The AI Stock Selloff: What a Nasdaq Correction Means for Your Business

If you checked the markets this week, you probably saw a lot of red. On Wednesday, 5 November, the Nasdaq fell 3.9% and slid into correction territory, more than 10% below the record it set just a month earlier. The selloff wiped trillions of dollars off US stocks in a couple of sessions, and almost all of the damage was concentrated in one place: companies connected to AI.

Nvidia dropped around 8% and briefly gave back its entire gain for 2026. AMD and Micron lost roughly 15% each in a single day. CoreWeave, Oracle, Broadcom, SoftBank and a long list of others followed. The main semiconductor ETF had its worst day since March 2020, and the pain spread to Asia, where SK Hynix and Advantest fell hard overnight.

So what happened, and more importantly, what does any of this mean if you are not a trader but a business owner who uses AI, pays for AI tools, or is halfway through building an AI-powered product? That is what this post is about. Let's sink in!

What actually happened on Wall Street

The numbers first. On 5 November the S&P 500 fell 2.8% and the Dow lost over 1,100 points, but the Nasdaq took the biggest hit at 3.9%, which pushed it into an official correction. Reports put the value wiped off US stocks in the trillions, with the so-called Magnificent Seven alone losing around a trillion dollars of market value in a day.

The selling was clearly an AI story rather than a general one. Memory makers like Micron fell hardest, reversing the huge rally they had enjoyed on the back of the AI memory shortage. Data centre operators and the companies that finance them with debt, such as CoreWeave and Nebius, dropped 14 to 18%. Even companies that reported strong results were punished: Viavi posted 57% revenue growth and still fell 18% the next morning.

Selling continued into Thursday and Friday. At one point Nvidia was trading below the level it started the year at, which tells you how much 2026 optimism had been priced into these shares. A few names recovered part of their losses later in the week, but the overall mood changed from greed to caution very quickly.

Why the midterm elections spooked investors

The trigger, oddly enough, was politics rather than technology. The US midterm elections handed Democrats control of both chambers of Congress, and several new governors campaigned on slowing down data centre construction. The incoming governor of Texas had called for a moratorium on new data centres, and Texas matters a lot here: by one Wells Fargo estimate the state accounts for about 27% of planned US data centre power.

Investors did the maths fast. UBS analysts estimated that if construction pauses spread, nearly a third of the new US data centre capacity expected in 2028 could be delayed or lost. J.P. Morgan's Michael Cembalest was calmer, pointing out that 2027 capacity is largely safe because 75% of it is already under construction, but he agreed the risk sits in 2028 and beyond.

There is a second layer underneath the politics. For over a year, analysts have asked whether the enormous sums spent on AI infrastructure will ever come back as profit. Microsoft and Alphabet have been trimming costs elsewhere while raising capital spending, and much of the newer data centre construction is financed with debt. When a political risk appeared on top of those doubts, holders of the most stretched stocks decided not to wait around and find out.

Is the AI bubble actually bursting?

This is the question every headline asked, so let's be careful with it. A correction in share prices is not the same thing as a collapse in demand. Nothing about this week's selloff says that companies suddenly stopped using AI tools, cancelled their API contracts or lost interest in automation. Chipmakers' own forecasts still point to growth: Nvidia guided to roughly $108 billion for its next quarter, and AMD, Micron and Broadcom all raised or confirmed strong outlooks shortly before the crash.

What changed is the price investors are willing to pay for future AI profits, and the discount they apply for risks like construction pauses, power constraints and debt. That is a repricing, not a verdict on the technology. We saw a similar pattern with the internet: the dot-com crash of 2000 wiped out share prices, yet internet usage and e-commerce kept growing right through it.

At the same time, it would be naive to pretend nothing is wrong. Some AI infrastructure companies are carrying a lot of debt against revenue that has not arrived yet, and a long market downturn would make that debt more expensive. If funding gets tighter, weaker AI startups will struggle, some products will be discontinued, and consolidation will speed up. Markets sometimes overreact, but they also sometimes notice real problems early.

What this could mean for AI prices and capacity

Here is the part that actually touches your business. If data centre construction slows in 2028 while demand keeps growing, computing capacity becomes scarcer, and scarce capacity usually means higher prices or longer waits. Businesses that rely on heavy AI workloads, such as video generation, large-scale document processing or model fine-tuning, could feel that first.

On the other hand, a cooler market has benefits for buyers. Over the past two years AI vendors have priced aggressively for growth, subsidised by cheap capital. If capital stops being cheap, expect clearer pricing, longer-term contracts being pushed harder, and fewer unlimited free tiers. None of that is bad for a business that plans properly; it mostly punishes those who built products on the assumption that AI compute would stay almost free forever.

Tighter conditions may also favour the biggest providers. The large cloud platforms already own data centres, generate cash and can fund construction from their own pockets, while newer debt-funded rivals cannot. Practically, that means the AI tools your business depends on are safer when they run on major infrastructure, and riskier when they depend on a single young provider burning borrowed money.

Practical takeaways for businesses building digital products

First, do not change your product plans because of one bad week on the Nasdaq. Stock prices moved; the models, the APIs and your customers' expectations did not. If an AI feature made commercial sense for your website or app two weeks ago, it still does. Decisions about architecture should follow usage and unit economics, not market headlines.

Second, check your dependencies. Make a short list of the AI services your product or your internal processes rely on, and ask a simple question about each one: what happens to us if this provider raises prices by 50% or shuts down with 30 days' notice? Keeping a second inference provider configured, even if you never switch, turns a potential crisis into a configuration change. The same goes for export options on any platform that stores your data.

Third, treat the next few months as a buyer's window. When markets wobble, vendors become more flexible: discounts appear, contract terms soften, and sales teams return calls faster. If you were planning to lock in an annual AI or cloud contract, a nervous market is a decent time to negotiate.

And finally, watch costs that are hiding inside "free". If capacity tightens in 2027 or 2028, usage-based prices are more likely to rise than fall. Build a basic view of your cost per user or per transaction for AI features now, so a price change is something you can model in an afternoon rather than discover in an invoice.

Final notes

The November selloff is a reminder that the AI boom runs on physical things: chips, buildings, electricity and borrowed money. Markets repriced the risk around those things this week, and they did it with their usual lack of subtlety. Whether this turns into a longer slide or gets bought back within a month, nobody honestly knows.

For businesses, the sensible reading is neither panic nor denial. AI demand is still there, the tools keep improving, and your competitors are not uninstalling anything. But the era of assuming infinite cheap compute and immortal AI startups is ending, and a little supplier discipline now will save you real money and stress later. If you want help reviewing how exposed your product or website is to any single AI provider, that is exactly the kind of work we do at Davydov Consulting, and we are happy to talk it through.