Groq’s $20BN NVIDIA Deal | Why Sam Altman Doesn’t Care About Dilution & Invisible Unemployment 2026
Jason Lemkin and Rory O’Driscoll on 20VC dissect Nvidia’s $20B Groq acquisition, Meta’s $2.5B Manus buy, and Jason’s thesis that AI-driven ‘invisible unemployment’ is accelerating in 2026.
- Nvidia paid $20B for Groq — less than 1% of its market cap and under 20% of annual free cash flow — primarily to eliminate a low-latency inference competitor before margin pressure materialized.
- Groq had ~$4M revenue in 2023, ~$40M in 2024, and ~$175M at acquisition; the deal closed in under two weeks after Jensen Huang personally demanded it done before Christmas.
- Meta acquired Manus for $2.5B (25x ARR); Manus was at $100M ARR / $125M run rate, founders retained ~80% equity and each walked away with ~$500M.
- Lemkin argues Manus founders correctly identified a local maximum — gross margins are structurally low for an LLM orchestration layer that rivals like Anthropic and OpenAI are replicating.
- Navan IPO’d at 4x ARR despite 27% growth and non-GAAP profitability; panelists say it IPO’d out of necessity ($700M debt, $200M cash) not strength, and the IPO window is barely cracked open.
- Lemkin defines ‘invisible unemployment’ as companies hitting record ARR-per-employee while eliminating entry-level and mid-tier hiring — Shopify flat headcount for three straight years is cited as the template.
- Stanford CS graduates who skipped AI coursework are already struggling to find jobs; Lemkin says top-0.1% AI engineers have infinite offers while the other 99% of their class has diminishing prospects.
- Rory O’Driscoll identifies two distinct late-stage venture categories: companies at $10M–$400M ARR that cannot yet IPO, and a separate class (Stripe, Databricks, Revolute) that could IPO but rationally choose not to given cheaper private capital.
2026-01-08 · Watch on YouTube
