INFERACT

AI Focus Areas

• LLM inference optimization
• Serving infrastructure for generative AI
• GPU utilization and scheduling
• Open‑source model acceleration

Key Products

• Inferact managed vLLM platform
• Enterprise‑grade LLM serving APIs

Market Position

Inferact sits in the rapidly growing layer of AI infrastructure focused on making large language models cheaper and faster to run in production. Spun out of academic work at UC Berkeley, it leverages vLLM’s strong open‑source adoption as a wedge to sell managed services and enterprise‑grade deployments. The company competes with other inference specialists like Modal Labs, Baseten, Fireworks AI, and in‑house solutions from hyperscalers, differentiating through cutting‑edge kernel and scheduling optimizations that can materially lower inference costs for heavy LLM workloads.

AGI Relevance

As models approach AGI‑like capabilities, the bottleneck will increasingly be safe, affordable, low‑latency deployment at scale rather than just training. Inferact’s work on making inference highly efficient—across GPUs and heterogeneous hardware—directly impacts how widely advanced models can be embedded into products and agents. Its technology can act as the ‘runtime’ for AGI‑class models, enabling interactive applications, autonomous agents, and continuous reasoning systems that must respond in real time.

Investment Highlights

Raised a $150M seed round co‑led by Andreessen Horowitz and Lightspeed at an $800M valuation in early 2026—exceptionally large for a seed‑stage company. The founding team includes core contributors to vLLM with strong ties to Berkeley’s RISELab ecosystem, and early users reportedly include large cloud and e‑commerce platforms.

Tags

• ai company