IBM Thinks Your Data Is Too Stubborn to Move (and AI Agrees)

For years, enterprise IT strategy followed a simple script: Move workloads to the public cloud. Amazon (NASDAQ: AMZN) Web Services, Microsoft (NASDAQ: MSFT) Azure, and Alphabet's (NASDAQ: GOOG) (NASDAQ: GOOGL) Google Cloud built massive businesses on that assumption.

But the AI era is throwing a spanner in the cloud computing works.

Image source: Getty Images.

The gravitational pull of stubborn data

"Hybrid cloud and AI are two sides of the same coin," said Dr. Hillery Hunter, CTO and general manager of innovation for IBM (NYSE: IBM) Infrastructure. "Where your data is, is becoming very much a grounding factor, because where your data is, is then where you naturally want to build out your AI."

That makes sense, right? It's the concept of data gravity in action.

Once enterprise data settles somewhere, it tends to stay put. It accumulates. It gets comfortable. And moving it becomes a difficult production. Taking a large company's data out of its natural habitat incurs egress fees, data transfer latency, security reviews, and the need to explain to regulators why sensitive information is taking a field trip, perhaps across international borders.

For some organizations, AI accelerates the move to public cloud; they've already committed to that architecture and will consolidate further. But Hunter sees an equally strong motion in the opposite direction: enterprises with significant on-premises data gravity that need to bring AI capabilities to where the data already lives.

"I see both motions equally developing," Hunter said in a recent interview with The Fool. "People are taking different decisions across that spectrum."

IBM has receipts for the hybrid cloud thesis

Research from the IBM Institute for Business Value supports this split. Nearly three-quarters of executives surveyed are backing away from cloud-first defaults and deeper into their corporation's data centers. And 72% of organizations said that cloud costs in production exceeded expectations by an average of 1.5 times. IBM's hybrid computing model supports both sides of this equation, pairing the centralized IBM Cloud services with mainframes and Power servers in the data center.

IBM isn't the only company leveraging the data gravity idea.

• Oracle (NYSE: ORCL) is making a similar bet from a different starting point. The database giant has positioned itself as a "data vault" provider, optimizing for enterprises whose mission-critical data already lives in Oracle systems and databases.
• CrowdStrike (NASDAQ: CRWD) argues that its Falcon security system creates useful data gravity wherever it goes. "This integration means data doesn't just accumulate; it's immediately actionable, contributing to threat detection and response," the company stated in a company blog post.
• Seagate (NASDAQ: STX) highlights how large data sets tend to attract smaller data collections to the same storage system, simply because it's easier to move the smaller ones. Enterprises should account for the data gravity effect when planning their large-scale storage setups, breaking up huge data buckets before they grow too large to manage.

The investment case for stubborn data

Different tech veterans are reaching the same observation from different angles: Data doesn't like to move, and pretending otherwise gets expensive.

IBM's infrastructure business is built around this premise. The company's Power servers and Z mainframes are designed for enterprises that need AI capabilities without relocating mission-critical data to third-party clouds.

Hunter emphasized that IBM's systems deliver "six nines of resilience" (meaning 99.9999% uptime, or just a few seconds of downtime per year) and twice the power efficiency of competing server architectures.

The hyperscalers aren't going anywhere; Amazon, Microsoft, and Alphabet have the scale and momentum to keep growing cloud revenues for years. But IBM is betting that "move everything to the cloud" was never the right answer for every workload. The company will help if you insist on cloud computing, but some data is just too heavy to lift.

For IBM investors, the hybrid cloud model is the thesis. If enterprises keep pulling AI-related workloads back from the cloud to their data centers, IBM's infrastructure business has a huge growth opportunity in that shift.

If cloud-first comes back in style over the years, IBM is swimming against the tide.

Place your bets accordingly. I, for one, see the value of local data collections rising as the datasets grow larger.

Anders Bylund has positions in Alphabet, Amazon, and International Business Machines. The Motley Fool has positions in and recommends Alphabet, Amazon, CrowdStrike, International Business Machines, Microsoft, and Oracle. The Motley Fool has a disclosure policy.