Even Sandwich Shops Are Riding the AI Wave? Jersey Mike's IPO Filing Reveals Just How Absurd the Bubble Has Become

A Sandwich Shop's IPO Filing Made Me Laugh — Then Made Me Uneasy

I'll admit it: before opening Jersey Mike's IPO prospectus, I had zero expectations of finding anything surprising inside. A submarine sandwich chain founded in 1956, built on some combination of marinades, cold cuts, and bread — what could it possibly have to do with AI?

I was less than ten pages in when I spotted "artificial intelligence," "machine learning," and "AI-driven operations" scattered throughout the risk factors, competitive advantages, and growth strategy sections like sesame seeds on a hoagie.

I screenshotted the passage and sent it to a friend who works in MLOps. His reply: "Hopeless."

---

AI Keywords Have Become the Standard Seasoning in IPO Filings

This isn't news in itself — to be precise, it was already common knowledge by 2025. According to tracking by financial data platform Sentieo, the proportion of U.S. IPO filings mentioning "AI" or "artificial intelligence" nearly tripled between 2022 and 2024. By the first half of 2026, that number was still climbing.

The problem isn't the word "AI" itself. The problem is the context in which it's deployed — and what it's being used to obscure.

In Jersey Mike's filing, AI is described as a potential "order forecasting and supply chain optimization tool." That sounds reasonable enough; plenty of restaurant chains are genuinely doing work in that space. But read a little further, and you'll find they haven't actually deployed any such systems at scale. What they've done is take a declarative statement — essentially, "we may explore AI opportunities in the future" — and package it as a competitive advantage in the prospectus.

It's a bit like saying: "We have a kitchen, and kitchens may one day use fire. Fire is a powerful tool."

---

This Isn't Jersey Mike's Problem — It's the Entire Capital Market's Problem

I have no particular interest in singling out this sandwich chain. They're simply doing what everyone else is doing. Investment banks, underwriters, legal advisors — the entire IPO industrial complex is quietly signaling the same thing: if your filing doesn't mention AI, you'll pay for it in your valuation.

This is a textbook bubble lexicon. In 2000 it was ".com." In 2007 it was "repackaged subprime structured products." In 2021 it was "the metaverse." Every era has its magic phrase — the one that makes an otherwise unremarkable business model suddenly glow in investors' eyes.

The difference this time is that there's genuine technology backing up the magic phrase. What OpenAI, Anthropic, and Google DeepMind are doing is real. The capabilities of GPT-4o, Claude 4, and Gemini 2.5 are real. That's precisely what makes this bubble more dangerous: because you can't flatly declare "AI is fake," companies that are merely riding the name have cover to muddy the waters beneath the halo of legitimate technology.

---

Three Quick Signals for Spotting "AI Washing"

Whenever I see a company talking up AI in an IPO filing, earnings report, or fundraising deck, my first instinct isn't excitement — it's to run three checks:

First: Is AI core architecture, or is it a retrofitted narrative? If the entire business model continues to function normally with AI removed, then AI is decoration. Jersey Mike's sandwiches sell perfectly well without AI. That's a fact, and that's the problem.

Second: Is there concrete evidence of technical investment and talent? Companies genuinely pursuing AI transformation tend to mention things like engineering hires, model training costs, and data infrastructure buildout in their filings — because these things cost money. If the entire document amounts to "we believe AI will drive efficiency gains," they're selling a dream.

Third: Where does AI appear — in risk factors or in competitive advantages? Companies that are actually using AI typically acknowledge the downside in their risk disclosures too: dependency on specific model providers, data privacy regulatory exposure, that sort of thing. If a filing has nothing but praise for AI and never a critical word, you can be fairly confident the people writing it don't really understand what they're talking about.

---

A Bubble Isn't the Same as a Crash — But It Makes Real Signals Harder to Read

I'm not predicting an imminent AI bubble collapse. In fact, I believe investment at the AI infrastructure layer remains substantially justified: Anthropic's expansion into scientific workflow platforms and NVIDIA's compute scaling plans for 2026 are not built on air.

But the noise at the application layer is making it increasingly difficult to hear the signals that actually matter. When Jersey Mike's and a manufacturer genuinely using AI to restructure its supply chain speak the same language to package themselves, how is an ordinary investor supposed to tell the difference?

That's the real problem. Not the sandwich shop appropriating AI, but the fact that the entire market tolerates it — and even incentivizes it.

I've never eaten a Jersey Mike's sandwich; by all accounts they're pretty good. But their IPO filing made one thing unmistakably clear: the AI bubble of 2026 is not some hidden, hard-to-detect systemic risk. It's written right there in the prospectus, brazenly, waiting for anyone who bothers to look.