The AI Stock Selloff Is Loud. Your Business Shouldn't Hear It.

Two things collided this week. First, a hotter-than-expected inflation print on October 17 briefly pushed markets toward pricing a Federal Reserve rate hike, with October hike odds spiking to around 58% intraday. By the next morning, those odds had fallen back to roughly 30% and Treasury yields had dropped to their lowest levels in nearly a month. That whiplash tells you the real fear isn't inflation anymore. It's growth. Investors have started asking whether AI spending, which has propped up corporate earnings for years, might slow down. Second, the revenue picture at the top of the AI world showed cracks. Anthropic disclosed $16.8 billion in third-quarter revenue. That's a huge number, but it implies growth is decelerating from the prior quarter, and markets treated the slowdown as a sell signal. Add in the FTC's subpoenas of Anthropic and OpenAI over safety and privacy practices, and you get a sector under pressure from three directions at once: money, regulation, and narrative. The damage spread through the hardware chain. Chipmakers from Intel to Micron fell 2 to 4% in premarket trading, and the semiconductor index has now dropped six sessions in a row. Gold rallied 1.3% as money ran for cover. This is what a genuine risk-off day looks like, and it started in Asia before US markets even opened.

It helps to separate two very different businesses that both get labeled "AI." The first is AI infrastructure. Companies in this group spend billions building data centers, buying GPUs, and training frontier models. Their economics depend on enormous capital outlays today in exchange for revenue that may or may not arrive at the scale projected. CoreWeave's distressed bonds are the clearest signal that some of that financing is under strain. Gartner's warning about trillions in potential impairments is the same concern stated at industry scale. When growth decelerates even slightly, the math on a $50 billion data center gets uncomfortable fast. The second is applied AI services. This is software that uses AI models, most of which are rented cheaply, to do specific jobs: answer a customer's question, draft a quote from a phone call, write a service page for a website, chase an unpaid invoice. There's no data center to finance. There's no frontier model to train. The capital intensity is close to zero, the revenue is a predictable monthly subscription, and the customer is a local business that judges the tool by one standard: did it get me more work? The selloff is happening almost entirely in the first category. The second category benefits from it, because the underlying models that power these tools keep improving while their costs keep falling. Competition among AI labs, the very thing scaring the stock market, is what makes your tools better and cheaper every month.

There's one detail in this week's data worth your attention even if you ignore the stock market entirely. IDC reported the worst quarterly smartphone shipment decline on record, down 22%, and attributed it largely to rising memory costs. Those costs are rising because AI data centers are buying up memory supply, crowding out consumer electronics. That's a small example of a bigger dynamic: the AI boom raises prices for the components and software around it. For a small business, that shows up as subscription creep. The scheduling app adds a fee. The review tool gates e-signature behind a higher plan. The field-service platform charges per seat and sells the features you actually need as add-ons. When every vendor in your stack is under its own financial pressure, those price increases arrive faster. This is why consolidation arguments tend to get stronger in uncertain markets, not weaker. If one managed platform replaces your website builder, CRM, chatbot, content tool, scheduler, review manager, and invoicing system, then a price hike from any single vendor stops being a crisis. You have one bill, one system, and one team accountable for whether it all works.

The Federal Trade Commission has subpoenaed Anthropic and OpenAI over safety incidents and privacy policies. That's a serious development for the companies building frontier models, and it could shape how those models are deployed and monetized in the years ahead. It is not a development that reaches your storefront. Businesses using AI-powered tools for marketing, scheduling, and customer service are not the subject of these inquiries. The regulatory exposure sits with the model builders, and a done-for-you platform absorbs that complexity on your behalf. You don't need to track which model powers your chatbot any more than you need to know which email server delivers your invoices. You need the chatbot to capture the lead and the invoice to get paid.

So what should a small business owner actually do while the headlines stay ugly? Three things. First, ignore the stock market as a signal about your tools. CoreWeave's bond prices tell you nothing about whether an AI receptionist can answer a missed call at 7pm on a Friday. Judge your tools by results: leads captured, quotes sent, invoices paid, reviews collected. Second, audit your subscription stack. List every software bill on your credit card, note what each one costs, and ask which ones overlap or go unused. Most service businesses find six to ten tools doing work a single managed system could handle. In a tightening economy, that audit is worth real money. Third, if you've been putting off AI because the headlines made it feel risky or complicated, reconsider the framing. The risk in AI right now sits with investors who bet on infrastructure. The opportunity sits with businesses that let someone else handle the technology entirely. You don't learn the software. You don't manage the models. A system is built for you, it runs, and you talk to it the way you'd talk to an employee.