Daily Digest
Daily Digest
Pomegra Startups
AI startup news: funding rounds, valuations, and exits across model labs, AI infrastructure, and applied-AI companies - and what each move means for the sector.
Optical interconnect startup Lumilens emerges from stealth with $700M+ in Series C funding at a $5.51B valuation to solve GPU networking bottlenecks inside AI data centers.
Chip interconnect startup Eliyan raises $145M Series C at a $1B valuation for its high-bandwidth die-to-die connectivity technology targeting AI chips.
Canadian biotech Biossil exits stealth with $70M to repurpose previously failed drug compounds for new therapeutic indications.
Stripe is reportedly in talks to acquire AI gateway startup OpenRouter for more than $7 billion.
Databricks secured $5 billion in fresh funding at a $190 billion valuation.
Salesforce veteran Efrat Rapoport's startup June exits stealth with $20M in pre-seed funding backed by Marc Benioff, Michael Dell, and Aaron Levie to automate enterprise AI deployment.
Open-source AI inference firm Together AI raises $800M Series C at an $8.3B valuation, led by Aramco Ventures, with 500 MW of compute capacity commitments secured separately.
AI inference startup Baseten raises a $1.5B Series F at up to a $13B valuation after achieving 20x revenue growth and processing over one billion inference requests per day.
Training and serving frontier models is capital-intensive in a way most software never was: compute, data, and specialist salaries all cost money before a product earns any. That is why AI rounds are larger and more frequent than the historical norm for the same company stage, and why the sector produces valuations that look disconnected from current revenue. The interesting figure in an AI round is rarely the raise - it is the multiple the round implies and who was willing to pay it.
The sector also splits into layers that behave differently: the model labs, the infrastructure companies selling compute and tooling beneath them, and the applied companies building products on top. A round in one layer often reprices the others, and coverage here notes which layer a story sits in - including when a private AI company's raise moves the listed chip and cloud names that supply it.
Most AI startups are private and not directly investable. The common routes are listed companies with significant AI exposure or stakes in private labs, funds that hold private positions, and waiting for an IPO. This is educational coverage, not investment advice.
AI valuations often price expected future capability rather than current revenue, and the capital required to train models means companies raise earlier and larger. That combination produces higher revenue multiples than most software categories.
Pomegra