Everyone's Watching Iran. Few Are Watching Beyond.

The stock markets are down. Oil prices are spiking as the Iran war enters its second month, Trump is floating a ground operation, and Morgan Stanley just downgraded global equities to hide in cash .

The Disconnect

The war is a genuine geopolitical crisis and a legitimate reason for risk repricing. But it's also acting as a thick fog over one of the largest technology shifts in a generation. And the SaaSpocalypse narrative — the idea that AI will destroy software companies and crater their stock prices — is compounding the confusion by lumping every tech stock into the same bear case.

I wrote about that two weeks ago. The earnings data showed roughly two-thirds of the tech companies I track are actually accelerating, not dying. Since then, the disconnect has gotten wider.

What the Numbers Say When Nobody's Listening

I follow a broad cross-section of US public tech companies through their quarterly filings. Not stock prices — the actual numbers. Revenue growth, bookings, margins, what management said versus what happened.

Here's what the latest filings show across the AI value chain, from silicon to software:

The infrastructure buildout isn't slowing. It's accelerating.

NVIDIA's Data Center revenue grew 75% year-over-year last quarter, adding $10 billion sequentially — the largest quarter-over-quarter jump in the company's history. Broadcom's AI semiconductor revenue hit 140% growth, with a $73 billion committed backlog. Micron is running a memory super-cycle with demand exceeding supply by 35–50%, and no meaningful new fab capacity until late 2027.

The hyperscalers — Amazon, Microsoft, Google, Meta — are collectively deploying roughly $600 billion in capital expenditure this year. That's war-footing spending. AWS reaccelerated to its fastest growth in thirteen quarters. Google Cloud grew 48% while expanding operating margins from 11% to 30%. These are not companies hedging their bets. These are companies that see something worth spending $600 billion on.

Enterprise software isn't being destroyed by AI. The best companies are being turbocharged by it.

Palantir grew revenue 70% year-over-year — a rate almost unheard of for enterprise software at scale — with a Rule of 40 score of 127%. CrowdStrike is accelerating revenue growth at $5 billion in annual revenue, which shouldn't be mathematically possible at that scale but is happening anyway. Datadog's growth reaccelerated to 29%, and here's the part that breaks the SaaSpocalypse thesis: even their non-AI customer growth inflected from 18% to 23%. Cloudflare grew 34%, with a widening gap between remaining performance obligations (48% growth) and recognized revenue — a classic leading indicator of further acceleration.

These aren't cherry-picked outliers. They're consistent with the broader pattern I documented in the SaaSpocalypse analysis: the companies integrating AI into their platforms are pulling away. The ones that aren't are the ones struggling. This is bifurcation, not apocalypse.

The Bear Case Gets the Timing Wrong

The most common argument I hear from AI skeptics is about ROI. Companies are spending hundreds of billions on AI infrastructure, and where are the returns? Enterprise AI adoption feels slow. Most companies are still running pilots. The revenue from AI-native products is a fraction of total revenue at most software companies.

These are not wrong observations. They are observations that fundamentally misunderstand how transformative technology cycles work.

Every major technology shift — the internet, mobile, cloud computing — followed the same pattern. Massive infrastructure spending first, adoption lagging by years, skeptics pointing to the gap between investment and returns, and then a tipping point where adoption curves went exponential. Internet infrastructure spending peaked in 2000. E-commerce didn't surpass 10% of retail sales until 2017. Seventeen years. Cloud computing was "the future" in 2006 when AWS launched. It didn't become the default deployment model until roughly 2018.

The timing objection isn't an analytical framework. It's impatience dressed as analysis.

AI is following this exact arc — except faster, because the infrastructure is more accessible and the capabilities are more immediately useful. But "faster" still means years, not quarters. Companies don't reorganize their workflows, retrain their employees, and rebuild their processes in a single budget cycle. The people who expect AI ROI to show up in 2025 or 2026 earnings reports are measuring a decade-long transformation with a quarterly ruler.

AI Is Not a Tool. It's a Different Framework.

This is the part most people get wrong, and it's where the skeptics reveal that they're not using the technology themselves.

AI is not a better spreadsheet. It's not a faster search engine. It's not an upgraded autocomplete. Treating it as a tool — "let me use AI to write this email faster" — produces marginal improvements. That's where most companies are today, and it's why the ROI skeptics can point to modest productivity gains and declare the revolution overhyped.

But that's like evaluating the internet in 1998 by measuring how much faster people could send faxes via email. The transformative value wasn't in doing old things faster. It was in doing entirely new things that weren't possible before.

AI requires a framework change. You don't bolt it onto existing workflows. You rebuild workflows around it. You design agentic pipelines where AI handles the high-volume, high-consistency tasks — reading every earnings filing, monitoring every data source, producing first drafts at machine speed — and humans handle the judgment calls that require context, conviction, and taste.

I know this because I built it.

What I Actually Built

I'm a professional tech investor. Over the past year, I built an AI-powered research system that fundamentally changed how I work. Not incrementally. Fundamentally.

The system reads every earnings transcript and press release for over 500 companies, every quarter, within hours of filing. It applies a consistent forensic methodology — tracking what management says against what actually happens, detecting growth trajectory shifts, identifying narrative changes that signal inflection points. It monitors news across the technology sector continuously, screens for relative price strength and weakness daily, and maps intelligence across thematic clusters weekly.

Every analysis follows the same structured framework. Every claim is cross-referenced against prior quarters. Every management promise is tracked against subsequent delivery. The result is a longitudinal record updated every quarter, with a consistency and depth that would be physically impossible for a single human analyst.

But here's the critical point: the system doesn't make investment decisions. I do.

The AI reads hundreds of earnings filings and tells me what changed. I decide which three changes matter this week. The AI surfaces a pattern across several companies where booking metrics are diverging from revenue growth. I decide whether that pattern is a leading indicator or statistical noise. The AI drafts a market analysis. I rewrite it with the conviction and context that comes from having actual money at stake.

This isn't AI replacing the analyst. It's an analyst operating at a scale and consistency that was previously only available to institutional research teams with dozens of employees. One person, augmented by machines, producing intelligence that competes with teams of twenty.

That's the AI opportunity that most people are missing. Not because the technology doesn't work — it works spectacularly well. But because extracting its full value requires rethinking how you work from the ground up. And most people, most companies, haven't done that yet.

The Consciousness Red Herring

There's a subset of AI discourse that I find particularly unhelpful: the debate about whether AI will become conscious, whether we should fear AGI, whether superintelligence will make humanity obsolete.

This is philosophy cosplaying as market analysis.

Increased computational performance doesn't evolve into consciousness. That's a category error. Your calculator didn't develop feelings when it got faster. The question isn't whether AI will "think" — it's whether AI, as it exists today, materially changes how work gets done and value gets created. And the answer to that question is unambiguously yes, if you've actually used it.

I find that the people most skeptical about AI's transformative potential are, almost without exception, the ones not using it intensively. Every power user I talk to — people who have spent hundreds of hours working with the technology, building with it, integrating it into their daily practice — comes away with the same reaction: this is bigger than people realize.

Does AI have flaws? Absolutely. It hallucinates. It loses context. It can be confidently wrong. Do humans have flaws? Obviously. We forget things. We get tired. We process information orders of magnitude more slowly than machines. We have emotional biases that cloud judgment.

The question was never whether AI is perfect. The question is whether a human augmented by AI is better than a human alone. And that question has been answered.

The Window

Markets are driven by narratives, and right now the dominant narrative is fear. War in the Middle East. Tariff uncertainty. Inflation anxiety. "SaaSpocalypse."

This is what opportunity looks like. Not when everyone agrees the future is bright and stock prices reflect that consensus. Opportunity appears when the market is looking somewhere else entirely, and the fundamental picture is moving in the opposite direction from the price.

The AI infrastructure buildout is accelerating. Enterprise adoption is following the same adoption curve as every prior technology revolution — slow at first, then all at once. The companies building on AI are posting accelerating growth. The ones ignoring it are being left behind.

The war will eventually end. The tariff uncertainty will resolve one way or another. The fear will subside. And when investors look up from the crisis headlines and check what happened to the technology landscape while they weren't watching, they'll find it changed more than they expected.

The biggest technology shift of my generation is happening right now. My opinion isn't based on what analysts think, what hedge funds are buying, or what the consensus narrative says. It's based on direct, daily experience building with and deploying this technology.

The future is already here. Most people are just watching Iran.