Baseten closed a $1.5B Series F round in June 2026, the largest single raise in a cluster of AI inference funding events.1 Groq pulled $650M to expand its inference cloud infrastructure.2 Upscale AI completed a Series A extension, bringing its total financing to $500M.3
The three rounds concentrate capital in one layer of the AI stack: inference — the process of serving trained model outputs to end users at scale.
The funding surge comes as enterprise AI shifts from pilot programs to production deployments. Running models at scale for millions of daily queries demands infrastructure investment well beyond what training pipelines require.
Nvidia moved in parallel. The chip giant acquihired Groq's founder and key team members,4 signaling strategic interest in the inference talent market. The move extends Nvidia's position from hardware into the inference services stack.
In semiconductor supply chains, Air Products secured a long-term liquid helium supply agreement with an Asian semiconductor manufacturer.5 Liquid helium is essential for chip fabrication equipment. The deal signals sustained fab demand, supporting the GPU supply pipeline feeding AI infrastructure growth.
For investors, the June cluster suggests consensus that inference infrastructure is the next capital-intensive phase of the AI buildout. Earlier funding cycles targeted model training and foundational research. Capital is now rotating toward the deployment layer.
Baseten, Groq, and Upscale AI are competing to own the API and compute layer where enterprises connect AI models to production applications.
Tracking signals over the next 12-18 months: GPU order volumes from Nvidia, AMD, and TSMC; data center buildout announcements from these inference providers; and capex versus revenue trends from cloud providers offering AI services. Hardware suppliers and semiconductor fabs are positioned to benefit if inference demand scales as the funding rounds imply.
