A post today on X made the argument that spending on a Super Bowl ad is an indication that a technology phase has passed its peak, and is in an era of excess and decline, using the .com cycle and crypto cycles as examples of how paying too much for ads during those busting boom cycles were indicators that in fact, those cycles were nearing their completion in retrospect. That assessment of those two cycles is accurate however it is inaccurate to say the same about what is happening now in AI.
As I’ve argued before, there is a massive spend happening on this technology and it’s promotion, but that does not mean that we are in an AI bubble. Super Bowl ads are the most expensive advertising that a company can buy, and it can at time seems like an irrational purchase. This Super Bowl is no exception; it was saturated with AI ads and there are companies that likely overspent on the attention that they bought. But directly connecting that to the deflation of previous cycles is erroneous pattern detection. Every hype cycle eventually produces its favorite shorthand: this looks like that. Dot-com ads, crypto ads, now AI ads – same stadium, same spectacle, same warning label. It’s an argument that feels reasonable because it borrows from history. But resemblance is not mechanism, and analogy is not causality. When we put distinct technological phases through the same narrative, we risk mistaking surface symmetry for structural sameness.
Super Bowl advertising has historically marked a peak in narrative confidence, not a verdict on technological validity. The dot-com boom did not fail because the internet was useless; it failed because valuation ran ahead of viable business models. Crypto’s collapse was not a repudiation of distributed systems; it was the implosion of financialized speculation detached from productive cash flows. AI’s current visibility spike similarly reflects mass awareness outrunning measured ROI, but this is a phase every general-purpose technology enters before institutions rewire processes to extract value.
This distinction matters because AI is not being sold primarily as a speculative asset, but as a labor-substituting and labor-augmenting capability embedded inside existing workflows. Productivity gains in such transitions rarely appear cleanly in CFO surveys during the installation phase. They surface later, unevenly, and often indirectly — through cycle-time compression, error reduction, and option creation rather than immediate margin expansion. Treating early measurement ambiguity as evidence of hollowness confuses accounting lag with technological absence.
The deeper pattern worth noticing is not that hype sectors advertise at cultural inflection points — that’s expected — but that analysts repeatedly over-index on narrative rhyme while under-weighting economic and operational mechanism. Sometimes a pattern really is a warning. Sometimes it’s just nostalgia wearing numbers. The harder work is telling the difference.
