OctoML, which helps enterprises optimize and deploy their ML models, raises a $85M Series C led by Tiger Global, following its $28M Series B in March
Context & Ripple Effects
OctoML is on its third raise in roughly eighteen months: a $15M Series A in April 2020, then a $28M Series B in March 2021 that framed it as an ML acceleration service built on the open source Apache TVM compiler framework. The new $85M Series C compresses that cadence further, arriving just seven months after the B.
The lead investor is the throughline: Tiger Global has been systematically buying into this exact layer of the stack, having led Moloco's $150M Series C in August and later Run:AI's $75M Series C for AI workload optimization — making OctoML one more bet that the tooling between trained models and production hardware is where enterprise ML spend lands.
First-order effects
- OctoML gets the balance sheet to scale its TVM-based optimization and deployment service beyond what the March Series B funded, while Tiger Global adds another ML-infrastructure C-round to a portfolio that already includes Moloco and Run:AI.
Second-order effects
- Rivals in model optimization and serving — including Run:AI, which raised its own $75M Series C months later under the same lead investors — now compete against a peer whose open-source TVM root lets it claim hardware-agnostic reach without proprietary lock-in as the wedge.
Third-order effects
- Tiger Global's 2021 cadence of leading large C rounds at speed was reported at the time to be inflating a unicorn cohort, and its later markdowns of portfolio companies like Superhuman (down 45%) and DuckDuckGo (down 72%) suggest startups priced in this window may face down-round pressure if the pattern holds.
The trend: Enterprise ML is splitting into distinct layers — models, optimization/deployment tooling, and hardware — with crossover funds like Tiger Global racing to own the middle layer before it consolidates.