The software sector has weathered a turbulent stretch. After a sharp sell-off earlier in the year — a period some described as a "SaaS apocalypse" in which strong companies were punished alongside weak ones — investor interest has returned, with software-focused indices recovering to levels last seen before the downturn. Yet recovery in price does not necessarily mean a definitive bottom has been reached. The more important question is whether the industry has genuinely changed how it operates, and there is reason to believe it has.
The Message Has Landed
Software companies appear to have absorbed a hard truth: the market for SaaS will not look the way it did six months or a year ago. The era in which a vendor could win by simply enumerating features — or by boasting about how many AI agents it had deployed — is ending. What customers now want is a clear line from the technology to the bottom line. It is one thing to claim that AI or agentic technology improves productivity and efficiency. It is another to demonstrate what that means for overall profitability and revenue generation.
Encouragingly, this shift is already visible. Across numerous industry events in recent months, vendors have moved away from the familiar marketing spiel and toward concrete proof points. In some cases, they have put customers on stage to describe measurable results — including actual revenue lifts attributable to the software. That move from abstract promise to documented outcome is the single most reassuring signal the market has received.
Buyers Demand Proof, Not Promises
Buyers are no longer satisfied with outcome claims buried in a slide deck. They want tangible, verifiable evidence, and increasingly that evidence must come from third-party validated assessments rather than vendor self-reporting. Anyone can assert that they "drove a little efficiency." Serious evaluation requires digging into the full picture: what the baseline looked like before implementation, what was done during deployment, how much it cost, and what the actual return turned out to be. This rigor raises the bar for vendors but ultimately strengthens the credibility of the entire category.
The Bridge Problem and the Shadow IT Risk
General-purpose AI tools illustrate both the opportunity and the danger of this moment. People are now using assistants like Claude in remarkably practical ways — one example being someone effectively treating it as a personal CFO that handles accounting tasks. Many sectors are being genuinely disrupted by this kind of usage.
It would be naive to bury one's head in the sand and insist these tools change nothing. At the same time, the claim that a general AI assistant can wholesale replace an entire CRM remains untrue. The more accurate description is that such tools serve as a bridge. If a user needs to pull up data in a way that suits their workflow, an AI assistant can quickly spin up a small, custom application to do exactly that. The obvious risk is that this proliferation spirals into a serious shadow IT problem for organizations — ungoverned, unmonitored applications multiplying outside official oversight. This is precisely why large software vendors are now emphasizing that customers can build applications on top of their platforms in a way that is properly governed, properly guardrailed, easier to manage, and — critically — instrumented so that value delivered can actually be tracked.
Who Survives the Transition
Two factors will determine which incumbents weather this shift well: the strength of the product and the depth of existing customer relationships. The largest players — Microsoft, Salesforce, ServiceNow, SAP — carry serious gravitational pull simply because they are deeply entrenched with their customers. But entrenchment alone is not the strategy. These vendors are working to demonstrate that they can incorporate not only data but the context surrounding that data. A customer information profile becomes far more powerful when enriched with ambient data — world events, weather activity, even something like an impending regional railroad strike. Layering that context into a platform is what turns raw information into insight that drives real business outcomes.
Consolidation Is Desired but Hard
As AI becomes embedded everywhere, a natural question is whether enterprises will consolidate around a small number of vendors. Buyers clearly want this, but it is easier said than done. Enterprises run many software platforms for structural reasons: companies acquire other companies, each bringing its own systems, and crises like the pandemic forced software to be stood up rapidly with little time for consolidation. Coalescing around fewer platforms takes time, which is why every vendor is competing fiercely to remain on the short list of two or three providers an enterprise seriously considers.
Even so, complete consolidation is unlikely and arguably unwise. Concentrating an entire technology stack with one or two vendors is a risk question as much as an efficiency question. Given cybersecurity threats, no enterprise wants all of its eggs in one or two baskets. Multiple platforms will remain a permanent feature of the enterprise landscape.
Disillusionment, Defense, and the Reality of Time
The headwinds are real, and much of the difficulty stems from overheated expectations. There has been a fair amount of hype, amplified by widely circulated studies — including one from MIT claiming that 95% of AI pilots fail. That figure should not be read as universally true; it likely skews toward smaller companies rather than every segment. But it captures a genuine mood of disillusionment. AI was sold as something that could be deployed and deliver returns almost immediately, and that has not been the experience.
This sentiment has produced some counterintuitive market behavior. Defense companies, for instance, have struggled despite defense budgets surging and active conflicts driving genuine replenishment needs. One observation is that defense firms which referenced AI or software on their earnings calls actually traded more negatively, their sell-offs correlating with AI-themed stocks rather than benefiting from the AI narrative. The market, in other words, has begun to treat unsubstantiated AI references as a liability rather than an asset.
The underlying reality is that returns take time. Vendors now recognize that convincing customers, assembling credible proof points, and translating those into stock performance is not an overnight process. And in the long run, AI will follow the path of every transformative technology before it: eventually it becomes table stakes, embedded so thoroughly into products that merely "having AI" means nothing. At that point, the only thing that will matter is the question that matters now — how does this software, and the company behind it, work with me as a customer to deliver the business outcomes I need to move my business forward.