AI's teaser period is ending: record highs mask a reset wall

The Nasdaq Composite hit an all-time high on September 23, 2026, extending its year-to-date gains past 17% as investors shrugged off AI safety concerns, geopolitical tensions, and sky-high valuations (Al Jazeera). Nvidia closed up 0.7%, Micron jumped 5%, and Monolithic Power Systems surged 8.1%. Meta dropped 0.63% after its Monday spike on the new Muse assistant. The market is betting that AI's growth curve remains steep enough to justify current prices.

But a week later, a different kind of signal emerged. On September 28, the Dutch consumer authority ACM demanded more transparency on how AI sets prices, after finding that some airline seats had 35 different prices and one-third of tickets cost 40% more than the average (DutchNews.nl). ACM chairman Martijn Snoep noted the same techniques are spreading to online retail, concert tickets, and even petrol stations. Two people looking at the same product at the same time can be charged different prices, and consumers have no way to see how those prices are derived.

This regulatory intervention is exactly the kind of concrete friction that the AI trade has been ignoring. The Nasdaq's record high is built on the assumption that AI costs will keep falling and adoption will keep compounding. But there is a growing body of analysis suggesting the boom is still in what we might call the teaser period — and a reset wall is visible on the horizon.

In August, an essay titled "The Teaser Period: Why the AI Boom Is Hitting a Reset Wall" drew a direct parallel between current AI economics and the subprime mortgage crisis (Hacker News). The argument is structural: the subprime boom ran on 2/28 adjustable-rate mortgages — two years of low teaser payments, then a reset to payments 30-50% higher. That reset was contractually inevitable from origination. When the industry aggregated those schedules, it produced the "reset wall": roughly a trillion dollars of mortgages automatically repricing across 2007-2008.

The essay argues the AI boom has its own equivalent. Cloud credits, subsidized inference pricing, free tiers, and generous API allowances are the teaser rates. They make AI adoption look frictionless and unit economics look attractive. But those subsidies are not permanent — they are marketing costs that vendors will eventually roll back, or that will be exposed when capital gets more expensive. The question is not whether the repricing happens, but when the wall hits.

For engineers and founders, the practical implication is that the current economics of AI are not a stable equilibrium. The ACM's finding on airline pricing is a leading indicator: when regulators start scrutinizing the opacity of AI-driven pricing, costs of compliance get added to the stack. The same technology that sets dynamic prices on flights is already embedded in everything from ad bidding to cloud auto-scaling. If the push for transparency spreads beyond airlines — and the ACM chairman explicitly said it should — it will hit margins that the market has not priced in.

The Nasdaq reset wall may not look like a mortgage chart. It could be an earnings season where a major vendor admits that gross margins on AI services are thinner than guided. It could be a regulatory fine or a forced overhaul of a pricing algorithm. Or it could simply be the point where the cost of capital rises and the free credits dry up.

Investors shrugged off the Iran war and AI safety letters in September. That is precisely what subprime investors did in 2006. The difference is that the reset schedule this time is not printed on a contract — it is buried in server costs, API rate cards, and adversarial debates about algorithmic transparency. The wall is still visible if you aggregate the right data. The question is whether the market will look at it before the prices reset, not after.