DeepMind's Alex Imas: agent token surge meets extreme frontier price war
FinanceYF5 · x · 2026-11-17
In an a16z interview, Google DeepMind's Alex Imas explains why lab revenue can stay flat even as token usage soars:
- The last few months have seen "extreme price competition" among frontier labs, far beyond six months ago
- Competition has shifted from who has the better model to price, as enterprises adopt models at scale
- Agentic AI uses far more tokens than chat, so rising usage raises costs and forces steeper price cuts
- Net effect: soaring usage no longer translates into revenue growth
This is the fuller original source behind the Chinese summary post of the same event.
More from Venture
- India's sovereign AI lab Sarvam slammed for mostly hosting Chinese models — itsOmSarraf_ · 2026-11-17
- Vibe-coded AI game Slotbound makes $356k in 48 hours from a first-time dev — tetsuoai · 2026-11-17
- Italian startup Mosaico raises €7.5M to fix robotics data for physical AI — lukas_m_ziegler · 2026-11-17
- OpenAI keeps charging a VIES-certified EU company 21% VAT it shouldn't owe, and support can't fix it — Creative_Strike8202 · 2026-11-17
- Snowflake's real fight isn't its Crunchbase list — rivals moved up to governance and semantic layers — YvesMulkers · 2026-11-17
- Fireworks AI tells investors chips are a commodity, its moat is software that runs models fast — hargup13 · 2026-11-17