A moment of clarity arrived in the markets this week: artificial intelligence can create winners — and devour businesses that sell yesterday’s software.
Traders reacted violently. Headlines tallied hundreds of billions — Bloomberg cited as much as $1 trillion wiped from software market caps in a single session — while sector-focused funds and a long list of cloud names slid sharply. The selloff has a blunt logic behind it: agentic AI systems are getting better at performing tasks that once required licensed software and human workflows, and investors are repricing companies accordingly.
Panic, profit-taking and a new market hierarchy
The market isn’t simply melting down across the board. Investors are shifting toward a narrower group: the model-creators and infrastructure providers (the chips, the cloud services, the firms building the big AI brains). At the same time, many pure-play software vendors — the folks who charge by the seat or module — have been treated as potentially expendable.
That dynamic helps explain the grim numbers: specialty cloud funds have plunged, and individual names like HubSpot, Figma, Atlassian and Shopify have been punished hard. The WisdomTree Cloud Computing Fund, for instance, was down sharply in 2026 as fears about AI disruption accelerated.
But not everyone agrees that this is a binary outcome. Box CEO Aaron Levie called it “the most exciting moment we’ve ever had,” arguing that businesses will still pay for specialist products rather than cobbling together internal systems and liabilities. Salesforce and ServiceNow executives have made similar points: if you control the customer data and integration layer, you can layer AI on top and retain customers rather than be displaced.
What changed this week: agents got real
Two developments crystallized investor nervousness. First, a wave of practical, agentic tooling arrived from model-makers and startups — products that stitch AI into workflows, automate repeated tasks and, importantly, can be customized. Anthropic’s recent push to extend Claude into coworker-style productivity tools with new legal and marketing capabilities — and a plugin approach that encourages customization — was cited by many as a major inflection point. The sense that AI could be turned into purpose-built agents, cheaply and quickly, cut straight through sellers of legacy workflows.
Second, capital flows and valuations concentrated on a handful of model owners. Reports of massive fundraises and sky-high private valuations for companies building foundational models signaled where the market’s faith now sits: with those who own AI’s core intellectual property and scale. That concentration has a feedback effect — investors who believe the winners will be few and giant then sell everything else.
Retail, ‘dumb money’ and the buy-the-dip instinct
Amid institution-led repositioning, retail investors have been an odd counterweight. Platforms that democratized trading have turned ordinary accounts into a steady source of “buy the dip” demand. JPMorgan analysts noted record retail activity in January, even as total purchases softened in recent days. Some market strategists borrow the old pejorative “dumb money” to describe these buyers; others see them as a persistent support that changes volatility and recovery patterns.
That tug-of-war — institutional reweighting away from software vs. retail buying into weakness — helps explain why broad market indices sometimes look less bruised than the software-heavy corners of the market.
Between panic and opportunity
The selloff is not an automatic handshake for value investors. Several analysts and venture investors argue that the chaos creates openings: companies with strong integration, sticky revenues, and clear AI roadmaps could be materially cheaper. Cantor and Stifel analysts have pushed back on blanket doom for particular vendors, saying on-the-ground enterprise buyers aren’t yet stripping out seats. Byron Deeter and others have tweeted that “chaos creates opportunity,” encouraging buy-the-dip bets for those who can pick durable franchises.
Still, skepticism is warranted. If agentic systems truly scale across multiple enterprise functions, incumbents that fail to adapt will see pressure — both on top-line growth and on pricing power. The speed and scope of that disruption remain the core debate.
What to watch inside companies
Executives who control key data flows and integration points — CRM systems, enterprise search layers or workflow platforms — have leverage because customers don’t want brittle custom solutions. Firms that reposition as the semantic or orchestration layer for AI may thrive rather than wither. For practical reading on how agentic features are being folded into consumer and business products, note Google’s push to add booking and agentic capabilities to Search and the deeper workspace integrations rolling out for Gemini, both of which show how big platforms are embedding agentic functions across product lines (Google’s AI Mode agentic booking, Gemini’s Deep Research linking into Gmail and Drive).
Investors should also pay attention to which model providers enable easy customization and plugin ecosystems versus those that lock functionality behind opaque APIs. That difference will determine how rapidly third parties can substitute for software vendors. Microsoft’s recent moves in model and media tooling are another piece of the puzzle as the competitive landscape shifts toward model-first strategies (Microsoft’s MAI-Image-1).
The markets are sorting themselves. Some software companies will be rewritten, others will be retooled, and a few will be swept aside. For investors and customers alike, the immediate task is not to predict who wins every matchup, but to understand whether a vendor is becoming the bedrock layer for AI-driven workflows — or merely a casualty of them.
This moment feels dramatic. But in technology, disruptive swings usually breed new businesses as quickly as they break the old. That doesn’t make the next quarter any less volatile. It just means the story is far from over.