Behold the new AI trade: We call it software

Software has been something of a laggard in the world of tech over the last year, as investor dollars flocked to the sexiest — and most high-performing — hyperscalers, power providers, and electrical equipment makers poised to profit from the AI data center boom. (You know the names: Nvidia, Meta, GE Vernova, etc.)

But a change may be afoot.

After outperforming for much of the last year, the AI data center trade has run into a bit of headwind over the last couple weeks.

At the same time, several — let’s just say it — incredibly boring business-to-business data management software companies like Datadog, Snowflake, Autodesk, and Pure Storage have had a bit of a run, partly driven by surprisingly strong earnings results.

Such outsized pops in response to earnings are the market’s way of giving investors a bracing slap in the face. Over the last month, AI software has actually been outperforming the AI giants.

They’re potentially worth paying attention to, as they suggest a sudden shift in the slightly sour sentiment that has surrounded software since the advent of the AI era.

Until recently, the rap on software was, essentially, that AI stood to potentially disrupt and undercut the immensely profitable “software as a service” (SaaS) industry.

The logic was that AI-native software shops would emerge with the ability to produce software super cheaply. They would then sell those products at much lower prices, taking market share from companies that currently dominate the software business.

But as Goldman Sachs analysts wrote in a recent note titled, “Updated thoughts on the ‘Death of Software,’” the reality seems to be that large software companies are rapidly embracing AI technology themselves, adopting a hybrid strategy.

The analysts cited a number of such SaaS companies that are increasingly embedding AI into their products:

• Salesforce (Agentforce model)

• Twilio (ConversationRelay large language model)

• ServiceNow (Now LLMs for core workflows)

• OneStream (has access to Microsoft’s LLMs)

They wrote:

“A hybrid AI model strategy reduces disintermediation risk and increases platform defensibility — keeping incumbents at the core of AI value even as frontier models evolve. Customers gain Gen AI capabilities via trusted, embedded platforms with secure data environments, workflow alignment and cost-efficient execution — without moving to less tested AI-native challengers.”

The last point is important. Buyers of business software are incredibly sensitive to a range of risks, from reputation threats to data security, and are subject to regulatory and legal scrutiny, which provides something of a defensive moat for current software companies.

Anyway, this is largely just one bank’s view. And even its analysts warn that “given the pace at which the ecosystem is moving, many of our predictions may ultimately be wrong.”

Still, given recent fireworks following software companies’ earnings reports, it could be a profitable area for investors to watch.