👋 Hey, Sahil here — Welcome back to Venture Curator, where we explore how top investors think, how real founders build, and the strategies shaping tomorrow’s companies.

---

Big idea + report of the week

How big is AI spend, really? Bigger than you think.

Fresh analysis comparing AI infrastructure investment to the largest capital mobilisations in U.S. history shows just how fast the AI economy is scaling and how far it still has to go.

• AI vs historical megacycles:

  • World War II remains the biggest economic mobilisation ever at 37.8 % of GDP, followed by WWI (12.3 %), the New Deal (7.7 %), and the railroad boom (6.0 %).

  • AI infra spend is already at 1.6 % of GDP — bigger than the entire telecom bubble at its peak (1.2 %).

• Corporate CAPEX is exploding:

  • Microsoft: $140 B

  • Google: $92 B

  • Meta: $71 B

  • OpenAI alone is reportedly planning $295 B by 2030, signalling the scale of the compute arms race.

• What 2030 could look like:

  If OpenAI represents ~30 % of total AI infra spend, the market would hit $983 B annually by 2030, roughly 2.8 % of U.S. GDP.

  Hitting the railroad era’s 6 % equivalent would require ~$2.1 T per year, meaning today’s tech giants would need to 5–7× their current spend.

AI infra spending has already become one of America’s largest investment categories, and it’s only in the early innings. While still far from the scale of wartime mobilisation, the compounding CAPEX cycle from OpenAI, Microsoft, Google, and Meta could reshape the U.S. economic landscape for the next decade.

---

Europe’s valuation gap with the U.S. is closing, but only at scale.

Yoram Wijngaarde’s latest analysis shows that European startups still face a 12 % valuation discount compared to U.S. peers at the Series A stage. The gap reflects a persistent cost and capital‑availability imbalance in early‑stage venture rounds.

• 12 % early‑stage discount: European Series A valuations remain roughly 12 % lower than U.S. equivalents.

• Parity at $100 M+ valuations: Once startups cross the $100 M pre‑money threshold, the gap nearly disappears — European unicorns are often priced on par with U.S. counterparts.

• U.S. still dominates deal flow: Despite convergence at the top, U.S. founders capture larger and more frequent rounds, maintaining a structural edge in access to late‑stage capital.

Europe’s venture ecosystem is maturing fast, but scale is the great equaliser; the valuation gap narrows only once startups break out of the early‑stage orbit.

---

VC valuations hit new highs, but investor returns are still near decade lows.

PitchBook’s new Q3 2025 US VC Valuations & Returns Report captures one of the most uncomfortable contradictions in the market right now: startup valuations look like 2021, but VC returns look like 2010.

• Valuations are back to peak levels

  Across every stage, the median pre‑money valuation is now equal to or above 2021 levels. AI dealmaking is the biggest accelerant, but a second driver is multi‑stage funds pushing aggressively into seed and Series A, bidding up prices with looser valuation discipline.

• Seed valuations hit a fresh record

  Median seed pre‑money valuation in Q3 2025: $13.9 M, a new all‑time high.

• Returns tell a very different story

  Despite rising valuations, VC’s rolling 1‑year IRR is only 3.1 %, and cash flows to LPs have been negative since 2022. The 2020 VC vintage now has the worst TVPI after 5 years of any vintage since 2010, showing how badly late‑cycle checks are struggling.

• Liquidity pressure is intensifying

  More unicorns have gone public in 2025 than in the prior few years, but the overall IPO count will still only match the last three years. Large‑scale M&A hasn’t meaningfully rebounded either, leaving many funds dependent on 2026 outcomes to avoid prolonged underperformance.

• Why valuations keep rising anyway

  Dealmaking is climbing, and AI startup costs are dropping, letting founders justify richer terms. The market is effectively betting that liquidity catches up later — a narrative that has not yet materialised, but investors are hoping 2026 cracks open the exit window.

Bottom line: VC is facing another valuation/returns disconnect. Prices are rising again, but distributions aren’t. Unless liquidity accelerates in 2026, many funds — especially those raised in the 2020–2022 boom years — will face steep pressure from LPs expecting real returns, not just paper markups.

---

Turning VC into a network‑effect business

Most founders think great VCs win because they have money, brand, or a famous GP. David Booth’s insight is sharper: the next decade of VC will be dominated by firms that operate like network platforms, not service providers.

Why traditional VC hits a ceiling

The old model is intimacy‑driven: partner time, partner intros, partner advice. That model collapses at scale. Every new investment dilutes attention. Founders feel it immediately: fewer intros, slower help, weaker leverage.

The shift: preferential attachment

Marc Andreessen calls it the core mechanic of startup success: when you’re winning, everything starts attaching to you—talent, customers, capital, media, momentum. VC firms can harness the same mechanism. The ones that build systems, not just relationships, start compounding.

What a “network‑effect VC” actually looks like

• Not just events or fancy platform decks.

• A firm with layered infrastructure that compounds with every new founder who joins.

What a16z got right (and others are now trying to copy)

• Talent pipelines continuously flowing into portfolio companies.

• Customers and executives moving between companies.

• Alumni founders becoming angels for new founders.

• Internal tools, research, and distribution that everyone can access.

• Vertical GPs + horizontal networks that multiply over time.

This invisible infrastructure turns a VC from a linear value‑add service into a self‑reinforcing network that amplifies returns for both the firm and its portfolio.

---

End of article.