General Compute announced Tuesday that it has raised $100 million in Series A funding to develop a heterogeneous inference cloud designed to accelerate artificial intelligence token generation.
The funding round was co-led by Blackbird and Abstract, with participation from Fuse Venture Partners, Village Global, and several other investment firms. The company’s cloud platform is designed to provide AI companies with on-demand access to both traditional GPUs and specialized frontier inference chips through flexible contracts.
By integrating advanced hardware from providers including Cerebras, SambaNova, and Positron AI, General Compute aims to deliver processing speeds up to 40 times faster than standard GPU-based solutions. The company provides a full-stack approach, managing the physical infrastructure, including hardware procurement, cooling, networking, and energy requirements.
The expansion arrives as the artificial intelligence industry shifts its primary focus from model training to inference. Market data highlights the scale of this transition, with Goldman Sachs projecting that demand will increase 24 times by 2030. Furthermore, Google reported that its token demand surged 330 times between 2024 and 2026.
To support its big-scale infrastructure buildout, General Compute has also secured a $400 million debt facility from Upper90. The company plans to deploy its hardware in colocation spaces and offers customers multiple consumption models, including dedicated bare metal access or service level agreements guaranteed by specific tokens per second.
The company’s first major deployments are scheduled to go live in the first quarter of 2027. General Compute currently has more than 50 MW of capacity in its development pipeline.
