Hark builds next-generation GPU infrastructure and AI hardware for enterprises scaling large language models.
$700M
Investor undisclosed
A $700M raise for GPU infrastructure in May 2026 signals the market still believes there's room for non-NVIDIA players in enterprise LLM deployment—but the co-investment from NVIDIA, AMD, and Intel suggests this is less about disruption and more about carving out a managed-services layer on top of commodity chips. If you're building anything that touches model serving, inference optimization, or enterprise GPU orchestration, this validates that customers will pay for abstraction layers that hide hardware complexity, even if they're not buying the silicon itself.