Gruve, the enterprise AI services platform founded by former Rahi Systems executives, has closed a $50M Series A extension that values the 18-month-old company at an undisclosed amount. Total funding now stands at $87.5M, following an April 2025 $20M round led by Mayfield and a previously unannounced $17.5M seed.
The company's pitch is straightforward: US data centers have stranded capacity from overprovisioned power builds and hardware sitting idle between training runs. Gruve aims to monetize that headroom by running inference workloads, which require less compute intensity than training but still generate revenue during off-peak periods.
This approach addresses a practical problem. Training large language models demands sustained GPU clusters running at 80-90% utilization. Inference workloads are bursty, often leaving GPUs idle 40-60% of the time. Data center operators looking to improve margins are experimenting with dynamic allocation strategies, splitting capacity between training tenants paying premium rates and inference customers filling the gaps.
Grouve's founders previously sold Rahi Systems to Wesco for $225M, bringing data center procurement credentials to the model. The firm positions itself as outcome-based rather than hourly consulting, charging for results like deployed AI agents rather than implementation hours. Services span AI-ready infrastructure design, Cisco and Juniper network partnerships, and Salesforce CX implementations.
The real test will be execution at scale. Idle capacity monetization sounds efficient in theory, but coordinating multi-tenant workloads across inference and training cycles introduces operational complexity. Power headroom also varies by facility design, particularly as operators evaluate 800 VDC architectures for next-generation AI infrastructure versus traditional 480V systems.
Grouve acquired Acuvity AI at launch in July 2024 to accelerate agent development. The funding round timing suggests momentum, though the company hasn't disclosed customer names or revenue figures. With data center inference demand forecast to grow 300% by 2028 according to industry projections, the market opportunity is clear. Whether Gruve can capture enough of it to justify the valuation remains to be seen.