Thomas Laffont: The $4T AI IPO Wave Is Coming… and We’ve Never Seen Anything Like It
In a Nutshell
Thomas Laffont argues the AI-driven private market is entering an unprecedented $4T wave, with OpenAI, Anthropic, and others growing faster than any prior tech cohort and poised to surpass hyperscalers in revenue within years. Exits are thawing via massive upcoming IPOs, funding is concentrating among a handful of winners, and the ecosystem is shifting from cash-burning to cash-generating. Power-law dynamics mean missing the top companies carries rising costs, while every sector faces disruption as AI scales toward superintelligence.
These notes were generated by AI and may contain inaccuracies.
Thomas Laffont of Coatue, a hedge fund with $55 billion under management, discusses the firm's decision to enter the business of finding great entrepreneurs and companies. Coatue is looking to raise an additional $1 billion to invest in AI. The firm operates in an idea business where truly revolutionary ideas can scale significantly.
Markets have recovered, with the unicorn economy up 70% on average since September 2024. The share of the unicorn economy relative to the NASDAQ has plateaued over the past few years, reflecting strong performance of public companies such as Palo Alto. AI continues to dominate fundraising, with its share of total funding increasing for multiple consecutive years.
The unicorn factory peaked during the ZERP era of 2021 and has normalized to much lower pre-COVID levels. As a result, funding per unicorn has increased 5x since 2021, with fewer unicorns each raising more capital.
The pre-ZERP unicorn cohort of 73 companies showed that 80% had either raised a new round or exited within 20 quarters. The 2021 cohort of 479 companies showed less than 20% had exited or raised after 20 quarters. The 2024 AI cohort raises the question of which prior cohort it will most resemble.
AI funding is concentrated among a small number of companies, with the top 10 capturing a significant share. This is evidenced by massive rounds from Anthropic and OpenAI.
Laffont proposes a new index called the "Magnificent 8," which includes SpaceX, Stripe, Anthropic, Databricks, Revolut, ByteDance, and Anduril. The group spans internet, AI, fintech, and space tech. This index represents nearly $4 trillion in value and has outperformed the traditional Magnificent 7, with nearly every constituent outperforming that benchmark.
Exit activity is improving. 2025 is on a strong trend, though not yet at 2021 levels, with half the year remaining. Three major upcoming liquidity events—SpaceX (weeks away) and Anthropic (confidential S-1 filed)—will together exceed the total of the prior 10 years combined. The ecosystem, previously consuming far more cash than it returned, is now significantly more balanced even before these events.
OpenAI and Anthropic are growing at unprecedented rates. Starting from January 2025, these companies surpassed Workday, ServiceNow, Adobe, Salesforce, Google Cloud, and Azure within roughly 18 months. Projections suggest OpenAI could exceed AWS by year-end and potentially surpass all of Microsoft by 2028.
The largest technology companies have made unprecedented investments to enable and create the current AI transformation.
The primary driver of SpaceX valuation is launch cadence. Valuation per launch has increased over time. Laffont introduces the "CODE" framework to explain this: pre-constellation (testing rockets, government customers, one-time revenue), initial ramp (one constellation providing recurring revenue), scale (multiple constellations), and platform (new businesses such as space data centers, lunar and Mars applications).
Analysis of three buckets shows: unicorns have an 8% chance of becoming decacorns; decacorns have an 8-13% chance of reaching $100 billion; centacorns ($100B+) have a 31% chance of achieving a 10x return. Three public companies moved from $500 billion to $1 trillion market cap in the same year, with two achieving it in weeks.
Cerebras, where Laffont served as board member and led the Series B, experienced multiple years without new capital before securing a massive OpenAI contract that quintupled its value. Semiconductors have significantly outperformed the broader market since the 2024 All-In Summit.
AI systems become more useful with greater knowledge of user preferences. Memory per user could quintuple due to AI service demands, explaining recent performance in memory companies.
The AI ecosystem is estimated at $140 billion today, projected to reach $300 billion in 2025 and double again in 2027. Revenue sources include consumer subscriptions (number of subs × ARPU), advertising (currently ~25% of Meta and Google ads are AI-enabled, expected to reach 100%, contributing $150 billion), and enterprise/cloud/code applications.
Nearly every sector is being transformed: telco (Starlink enabling global phone calls, capturing the global broadband and wireless profit pool), semis (compute driving massive changes), energy (data centers altering the energy equation in Pennsylvania), automotive (Ferrari's electric/autonomous challenges), and consumer (GLPs impacting food, alcohol, diet, and wellness).
The new unicorn economy is healthier due to AI. Winners are compounding faster than ever, raising the cost of missing winners. Disruption affects every part of the global economy, and superintelligence has not yet arrived.
Power law effects are consolidating gains into a small number of companies. The private market ecosystem is evolving with companies staying private longer and achieving extraordinary outcomes. Questions arise about capital allocation strategy for LPs, angels, and venture firms given this new environment.
Big outcomes in private companies are positive for the ecosystem. Both Anthropic and OpenAI have stated intentions to go public. The market will serve as the ultimate test and equalizer for these companies. Public markets will apply scrutiny through short sellers, analysts, and regulators.
The timing of when public markets fully price these companies may extend beyond T+1 due to passive buying flows, potentially 6+ months post-IPO. The small sample size of trillion-dollar companies makes it difficult to determine whether their acceleration reflects structural inefficiency or survivor bias. Pre- and post-Cloud Code Anthropic represent fundamentally different companies.
The narrative that AI models are commodities has been largely disproven.
Creating detailed presentations re-anchors investment conviction through data and numbers rather than external noise. Companies reaching trillion-dollar valuations in weeks are real businesses with decades of history trading at low multiples relative to the S&P 500.
Memory companies lack an equivalent to TSMC's role in chips, raising questions about appropriate multiples for memory versus ASIC chips.
Policy and regulatory decisions, including actions by figures like Lina Khan, significantly affect outcomes in the technology sector.
The 31% probability of centacorns achieving 10x outcomes prompts consideration of whether trillion-dollar companies have greater than 30% odds of reaching $10 trillion. The filtering mechanism at each stage requires dominant businesses with compounding advantages and durable earnings.
To reach the trillion dollar club, companies need a dominant business. The key question becomes at what point saturation occurs. All of these markets have ended up being significantly bigger than anyone predicted.
Government intervention is another factor to consider. The breakup of the Bell system serves as a historical example—had it not occurred, the company could have established monopolies on the internet, commerce, and e-commerce.
A trading strategy involves having a bot that automatically starts buying shares once a company hits one trillion dollars in market cap. Apple was the first company to reach this milestone, and now there are approximately five companies that have achieved this valuation.
A study showed that buying the NASDAQ over a 10-year period yields approximately a 3x multiple. However, if an investor rebalanced every year to hold only the top 10 companies in the NASDAQ by market cap, they would outperform by approximately 3x over a decade.
When three or four trillion dollars gets distributed back to general partners, then to limited partners, and then recycled into the ecosystem, several dynamics may shift.
When David was bearish on California real estate, the question was whether this influx of capital would create opportunities. San Francisco homes are currently selling, with one speaker jokingly referencing a 40,000 square foot property described as a mausoleum, noting protesters are not included.
Regarding SpaceX, the speaker notes uncertainty about whether 1.75 is the right price for an IPO. The global profit pool of telco and service providers worldwide ranges between 200 to 400 billion dollars depending on the addressable market. A company that wasn't even in this business a couple of years ago is now addressing a profit pool of multiple hundreds of billions with a substantially better product.
Starlink works reliably without radio towers. This situation is unprecedented—we've never had anything like this before.
Looking at the ride sharing and food delivery wars, excess capital was eventually deployed into price competition. The question arises whether similar dynamics could occur between OpenAI and Anthropic. If these companies accumulate substantial capital, one might eventually pull a price lever to compete.
Rationally, they should engage in price competition. Companies with 200 billion in cash could theoretically use it strategically. However, these companies are spending heavily on infrastructure, making outcomes less obvious. Counterintuitive changes are likely to emerge that cannot be predicted today.
The speaker hopes to return in two years to analyze what went right and what went wrong.
Keep All-In Podcast in your library
Save the videos and channels worth coming back to, and find them again in one place.
