The launch of OpenAI's GPT-6 Astra has done more than reset the AI benchmark leaderboard — it has reopened one of Wall Street's oldest fears about the technology: that frontier models will eventually eat the software industry's lunch. Last week, Salesforce and Intuit fell about 4% while ServiceNow lost 5% in a single session, as investors repriced the risk that general-purpose AI could substitute for specialized enterprise software.
The selloff was broad. The S&P 500 software and services index fell 1.4%, its second consecutive day of declines, even as the broader market held up.
Why Astra Spooked Software Investors
The proximate trigger was OpenAI's launch of GPT-6 Astra the week prior, which reignited speculation that increasingly capable AI systems could compete directly with the services now sold by specialized software companies, Reuters reported. Astra's demonstrated strengths — long-horizon agentic task completion, code generation, and autonomous workflows — map uncomfortably well onto the products that enterprises buy from CRM, IT service management, and finance software vendors.
"Astra has kind of reignited the software disruption fears and it's resumed that old trend that we got used to for a while, where semiconductor stocks and data center capex beneficiaries do well, while software stocks do poorly," said Jed Ellerbroek, portfolio manager at Argent Capital Management.
The same session offered a vivid illustration of the other side of the trade. Intel jumped 9% and Qualcomm rose 3.2% after striking a deal with Amazon to develop custom AI chips, Reuters reported — a reminder that capital is chasing every layer of the AI hardware stack while the application layer bleeds.
That bifurcation has become the defining feature of AI-era equity markets. Companies that sell the physical inputs of the AI boom — chips, networking gear, data center capacity, and power — have absorbed the bulk of investor enthusiasm, while companies whose products could theoretically be replicated by a sufficiently capable model have lagged or sold off.
The 'SaaS Extinction Event' Debate
The anxiety has intensified into this weekend. A widely read Yahoo Finance analysis on Sunday asked whether Wall Street is now pricing a "SaaS extinction event," noting that Astra's launch wiped 4% to 5% off Salesforce and ServiceNow in a day and arguing that the market is beginning to distinguish between software companies that deploy AI as a moat and those that could be displaced by it.
The bear case is straightforward: if a frontier model can perform the core workflow — drafting the sales email, triaging the support ticket, generating the financial model — then the per-seat subscription that funds the software industry comes under structural pressure. In that world, the value of software companies migrates from the application layer to whoever controls the underlying models and the compute they run on.
The bull case is equally familiar. Enterprise software companies argue that their value was never the code — it is the data integrations, compliance certifications, workflow lock-in, and institutional trust accumulated over decades. Salesforce has spent two years rebuilding its product line around AI agents, and ServiceNow has made agentic workflow automation the centerpiece of its platform strategy. Both companies maintain that frontier models are their raw material, not their replacement.
A Market Pattern With Precedent
Veteran investors will recognize the shape of this debate. Every major platform shift — cloud, mobile, and now generative AI — has produced a period in which incumbents trade at a discount while markets sort genuine disruption from headline risk. So far, the AI era has favored infrastructure providers almost exclusively, a pattern Ellerbroek described as the "old trend" that Astra has now resumed.
The next test comes with quarterly earnings. Software companies that can demonstrate AI-driven revenue — not just AI-driven messaging — should decouple from the disruption narrative. Those that cannot may find that GPT-6 Astra's biggest impact was not on their products, but on their multiples.
For now, the market is voting with the sector index: software down, semiconductors up, and the question of whether AI will be the software industry's greatest tool or its greatest threat is no longer academic. It is priced in daily.
There is also a competitive dimension that pure market analysis tends to miss. DeepSeek's release of V4.1-Flash — a lean, low-cost model that analysts immediately benchmarked against OpenAI's flagship — has reinforced the sense that frontier capability is becoming a commodity, arriving from multiple directions at once. If cheap, capable models are available to any vendor at near-zero marginal cost, the advantage shifts to whoever embeds them deepest into existing enterprise workflows. That argument favors the software incumbents, not the model labs — and it is the core of the case Salesforce and ServiceNow executives will make to investors in the coming earnings season.
Either way, the repricing is unlikely to be the last. OpenAI's decision to rule out a 2026 IPO removes, for now, the prospect of a public-market check on the model developer's own value — leaving public investors to trade the disruption through the very software stocks caught in the crossfire. Until there is a clear answer on whether AI agents replace SaaS workflows or simply power them, expect every frontier-model launch to land twice: once in the product reviews, and once in the software sector's share prices.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
