Software is not the next AI casualty — it is where the AI trade is quietly being repriced
The market spent months pricing software as if generative AI would flatten moats and crush pricing power. This week’s rebound argues the trade is shifting: the likely winners are incumbent workflow and data-layer platforms that can fold AI into existing distribution, not the companies AI is supposed to replace.
The market’s software debate is stale. Investors are still arguing as if generative AI automatically destroys application software and information-services businesses, even as price action and company updates are pointing the other way. What changed is not that AI risk disappeared; it is that the market is starting to separate fragile feature vendors from platforms that already own workflow, trusted data, and customer distribution. That is why software is not the next AI casualty — it is where the AI trade is being repriced toward incumbents that can make AI more useful inside systems customers already pay for.
The clearest evidence is in the sector move itself. Reuters highlighted that the software group had gone from being down roughly 30% earlier in 2026 to down less than 2% by June 3 after a sharp rebound. That is not the tape of a market discovering software is obsolete. It is the tape of a market that first priced in maximum disruption, then started buying back the names with credible AI monetization and durable distribution. The important shift is qualitative: the debate is no longer whether AI touches software, but whether lower-cost functionality actually makes the platforms that organize work more valuable.
Datadog is the cleanest proof point because it sits in a category where AI creates complexity before it creates simplicity. The stock is up 73.7% year to date, trades at 22.57x sales, and still only posts a 3.7% net margin. On the surface, that looks like the kind of valuation bears would target in any "AI commoditizes software" argument. But the company just raised guidance, and Reuters tied its strength to AI-driven demand in cloud security and observability. That matters because AI agents do not reduce the need to monitor systems, permissions, model behavior, and workloads; they increase it. If AI expands the number of things enterprises must watch and secure, then observability is not a casualty of AI adoption. It is part of the toll road.
The same logic applies, more quietly, to information and workflow incumbents that the market treated as obvious losers. Thomson Reuters at 24.32x earnings and 4.83x sales is not priced like a broken franchise, but its shares are still down 33.9% year to date. RELX trades at 23.11x earnings and 4.78x sales, with a 21.5% net margin, yet the stock is down 12.9% this year. Those are not numbers that scream structural collapse. They look more like a market discounting disruption risk into businesses that still have what AI lacks on its own: embedded professional workflows, proprietary content, compliance sensitivity, and customers who care about auditability more than novelty. When Thomson Reuters says professionals accountable for high-stakes outcomes are choosing its AI products, that is the moat in plain English. In legal, tax, risk, and research, the winning product is rarely the cheapest answer. It is the answer that can be trusted inside an existing workflow.
That is also why the market should stop lumping software together. There is a real difference between a horizontal app that sells convenience and a platform that sits at the data layer or inside mission-critical process. Microsoft helps frame the distinction. At 24.58x earnings and a 39.3% net margin, it is not being valued as if AI will wreck software economics; it is being valued as if distribution and infrastructure will capture more of the value stack. That same principle, in smaller form, is what the rebound in software is starting to recognize. If AI lowers the cost of generating features, then standalone features get cheaper. But the systems that authenticate, govern, integrate, monitor, and distribute those features often get more important, not less.
Yes, the bear case is real. AI agents can compress interfaces, automate routine tasks, and make some premium software look overpriced. That is exactly why names tied to legal and information services sold off so hard earlier this year, and why a stock like Datadog still invites skepticism at a premium multiple. But that comparison often ignores where monetization actually sits. Customers do not just buy answers; they buy workflow continuity, trusted data, security, and accountability. AI may commoditize some front-end functionality, but it can simultaneously raise the value of the platforms that make machine-generated output usable inside regulated or operationally complex environments.
If anything, this week’s rebound suggests the market is finally pricing that distinction. The winners are unlikely to be every software company with an AI slide deck. They are more likely to be the incumbents that can absorb AI into products customers already rely on every day. Datadog fits because AI expands the monitoring burden. Thomson Reuters and RELX fit because AI increases the premium on authoritative content and defensible workflow. Microsoft fits because distribution and infrastructure remain the easiest places to monetize a platform shift. The old trade was "AI kills software." The better trade is that AI lowers the cost of functionality and raises the value of distribution, data, and workflow control.
That is the reframe investors need. Software is not being uniformly disintermediated; it is being sorted. The companies most exposed are the ones selling isolated features without proprietary data, embedded workflow, or meaningful distribution. The companies with those assets are not watching AI hollow out their business models — they are using AI to deepen them.
What we would watch from here is not broad sector rhetoric but proof of adoption inside existing customer bases: raised guidance, stable margins, and evidence that AI features are improving retention or wallet share rather than just inflating costs. If incumbents start losing usage while AI-native tools win the workflow outright, this thesis weakens fast. For now, the market’s rebound is saying something simpler: software is not the next AI casualty. It is where the AI trade is finally being repriced with more nuance.
Our take, not advice. This is opinion commentary — informational only, not personalized investment recommendations. Markets carry risk. Do your own research and consider your own situation before any trade.
▌The Daily Briefing · Free
A new stock idea, every evening.
One stock worth watching each weekday, plus the analysis behind it. Free, in your inbox.
