Jersey Mike’s IPO illustrates how bad the AI hype has become

What the Jersey Mike’s IPO Reveals About the State of AI Hype

When a regional sandwich chain with Danny DeVito as its public face mentions the term “artificial intelligence” 22 times in its IPO documents, the AI hype cycle has officially reached peak saturation. Jersey Mike’s, a company whose primary product is submarine sandwiches made with real ingredients, filed an S-1 that invokes AI not as a core technology, but as a risk factor and a nod to investor appetite. As reported by TechCrunch, the company’s use of AI is described in a single hand-waving phrase: “We are beginning to use AI Technologies in our business.” That’s it. No specifics about inventory optimization, customer personalization, or supply chain forecasting. Just the label.

For developers and builders who actually ship AI products, this is both a validation and a warning. On one hand, the market’s hunger for AI is so strong that even a franchise-operated sandwich shop feels compelled to signal AI in its SEC filings. On the other hand, the fact that the risk of an AI disaster for a sandwich shop is about as likely as a franchise getting hit by lightning — a real event reported in Texas in 2021, given zero mentions in the S-1 — underscores how much of the current AI conversation is detached from practical reality. The hype is no longer about capability; it’s about signaling.

The TechCrunch piece points out that Jersey Mike’s mentions software 52 times and data 112 times — both legitimate concerns for a modern franchise operation. But the 22 mentions of AI, especially in a risk-warning context, are pure boilerplate. This mirrors a broader pattern: non-tech companies scrambling to sprinkle “AI dust” on their pitches, as the article puts it, to satisfy investor thirst. Even Bending Spoons, a company known for acquiring aging, non-AI tech assets, is positioning itself for the AI narrative. For developers, this means the signal-to-noise ratio in the market is deteriorating rapidly.

The Real Cost of AI Hype for Developers

When a sandwich shop feels the need to talk about AI, it directly affects the credibility of every developer who is genuinely building something novel. Investors and executives trained on this kind of hype may start expecting “AI” in every product roadmap, even when traditional deterministic algorithms or simple rule-based systems would be more reliable, cheaper, and easier to maintain. The pressure to label everything as AI can lead to overengineered solutions that are harder to debug, more expensive to run, and ultimately less trusted by end users.

The article’s reference to Starbucks’ failed AI inventory tool is instructive. That tool, which the source describes as “half-baked” and “couldn’t count,” was recently scrapped. It’s a concrete example of what happens when an organization adopts AI for the sake of signaling rather than solving a real problem. For developers, the lesson is clear: your AI tooling needs to be genuinely useful and rigorously tested, not just defensively marketed. If a company as large as Starbucks can fumble an AI inventory system, the risk for smaller teams is even greater — especially when the underlying data or operational processes aren’t ready for AI-driven decisions.

This is where practical tooling becomes essential. Developers building AI features should have a clear way to track costs, especially for API calls to large language models. Using a resource like our LLM API cost calculator can help you understand the real expense of each prediction, inference, or completion — and make better decisions about when to use a full generative model versus a simpler heuristic. The hype may push you toward the most expensive, flashiest approach; the data should pull you back toward the right one.

Practical Takeaways for AI Builders

Drawing from the Jersey Mike’s case and the broader pattern it illustrates, here are actionable insights for developers and technical leaders:

• Focus on measurable outcomes over buzzwords. When a sandwich shop can claim “AI” without specifying what it does, the term has lost meaning. Your product should be able to demonstrate specific improvements — reduced latency, higher accuracy, lower cost per transaction, or better user engagement — that hold up under scrutiny. Be prepared to answer: “What exactly does the AI do, and how do we measure its impact?”
• Understand your real failure modes. Jersey Mike’s AI risk warning was boilerplate, but some of those risks are real for tech companies that depend heavily on AI. Build robust monitoring, rollback procedures, and fallback logic. The Starbucks inventory tool couldn’t count; your recommendation engine or customer support bot might have equally basic failure modes if not properly tested against real-world edge cases.
• Treat data quality as a prerequisite, not an afterthought. The article notes that Jersey Mike’s S-1 mentions “data” 112 times, far more than AI. That ratio is telling. Any AI system is only as good as the data it’s trained on. Invest in data pipelines, validation, and governance before layering on models. If your data is messy or sparse, no amount of AI marketing will save the project.
• Make cost-aware decisions from day one. AI infrastructure is expensive, especially when using external APIs or running large models. Use a pricing reference like our LLM API pricing reference to compare providers and estimate ongoing costs. Factor in not just per-token pricing but also latency, retries, and fallback logic. The hype will encourage you to start with the most powerful model; practical engineering often requires a lighter touch.

How to Navigate the AI Hype Cycle

The Jersey Mike’s IPO is a symptom, not the disease. The disease is a market environment where the label “AI” carries more weight than the actual utility. For developers, the response should not be to abandon AI, but to build with discipline. That means being transparent about what your system does and does not do, benchmarking against non-AI baselines, and communicating honestly with stakeholders.

The hype cycle will eventually cool, and when it does, the teams that survive will be those that delivered real value — not just S-1 mentions. If you’re a developer evaluating AI tools or building your own, ask yourself: Does this solution solve a genuine problem that cannot be solved with a simpler, cheaper approach? If the answer is no, you’re better off avoiding the AI label entirely. If the answer is yes, then invest in the measurement, monitoring, and cost management that make the system sustainable.

In the meantime, take the Jersey Mike’s story as a reminder that AI has become a near-universal signifier in business communications. That doesn’t mean you should stop using AI; it means you should be more critical of AI claims from others and more honest about your own.

Source: TechCrunch. Details as reported; verify specifics at the source.