Wall Street is grappling with a fundamental question: will artificial intelligence create the next generation of tech giants or simply make existing software companies more powerful? The answer could determine the fate of hundreds of billions in market value as investors flee traditional software stocks amid uncertainty about AI’s disruptive potential.
The $100 Billion Question
ServiceNow CEO Bill McDermott’s recent defense against “speculation that AI will eat software companies” did little to calm investor nerves. Despite his spirited earnings call remarks, ServiceNow’s shares have fallen 22% since, part of a broader software rout that has wiped out massive market value across the sector. This investor anxiety reflects deep uncertainty about what to expect from what some executives privately call “the most disruptive shift since the cloud.”
“This is more disruptive than the cloud, this is inference, judgment, reasoning,” said an executive at one of the biggest software-as-a-service groups. “Those other big changes happened over several years: this is happening at lightning speed on many fronts.” Yet established software companies have extensive sales organizations, customer relationships, and integrated systems that new AI startups lack. As Box CEO Aaron Levie notes, “The business model post-cloud was even better for the incumbents.”
The Productivity Paradox
While executives debate AI’s long-term impact, evidence about its current effectiveness presents a mixed picture. A European Investment Bank study found AI increases productivity in EU companies by about 4%, with no evidence of job losses and some wage increases. Similar productivity gains are noted in the US, with annual growth doubling to 2.7% in 2025, according to Stanford Digital Economy Lab director Erik Brynjolfsson.
Yet a National Bureau of Economic Research study reveals that many business leaders see little measurable impact from AI adoption in their companies over the past three years, with over 80% reporting no effects on employment or productivity. This disconnect between potential and current reality creates what AllianceBernstein’s Jim Tierney calls “the $64,000 question” for investors: how many software companies face being “disintermediated” as AI agents step between existing systems and human workers?
The Agent Revolution and Security Challenges
The emergence of AI agents represents both opportunity and threat. Companies like OpenAI and Anthropic are developing agent-based systems that could either integrate with existing software or replace it entirely. OpenAI’s Frontier framework aims to solve “fragmentation” between today’s “clouds, data platforms and applications,” potentially putting its technology at the center of enterprise IT.
However, these new systems face significant security challenges. A recent MIT-led study analyzing 30 agentic AI systems reveals widespread security and transparency issues, including lack of risk disclosure, monitoring capabilities, and safety evaluations. Open-source projects like OpenClaw have demonstrated vulnerabilities to prompt injection attacks that could compromise sensitive user data, highlighting the privacy and security challenges of agentic AI.
Financial Services: A Case Study in Disruption
The wealth management industry provides a tangible example of AI’s disruptive potential. When fintech company Altruist developed an AI-led tool to help financial advisers personalize clients’ investment strategies, shares of major wealth managers on both sides of the Atlantic suffered double-digit declines. Altruist CEO Jason Wenk warned the technology makes “average advice a lot harder to justify,” demonstrating how AI can undermine traditional business models by delivering superior personalized service.
The Orchestration Imperative
As AI agents proliferate, the technology needed to organize them alongside human workers – known as “orchestration” – could become the key to controlling the next computing platform. ServiceNow CFO Gina Mastantuono argues her company already has much of what will be needed to control these new AI-based work systems, including governance and security controls that customers will be wary about switching from.
Yet RBC Capital Markets analyst Rishi Jaluria warns that agents present a potential “long-term disaggregation risk.” Companies might continue to hold data in their current systems but use AI to “build custom applications” around that core, pushing old software into the background as more value moves to the new agent layer.
Investment Implications
For now, few investors are prepared to bet on any particular outcome. The uncertainties echo the move to the cloud, when software companies struggled to persuade Wall Street they could thrive in the new environment. Yet as Tierney notes, “This is certainly a headwind, but not necessarily deserving of the sell-offs we have seen in software.”
The transition to AI-centric computing may cause upheaval, but it’s far from the existential crisis the market is pricing in. Companies are wary about uprooting software systems that hold their most important corporate data or embed their core work processes. As the SaaS executive bluntly puts it, AI companies are “maths nerds, they’re not customer-focused systems builders.”
Ultimately, the question isn’t whether AI will transform enterprise software, but how quickly and who will capture the value. With evidence mounting about productivity gains but security concerns persisting, and established players facing both disruption and opportunity, the only certainty is that the software landscape will look very different in five years – whether today’s leaders are still standing or not.