The Great Software Reshuffle: Why AI Isn't Killing Apps, It's Changing Customers

$1 trillion vanished in a single week. That's how much market cap enterprise software companies lost this year, and at Morgan Stanley's tech conference, CEOs weren't asking "How can AI make us more efficient?" anymore. They were asking "Will we survive?"

The shift was palpable. What started as a conversation about trimming costs with copilots has evolved into an existential reckoning.

The Question Changed Everything

David Chen, Morgan Stanley's head of global technology investment banking, noticed the biggest change wasn't in the answers—it was in the questions. Last year, companies talked about shaving a few percentage points off operating costs with AI automation. That's now table stakes.

"I don't think investors really wanted to hear about how people are being more efficient with AI," Chen told us. "They really, really wanted to hear, are you a beneficiary, or does AI threaten your overall business?"

And companies were finally ready to answer honestly. Especially the enterprise software companies watching their valuations crater.

The Great Divide

Chen drew a stark line between two types of software companies: those doing deterministic work—calculating payroll, sending invoices, tasks where being wrong by 2% is a real problem—and those essentially organizing public data behind a nice interface.

The former still has a moat. The latter is in serious trouble.

"AI doesn't kill software," Chen said. "It's reshuffling it."

But he didn't sugarcoat the reality for companies on the wrong side of that line. He called it "wartime, not peacetime." Interestingly, boards are starting to prefer product-oriented CEOs over sales-and-marketing types. When you need to reinvent your back-end to be AI-native, you want someone who understands architecture, not just pipelines.

From Humans to Agents

CNBC producer Jasmine Wu coined a phrase that captures the shift perfectly: we've moved from SaaS (Software as a Service) to SaaaS (Software for agents as a Service).

Box CEO Aaron Levie told us agents are now his new customer base, and he could see that business becoming 10 times bigger than the existing one. The implication is massive: the software that survives won't be the software humans use. It'll be the software agents use.

This isn't just a technical shift—it's a fundamental reimagining of who software serves.

The Infrastructure Reality Check

When asked whether AI infrastructure spending would be higher or lower in 2027, Chen's answer was telling: "Probably a similar level." If accurate, the AI capex cycle from hyperscalers may be approaching its peak.

For the year ahead, Chen predicts a rebalancing of winners and losers in enterprise software. Cybersecurity stands out as having the right competitive moat characteristics and being a clear AI beneficiary rather than victim.

He also flagged a wave of next-generation companies in semiconductors and systems focused on solving connectivity, compute, and energy bottlenecks constraining the AI buildout.

The American Software Reckoning

For US companies, this reshuffling has profound implications. Traditional SaaS giants like Salesforce, ServiceNow, and Workday are scrambling to prove they're on the right side of Chen's divide. Meanwhile, cybersecurity players like CrowdStrike and Palo Alto Networks are positioning themselves as AI beneficiaries.

The startup ecosystem is also adapting. New companies are emerging specifically to build software for AI agents, not humans—a fundamental shift in product design philosophy.