The Twenty Minute VC (20VC) — November 20, 2025

“Cursor Raises $2.3BN: Who Wins the Coding War | Peter Thiel and Softbank Sell NVIDIA: Analysed | Why Venture Capital Will Hit $1TRN and the Opening of Retail | Why Stripe and the Best Companies Will Never Go Public”

Guests: Rory, Jason, Tom Tunguz
Host: Harry Stebbings

Episode Overview

This episode dives deep into the hyper-charged state of venture capital, exploding AI valuations (particularly Cursor’s $2.3B raise at a $29.3B valuation), and the fundamental shifts in how software, capital, and liquidity flows are transforming the industry. The trio—Rory, Jason, and Tom Tunguz—bring wit, data, and skepticism to big questions: Is a trillion-dollar venture market likely? How durable are today’s AI advantages? Is public-market IPO liquidity a relic? What does retail capital’s “opening up” mean? And how do legendary investors read today’s “top-of-the-market” vibes?

Key Discussion Points & Insights

1. The Cursor Mega-Round and the Coding Agent Arms Race

• Cursor just raised $2.3B at $29.3B post, with the who’s who of venture participating.

  • Tom: “Product market fit for agentic coding is probably the best of any use case aside from search.” (05:08)
  • Jason: “This is just default and necessary—the way we code now… We’re going to approach 100% penetration per developer.” (06:26)
  • Rory: “Multiply 100 million by $5,000 a pop and you get a huge number. Even narrowed to ‘serious’ developers, that’s a huge company.” (10:04)

• TAM Calculation and “AI Coding” Productivity:

  • Market expansion is jaw-dropping: 100-200M developers globally × $5K/year per seat.
  • Jason: “I think Cursor can do a trillion if it has its current [penetration].” (07:33)
  • Debate over the durability and profitability (gross margins, switching, moats).

• Competitive Risks:

  • Cursor’s raw ingredient (large models) comes from competitors—platform risk.
  • Tom: “As the models improve in performance dramatically, people switch … but as the improvements start to asymptote, I’m going to stay where I am.” (12:44)
  • Rory: “Profitability and durability: those are the only two [key] issues.” (11:45)

Notable Quote

“Entry price counts when TAM is unclear. Winning is the only thing that counts when TAM is huge.”
— Rory (76:08)

2. Margins, Moats, and Commoditization in AI Software

• Gross Margin Questions:

  • Can Cursor’s business model sustain software-level margins or will model costs cap out?
  • Tom: “Do [AI coding companies] need 70% gross margin? … If you can get to 60% GM and sell a billion dollars in revenue with 100 headcount, you’re going to be kicking off cash totally.” (15:51)
  • Microsoft’s modeling efficiency improvements offer hope—“90% more tokens per GPU hour than 12 months ago.” (16:45)

• Moats and Switching Costs:

  • Inertia is real once tools embed into workflows and orgs.
  • Integration depth = lower churn.

• Risk of Price Wars:

  • Rory: “If anything like that semiconductor DRAM commoditization took place… it’d be terrifying.” (26:04)
  • Jason: “I’m hoping for deflation! 100 grand per agent… I need a bit of that theory fee stream to go beyond 12 agents in production.” (23:47)
  • The scenario: fierce pricing battles could wreck venture economics.

Notable Quote

"As the models improve, people switch. But as improvements start to asymptote, I'll stay where I am... Switching diminishes."
— Tom (12:44)

3. “Top 3” in Coding AI – Who Wins?

• Predicting Market Share:

  • Tom: “Codex [OpenAI] 60%, Anthropic 20%, Cursor 20%.” (16:54)
  • Rory: “My gut—Cursor, GitHub (Microsoft), and Anthropic or Cognition. OpenAI might not even be a top-three player here.” (17:11)
  • Jason: “But the rate of change on the tooling side—Replit V3, for instance—is so high, in 18 months maybe someone else blows everyone out.” (18:43)

• Market Settling and Inertia:

  • There’s a “bacon and the skillet” moment—once the fat (market share) congeals, it’s hard to change.
  • Rory: “There’s this wild period… but after three or four years you grab what share you can.” (20:15)

Notable Quote

“I think we’re at a place where agentic coding is no longer on this extremely steep improvement path… So people will stay where they are.”
— Tom (14:16)

4. Is Venture Becoming "The New Public Market"? The Trading Mindset, Velocity, and Late-Stage Entry

• Late-Stage Venture is Hyper-Accelerated, Possibly “Easier” Money:

  • Secondary rounds (like Ramp’s 13B → 32B in a single year) have returned to the “best business in the world” for those with access.
  • Harry: “I could be doing 10 to $25M checks into these high flyers ... and I could get the step up. But no, I go back to the craftsmanship of seed… why the fuck do I do that?” (32:44)

• Venture Traders, Not Just Holders:

  • Rory: “The secret to success in that business is just being a trader… It’s fucking trading.” (36:01)
  • But warns: unlike the public markets, you can’t trade out as easily on the downside.

5. Market Top? Reading the Nvidia, Thiel, and Softbank Moves

• Thiel and Softbank Selling Nvidia—a Sign of Market Top?
  • Rory: “People rarely sell stocks because they think they're going up, but it’s a small portion of their net worth. Softbank’s move is about rolling into risk (AI), not de-risking.” (37:49)
  • Tom watches credit markets for more telling signs: “Oracle CDS are 3x higher than Google/Microsoft… subprime auto delinquencies at record highs.” (38:37)
  • Conclusion: Red lights are flickering, but not yet screaming. A correction will be “fast and brutal” if demand wobbles. (43:27)

6. The Trillion Dollar VC Market, Retail Capital, and Bifurcation

• Will Venture Hit $500B or $1Tn by 2030? What Drives It?

  • Tom’s data: “In '08 it was ~$8B; in ‘21 about $300B; today, about $270–$275B. 100% chance we hit $500B.” (50:12)
  • Rory notes prior boom-and-bust (1999: $100B → $8B post-crash).

• Concentration:

  • Tom & Rory: “Half the market is driven by 4–5 companies (OpenAI, Anthropic, Stripe, SpaceX). If those work, the rest doesn’t matter much for pooled returns.” (53:02)

• Retail Capital Flood:

  • GPs and LPs see the opening of retail (CO2 has $3B in retail, GC is aggressive here) as huge inflows possibly decoupled from long-term returns.
  • Rory warns: eventually, poor returns will slow that flow, but the lag may be 5–7 years: “When people lose money, they figure it out… but it takes longer.” (64:40)

Notable Quote

“If OpenAI trades up at IPO, it’s roses for everybody.”
— Tom (54:16)

7. IPOs: Dead or Just Delayed? The Rise of Secondary Markets

• Are IPOs Going Extinct?

  • Tom: “Why go public? Transaction costs are higher, private secondary market is flush with cash, and there’s now an access premium over public multiples.” (59:23–60:53)
  • Massive late-stage rounds mean companies like Stripe, OpenAI, and SpaceX just don’t need IPOs to raise capital.

• The New Liquidity:

  • Tom: Secondary market transaction volumes are 10–12% of VC, up from 2–3%. “Liquidity dollars are up meaningfully… Nobody wants to go public.” (67:55)
  • Instead, “dollar cost averaging” secondaries—selling slices over time, as in the public market, becomes the norm.

• Will the Big Exits Eventually IPO?

  • Rory: “I think in the end…the big exits will IPO. The trends favor more private late-stage trading, but major liquidity for the top companies will, at least occasionally, require IPOs.” (69:27)
  • Tom: “Unless the public market is willing to pay a significant premium, the number of publicly traded companies will dwindle as PE picks them off.” (73:23)

Notable Quotes and Memorable Moments

| Timestamp | Speaker | Quote | |---|---|---| | 05:08 | Tom | “Product market fit for agentic coding is probably the best of any use case aside from search.” | | 06:26 | Jason | “This is just default and necessary—the way we code now… We’re going to approach 100% penetration per developer.” | | 10:04 | Rory | “Multiply 100 million by $5,000 a pop and you get a huge number…you can still multiply by five grand a year and get a huge company.” | | 11:45 | Rory | “Profitability and durability: those are the only two [key] issues.” | | 14:16 | Tom | “As the models improve in performance dramatically, people switch… but as the improvements start to asymptote, I’m going to stay where I am.” | | 16:45 | Tom | “Microsoft…now producing 90% more tokens per GPU hour than 12 months ago.” | | 36:01 | Rory | “The secret to success in that business is just being a trader… It’s fucking trading.” | | 43:27 | Tom | “If there’s some wobble, the magnitude of the correction will be fast and brutal. Everyone knows the tachometer is at a red line.” | | 50:12 | Tom | “In 2008, it was about $8 billion. In 21, it hit about 300. Today, 270. 100% chance [of $500bn].” | | 64:40 | Rory | “In the end, the only thing that matters is returns. The only question is how long does in the end take.” |

Timestamps for Key Segments

• Cursor’s Mega-Round/Bull vs. Bear Case: 04:39–11:36
• AI Model Competition/Moats: 11:36–18:43
• Market Share Debate: “Top 3” in Coding AI: 16:54–22:53
• Venture as “Trading”/Late-Stage Velocity: 32:44–36:37
• Market Top? Nvidia, Thiel, and Macro Market Risks: 37:06–43:27
• Trillion-Dollar Venture and Retail Flows: 50:03–66:00
• The New Public Market & Secondary Liquidity: 59:18–74:32
• Quickfire “Would You Rather”: 75:05–77:34

Summary Table: AI Coding Market Outlook

| Company | Predicted Share | Rationale (Condensed View) | |-------------------|----------------|---------------------------------------------------------------| | Cursor | 20–60% | First-mover, strong product, early enterprise uptake | | Codex (OpenAI) | 0–60% | Model power; unclear on final market dominance | | Anthropic | 20% | Top in coding, focus/traction | | Microsoft/GitHub | Major player | Bundling, enterprise distribution, possible step function up | | Cognition | Potential | “Clever guys in the corner” | | Replit/Lovable | Niche | Prosumer and agentic/low-end markets |

Tone and Closing Thoughts

The episode is fast-paced, delightfully irreverent, and crammed with opinionated, high-context debate. The guests approach the current market with both visible awe (“Product market fit for agentic coding is probably the best of any use case aside from search…”) and skepticism (“Profitability and durability: those are the only two issues, which is amazing…”), with a constant undercurrent of self-awareness about the potential for exuberance to lead to pain in the future.

• Irony: Veterans bemoan how “easy” late-stage can seem (“It looks easier with the brand and the platform…” — Rory, 32:44), while also warning constantly about the absence of liquidity on the downside and the inevitability of cycles.
• Wisdom for Founders & VCs:
  • Markets can be wild for much longer than you anticipate before correcting
  • Being human and maintaining relationships matters even in hot markets (49:09)
  • Entry price matters when TAM is unclear; when TAM is enormous, “winning is all that counts” (76:08)

For Listeners Who Haven’t Tuned In

If you want a rare, candid, and numbers-rich look at how the upper echelons of venture capital view today’s AI and software markets—plus expert skepticism of the froth, and pragmatic lessons on margin, moats, and the new era of perpetual private liquidity—this is your episode.

Listen to:

• The breakdown of Cursor’s mega-valuation and whether it signals a new era or bubble (04:39–11:36)
• The razor-sharp banter on AI agent moats and how fast markets settle (12:44–22:53)
• Segments on the “trader” mindset, market timing, and velocity of late-stage step-ups (32:44–36:37)
• The closing “quickfire” series—sharp, funny, and full of insight (75:05–77:34)