The Twenty Minute VCOpenAI's Multi-Billion Deal with AMD & Polymarket, Vercel and Supabase Raise Mega Rounds
CHAPTERS
OpenAI–AMD chip deal: warrants, leverage, and why Sam Altman “understands power”
The panel breaks down OpenAI’s multi‑gigawatt chip supply deal with AMD and the unusual structure: OpenAI receives warrants tied to AMD’s performance and shipment/buy-through conditions. They interpret the deal as a real-time display of negotiating leverage among OpenAI, AMD, and NVIDIA.
Déjà vu of the PC era: OpenAI as Microsoft, NVIDIA as Intel, AMD as second source, Microsoft as IBM
Rory draws a historical analogy to the DOS/Windows–Intel–IBM dynamic, mapping today’s AI stack onto the PC monopoly era. The discussion suggests Microsoft may have enabled a powerful partner that could evolve into a platform competitor.
How NVIDIA can tolerate competition—and why OpenAI still has leverage while losing money
They examine how NVIDIA likely views AMD as controlled competition and why Jensen Huang may accept some erosion to keep regulators and customers at bay. Simultaneously, they argue OpenAI’s user demand gives it leverage even with massive losses.
Dev Day apps inside ChatGPT: big vision, but ‘Slack integration’ déjà vu and unclear killer use case
The group debates OpenAI’s push to make ChatGPT a place to run and invoke third-party apps (e.g., Canva, Zillow, Spotify). They’re skeptical that conversational UI beats purpose-built UI for complex workflows, and note similar past hype cycles for app marketplaces and integrations.
AgentKit and fast-built agents: threat to automation platforms, but enterprise reality adds complexity
They shift to AgentKit and the promise of building agents quickly, questioning whether it commoditizes orchestration tools (e.g., n8n-like products). The conclusion is that enterprise agents need governance, orchestration, and product surface area beyond a quick demo.
Mega-raises at mega-valuations: Naveen Rao’s $1B raise and whether venture math is breaking
Harry challenges the viability of venture returns when founders raise enormous sums at very high entry prices. Rory and Jason argue that in deep infrastructure, “proven builders” compress early uncertainty, but the remaining question is whether markets are big enough to justify 50B–100B outcomes.
Venture is ‘forgiving on price’—until it isn’t: variance, comps, and growth persistence
They discuss why venture can sometimes survive bad pricing due to power-law outcomes, unlike PE or public markets. But Rory warns that comps-based logic is dangerous because investors must underwrite future value, and growth persistence assumptions can fail abruptly.
LP liquidity and secondaries: endowments selling stakes, friction, and why more liquidity could help
The conversation turns to endowments (Yale/Harvard and others) selling venture stakes and whether it becomes normal. They argue liquidity needs are rising as holding periods lengthen, but the secondary process is still messy and constrained by fund agreements.
IPO bar in 2025 and the Snyk case: growth slowdown, valuation resets, and PE as imperfect outlet
They assess what it takes to IPO now, using Snyk’s rumored ~$300M ARR and ~26% growth as a borderline case. They debate valuation implications and note that PE interest appears muted versus 2021–2023, making “take control of destiny” the dominant strategy.
Founder longevity and CEO transitions: when changing the CEO helps—and when it’s delusion
They reflect on how rare 15-year founder journeys are and whether replacing founders can re-accelerate growth. Rory distinguishes between post-PMF execution gaps (where a manager CEO can help) and pre-PMF struggles (where replacement rarely works).
Vercel ($300M at ~$9B) and Supabase mega rounds: ‘Captain Obvious’ infra bets vs ‘suicide round’ fears
They argue Vercel and Supabase are structural picks-and-shovels for the explosion of app building and vibe coding, making them intuitive “follow the developers” bets. The main risk isn’t the fundraising itself but whether market size and growth persistence hold at scale.
Does ‘kingmaking’ work? Early momentum, brand-name capital, and Legora vs Harvey in legal AI
They debate whether venture capital can ‘kingmake’ winners by piling in early and deterring competition. The legal AI example (Harvey vs Legora) illustrates that even heavily funded leaders can be challenged by superior product execution, though capital intensity raises barriers for #3/#4 entrants.
SPACs re-emerge: Chamath’s revised terms, incentive alignment, and IPO alternative dynamics
They discuss a new SPAC structure with less egregious sponsor economics than 2021, tying sponsor gains to post-merger performance. While still skeptical SPACs beat well-run IPOs, they see improved alignment and renewed experimentation in going-public pathways.
Polymarket’s $2B raise at ~$9B: regulation reversal, strategic NYSE/ICE involvement, and why so much capital
They unpack Polymarket’s massive round as a deregulation-driven legitimacy shift and a strategic move by exchange operators to own emerging electronic markets. The key question becomes whether prediction markets expand meaningfully beyond sports betting into broader ‘information markets.’
