Scott Galloway: 95% of enterprise…

1. Key Themes

Theme 1: Enterprise AI Spend Has a Catastrophic ROI Problem

The core argument is that enterprise AI spending is almost entirely disconnected from measurable returns — and CFOs are beginning to notice.

"About 5% of the projects that people are using tokens for, CFOs can connect to some sort of return. They're really intoxicated. Using AI as much as you can and talking about it in your earnings calls, it's like adding .com back in the 90s."

Real-world evidence is already accumulating: Uber exhausted its entire 2026 AI budget before Q1 ended. Microsoft canceled cloud code licenses across multiple divisions. ServiceNow burned through its entire Anthropic budget. Shopify reported earnings partially offset by increased LLM costs.

"The technology is actually more expensive than the humans it is supposed to replace."

Theme 2: AI Valuation Correction Is Inevitable — The Only Question Is How Fast

Galloway offers a precise, math-grounded prediction: a 50–70% AI valuation correction within 24 months, framed as the less painful of two possible outcomes.

"One of two things is going to happen. Either the valuations of AI are going to come down by 50% or 70%, or we're going to have labor chaos in these industries. I absolutely think it's the former."

The arithmetic: 155 million working Americans, ~75 million AI-vulnerable jobs, $100K average labor cost = a $5 trillion pool of potential savings. To justify current AI valuations through labor replacement alone requires 5–7 million layoffs in 2–3 years — roughly 10% labor destruction. That hasn't happened. The trigger event will be the first credible Fortune 500 CEO announcing an AI pullback on an earnings call, followed by NVIDIA's first earnings miss, which then takes the broader market down 5–10%.

"Being right about the technology and wrong about the timeline produces the same financial result as being wrong entirely."

Theme 3: Chinese AI Is a Structural Threat to American AI Valuations

Chinese LLMs are 10–30x cheaper than American counterparts — and 80% of American AI startups have already migrated to them.

"Every developer in the world has heard of them because 80% of American AI startups are now using Chinese models."

The cost advantage is structural: direct government subsidies plus industrialized "distillation" — harvesting outputs from American frontier models to train their own. Galloway's prediction: mass Chinese LLM adoption surfaces in earnings calls → American AI companies lobby for protection → Trump administration bans Chinese LLMs within 90 days, using the BYD vehicle ban as the template.

Theme 4: The IPO Window for AI Giants Is an Insider Exit Mechanism

SpaceX, OpenAI, and Anthropic are expected to IPO at a combined $4 trillion valuation — none of them profitable — and NASDAQ rule changes accelerate passive fund inclusion to just 15 days post-listing.

"When these companies go public, it's effectively the smartest people in the room saying: we've squeezed as much juice out of this as we can. We need to find people stupider than us to invest at this valuation."

The risk is compounded by the NASDAQ rule change: previously, companies needed 12 months of public trading before joining passive index funds. Now it's 15 days — meaning retail investors are forced into these positions before a single public quarter of results is reported.

Theme 5: The China Playbook on AI Mirrors What It Did to German Industry

China deployed the same pricing-and-dependency strategy against German manufacturing over 30 years. AI is the next target.

"What China has done to Europe economically, they're going to try to do to the AI market what they tried to do to the steel market here in the 80s and 90s."

The mechanism: subsidize models to below-market pricing → attract developers with cost savings → harvest outputs via distillation → create economic dependency → use dependent American companies as lobbying assets against any proposed ban. Volkswagen, Daimler, and Siemens became so dependent on Chinese arbitrage they lobbied against IP theft restrictions.

2. Contrarian Perspectives

The AI Apocalypse Won't Arrive — Valuations Correct Before Labor Markets Collapse

Consensus says AI will destroy 75 million jobs rapidly. Galloway flips this: the disruption required to justify current valuations would cause visible political chaos and regulatory response. The quieter, likelier outcome is valuation compression.

"The AI fear machine runs on catastrophizing as a business model. Fear is the product. Capital is the outcome."

Supporting evidence: AI eliminated nearly 50,000 jobs in 2025 — far below the 5–7 million annually required to justify AI valuations. The layoffs arrived on schedule. The savings did not.

Token Leaderboards Are Destroying Enterprise AI ROI From the Inside

While companies race to prove AI adoption, they've built internal incentive structures that reward consumption rather than outcomes — actively accelerating the cost problem.

"They'll create leaderboards at these companies, like Meta, like Amazon, where they will track how many AI tokens you're using. The people who are using the most tokens are the most AI-deployed. Those are the ones who get recognized. Maybe they'll get a promotion."

One Anthropic employee ran up a Claude Code bill of $150,000 in a single month. Stripe's technical staff spends $100,000 on AI tokens every day. Salesforce is on track for $300 million in Anthropic spend this year. These numbers aren't the result of rogue behavior — they're the rational response to the incentives companies created.

Relationships Are the Last Non-Commoditizable Career Asset — Not AI Fluency

The consensus career advice is to master AI tools. Galloway argues AI fluency is already commoditizing and that human relationships compound in a way AI literally cannot replicate by design.

"The only competence that's really important is storytelling and relationships."

Supporting evidence: Of the top 10 candidates brought in for roles at Google, 70% of the time the person hired had an internal advocate — not the strongest resume. AI drives output toward the median by architectural necessity: it averages previous words to predict the next one. It cannot manufacture the advocate in the room when your name comes up.

3. Companies Identified

Uber Ride-sharing giant. Mentioned as a leading indicator of enterprise AI cost failure — exhausted its entire 2026 AI budget before Q1 ended. COO acknowledged the costs are getting hard to justify.

"Uber exhausted its entire 2026 AI budget before the first quarter ended."

Stripe Payments infrastructure company. Cited as an extreme example of uncontrolled AI token spend.

"Stripe's technical staff spends nearly $100,000 on AI tokens every day."

Salesforce Enterprise software company. On track for $300 million in Anthropic spend this year and exhausted its Anthropic budget. Also cited as a company where CFOs can't connect spend to return.

"Salesforce is on track for $300 million in Anthropic spend this year."

Microsoft Technology conglomerate. Canceled cloud code licenses across multiple divisions — a concrete sign of AI spend retrenchment.

"Microsoft canceled cloud code licenses across multiple divisions."

ServiceNow Enterprise IT platform. Exhausted its entire Anthropic budget — another data point in the cost-without-return pattern.

"ServiceNow exhausted its entire Anthropic budget."

Shopify E-commerce platform. Reported earnings partially offset by increased LLM costs, publicly surfacing the cost problem.

"Shopify reported earnings partially offset by increased LLM costs."

NVIDIA Semiconductor company. Central to the repricing thesis — has beaten earnings estimates 15 consecutive quarters. Galloway identifies the first NVIDIA earnings miss as a market-wide trigger event.

"NVIDIA has beaten estimates 15 consecutive quarters. The streak is the pressure."

SpaceX Rocket and satellite company. Described as one of the greatest businesses built in 50 years — but the IPO structure bundles it with xAI, a division losing $2.5B per quarter.

"If you want to hang out with Snow White, SpaceX, you have to also invest in this money furnace called xAI." SpaceX stats: 90% global launch capacity, two-thirds of all low-Earth orbit satellites, $16B revenue, $8B operating profit, 30% annual growth.

xAI Elon Musk's AI company. Identified as a $2 trillion valuation at 107x sales, growing revenue at 15% annually, losing $20 billion per year. The "furnace" bundled with SpaceX.

"At $2 trillion, at 107 times sales, growing revenue at 15% annually, losing $20 billion per year, the numbers collapse."

OpenAI AI research company. One of three companies expected to IPO at a combined $4 trillion valuation — none profitable. Identified as an insider exit vehicle.

"When these companies go public, it's effectively the smartest people in the room saying: we've squeezed as much juice out of this as we can."

Anthropic AI safety and research company. Closing in on a $1 trillion valuation. Referenced both as a major beneficiary of current enterprise AI spend and as one of the three companies in the IPO exit thesis.

"Anthropic just passed OpenAI in revenue spending 4x less."

DeepSeek Chinese AI company. Leading example of the 10–30x cost advantage Chinese models hold over American counterparts.

"DeepSeek performs comparably at 10-30x lower cost."

Meta Social media and technology company. Cited as a company that built AI token leaderboards to reward consumption — rewarding spend over results.

"They'll create leaderboards at these companies, like Meta, like Amazon..."

Opal Identity governance company (sponsor). Mentioned in the context that 80% of enterprise systems are exposed through stale access — a risk compounded by AI agents layering on top of existing permissions.

"80% of systems are exposed through stale access. Access that should have been removed and was not."

4. People Identified

Scott Galloway NYU Stern professor, entrepreneur, and author. The central voice of the article — presenting his thesis that AI valuations will correct 50–70% within 24 months, drawing direct parallels to his dot-com crash prediction.

"The man who called the dot-com crash at a faculty meeting while his colleagues were still buying Pets.com is making the same argument about AI valuations today."

Xi Jinping President of China. Identified as the key actor who could strategically accelerate cheap Chinese AI adoption in the U.S. to destabilize American AI valuations and the market's primary support mechanism.

"AI is the only thing that feels like it's propping up the economy right now. The Trump administration has too much to lose."

5. Operating Insights

1. Map Every Dollar of AI Spend to a Named Outcome Before Your CFO Does It For You

The CFO audit is coming regardless. The founders and operators who survive it are the ones who run it on themselves first. The article is explicit:

"Map every dollar of AI spend to a specific outcome a CFO can name. Produce one number this week that a board member can connect to a result."

The current baseline is damning: 95% of enterprise AI projects connect to no return a CFO can name. Getting into the 5% is a defensive necessity, not a competitive advantage.

2. Reward AI Outcomes, Not AI Activity — Before the Leaderboard Destroys Your Budget

Companies that built token consumption leaderboards created the exact incentive failure that accelerated their cost crisis. The article diagnoses the mechanical failure clearly:

"Token leaderboards are a company car with no fuel budget. Hand someone a tool, measure them on how much they use it, and the bill becomes the metric."

The four failure modes: a KPI measuring activity instead of output; a leaderboard rewarding cost generation instead of value creation; a promotion system tied to spend instead of return; and a budget with no anchor to any identifiable result. Fix the incentive or the spend will find its ceiling on its own — painfully.

3. Build Relationships Before You Need Them — They Compound and AI Cannot Replicate Them

While competitors invest in AI fluency, the compounding asset is human. Galloway's four-part framework for anyone building a career or network:

"Be as social as possible. Out of the house, not on a screen. Relationships compound like money."

Supported by the hiring data: 70% of the time, the person hired at Google had an internal advocate, not the strongest resume. AI cannot produce the internal advocate. The person who gets the job had a friend in the room before the job existed. The median is crowded. The internal advocate is not.

6. Overlooked Insights

1. NASDAQ Rule Change Structurally Accelerates Retail Exposure to Unproven AI Companies

The NASDAQ passive fund inclusion rule change — from 12 months of public trading to 15 days — is mentioned briefly but has significant structural consequences. SpaceX, OpenAI, and Anthropic will represent approximately 6% of global public equity markets on listing day before reporting a single public quarter. This isn't incidental market mechanics; it's a policy change that mechanically routes pension funds, index investors, and retail portfolios into these positions at peak valuation before any accountability data exists.

"NASDAQ changed its rules specifically for these companies... These three will represent approximately 6% of global public equity markets on listing day, before reporting a single public quarter."

2. China's Strategy Depends on Making American Companies Its Lobbyists — Not Just Its Customers

The most strategically significant element of the Chinese AI playbook is step five: once American startups are economically dependent on cheap Chinese LLMs, they become lobbying assets against any proposed ban. This mirrors exactly what happened with Volkswagen, Daimler, and Siemens in Germany, who lobbied against IP theft and product dumping restrictions because their own profitability depended on the Chinese arbitrage. The 80% of American startups already using Chinese models are being positioned — perhaps unknowingly — as future political assets for Beijing.

"Use dependent American companies as lobbying assets against any proposed ban... Industrial strategy disguised as market competition."