What it sells
Custom AI chips.
Business model breakdown
Buyer: who owns the budget?
UnverifiedAI infrastructure or inference buyers (data-center operators, model developers) licensing a model-specific chip; HC1 offers API access.
Trigger: what makes them buy now?
Need for ~1000x efficiency vs software-simulated models; HC1 claims 17k tokens/s per user.
Workflow: what inputs become what output?
Any AI model is made hardcore through Taalas Foundry into a chip; fine-tuning supported; apps written in human language.
Economics: what it saves and how it charges
Sells model-to-silicon conversion (Taalas Foundry) and hardwired chips claiming 1000x efficiency; HC1 demonstrator runs Llama 3.1 8B at 17k tokens/s per user.
Distribution: how it reaches customers
Request API access and public chatbot demo; contact via info@taalas.com.
Funding history and coverage
| Round date | Stage | Amount | Investors | Source |
|---|---|---|---|---|
| 2026-02-19 | — | $169M | — | news.google.com |
Profiles summarize public reporting and company websites. Business-model notes are MaxBC's reading of that material, not confirmed facts; unknowns are left blank rather than guessed. Updated 2026-10-12.