Public Investors Missed the AI Growth, Baillie Gifford’s Singlehurst Says

SpaceX, Anthropic, and OpenAI are cited as high-valuation private companies expected to go public in the coming months, but Baillie Gifford's Peter Singlehurst says public market investors have already missed much of the growth. The piece is primarily commentary on private-market value creation and IPO timing rather than a new corporate event or earnings update.

Analysis

The real opportunity is not the IPO itself but the rerating of the entire private-markets ecosystem. If marquee AI and frontier-tech names come public into receptive conditions, the beneficiaries are the late-stage holders, secondary desks, and crossover funds that can crystallize gains and recycle capital into the next cohort; the losers are public-market incumbents that have enjoyed a de facto scarcity premium on “AI exposure” and may see that premium compress once cleaner, larger, more liquid vehicles arrive.

Second-order effects matter more than the headline. A strong IPO window raises the cost of capital for private competitors that are still dependent on venture funding and deferred revenue optimism, while also forcing public comparables to justify valuations against names with superior growth, brand, and optionality. If the listing process is disorderly, though, it could expose revenue concentration, compute dependence, and governance complexity, which would hit adjacent software and semiconductor suppliers first through multiple compression rather than fundamental demand weakness.

The key timing variable is months, not days: the setup only becomes tradable when the market has a few visible filings and price ranges to anchor expectations. The main downside catalyst is a rate back-up or a failed bookbuild that resets the market’s willingness to pay for long-duration growth; in that case, the “missed growth” narrative can flip into a liquidity-overhang story, with private holders rushing to de-risk into weaker public demand.

Contrarian takeaway: consensus is likely underestimating how much capital formation this could unlock for the rest of 2025. A successful cohort of AI IPOs would not just monetize winners; it could re-energize venture markups, secondaries, and employee retention at private firms, creating a self-reinforcing cycle that delays the public market’s catch-up even as it broadens the set of investable names.

Key Decisions for Investors

• Build a basket long of high-quality public AI enablers into the IPO window (e.g., MSFT, NVDA, AMZN) and hedge with a small short in expensive, revenue-light software names; thesis: cleaner public comps can compress premium multiples by 10-20% over 3-6 months if IPOs price well.
• For crossover exposure, buy late-stage private secondaries or related liquidity vehicles only after filings commence, not before; target a 6-12 month horizon where price discovery and allocation pressure improve entry points by 5-15% vs pre-IPO marks.
• Pair trade: long the strongest profitable AI infrastructure names / short the highest-multiple private-to-public converters in public markets; use a 2-3 month catalyst window around the first major pricing to capture multiple dispersion.
• If an AI IPO comes with weak order book signaling or a large discount, buy short-dated downside protection on broader growth indices (QQQ or IWM) for 4-8 weeks; failed books tend to spill into adjacent high-duration sectors faster than fundamentals deteriorate.
• Avoid chasing the first print; wait for the second-day trading range to establish whether public demand is real or just scarcity-driven. The risk/reward improves materially once lockup and secondary supply become visible.