected to support cloud operations and hiring as the company scales.
One cloud can span several chip architectures
Gimlet Cloud supports processors from NVIDIA, AMD, Intel, Arm, Cerebras, and d-Matrix. Its software decides where compatible portions of an inference workload should run and coordinates the work across the available hardware, allowing customers to use different processor types through one managed service.
Customers can access that service through Gimlet Cloud or deploy the software inside their own infrastructure. On-premises users can therefore keep workloads within customer-owned systems while Gimlet manages the underlying compute through the same platform.
Agentic AI is a particular target for the platform. These applications can string together repeated model calls and tool use, creating several compute stages within one job. Gimlet says agents can combine model calls with tool use and code execution within a single workflow.
Channel partners may see sourcing and integration work
MSPs, VARs, and systems integrators evaluating inference options for customers would still need to compare Gimlet with hyperscalers and other specialist AI clouds on cost, availability, and workload fit. No formal channel program accompanied the round, so any immediate role for partners is more likely to start with advisory and implementation work than resale.
Customer-hosted deployments could create a clearer opening for SIs and infrastructure-focused MSPs. Partners may be asked to connect the managed stack with existing systems, benchmark workloads before production, or support deployments after launch. VARs could also see procurement work where projects call for on-premises hardware.
Partners considering the service should verify live capacity and delivery windows against customer project timelines. Processor availability should also be confirmed for each deployment rather than assumed across every location or environment.
More news: ByteDance’s $29.6 billion loan could fuel new demand for data centers, compute, and infrastructure suppliers across Southeast Asia.