Baseten and Fireworks each raise $1.5B within weeks, marking capital's rotation from training to inference serving
Baseten closed a $1.5B Series F and Fireworks AI raised $1.5B within weeks of each other, both for enterprise inference, with Together AI raising again in early July. Meanwhile OpenAI and Anthropic together took $217B — 43% of all startup funding in H1. The barbell is now unmistakable.
Two $1.5B rounds into inference serving inside a single quarter is a thesis being expressed, not a coincidence. The bet is that model weights commoditise faster than the operational competence to serve them at latency, scale and cost — and the open-weight release cadence of the last month is the strongest evidence yet for that view.
The concentration figure is the other half of the picture. Two labs absorbing 43% of all startup funding means the middle of the market has effectively been cleared: capital is going either to the two firms training frontier models or to the firms serving everyone else's. There is very little being funded between those poles.
For anyone building on models rather than building them, the practical implication is that serving is where the negotiating leverage is accumulating. A well-capitalised inference layer sitting between open weights and enterprise buyers is positioned to set terms in a way neither the labs nor the buyers can easily route around.
Crunchbase News — North American Startup Funding Shattered Records In First Half Of 2026 → · Crescendo AI — Latest AI Startup Funding News and VC Investment Deals — 2026 →