AI’s Having a Moment Again (And Your Infrastructure Is Sweating)

Okay, so what’s “viral” in AI in the last 24 hours?

Not a dancing robot on TikTok. Not an app that turns your cat into a Renaissance duke.

It’s enterprise AI news. The kind of “viral” that spreads through boardrooms, budget spreadsheets, and the haunted hallways of legacy IT.

CEOs are throwing money at AI… while the wiring in the walls smokes

According to a January 14, 2026 survey from NTT and WSJ Intelligence (359 execs at large firms), 68% of global CEOs plan to increase AI investments over the next two years. Source

Alright, cool. Except only 18% say their tech infrastructure is actually optimized to scale AI.

Just to be clear: that’s like buying a speedboat because you love the ocean, but your driveway is still gravel and your car is a 2006 Corolla with one hubcap.

What’s getting in the way?

• Computing constraints (translation: GPUs are expensive and everyone wants them)
• Legacy systems (translation: your “modern stack” still has a mainframe energy)
• Data bottlenecks (translation: you can’t do AI magic with data that looks like a junk drawer)

Budgets are going up, and not by “a little”

Nearly half of these execs expect AI budget hikes of 11% or more. Not “we’ll see what happens.” Not “let’s run a pilot.” This is “we’re buying the thing and we’ll figure out the instructions later.” Source

They’re prioritizing predictive AI right now, but they believe the biggest value comes from combining predictive + generative + agentic AI.

Which makes sense. Predictive AI tells you what might happen. Generative AI writes the email about it. Agentic AI… goes and does something about it while you’re microwaving leftovers.

Surprise: AI scaling has an environmental tab (and it’s not small)

Also in that same survey: 83% of respondents say they’re concerned about the environmental costs of scaling AI fast. Source

So now we’re not just asking, “Can we scale AI?” We’re asking, “Can we scale AI without turning the data center into a space heater with a PR problem?”

Enter photonics, the “maybe we don’t have to melt the planet” option

This is where it gets interesting: 91% of execs are familiar with photonics, and 55% are interested in adopting it. NTT is pushing optical networking tech (IOWN) that could deliver up to 100x power savings. Source

Okay, I’m not saying photonics is a magic wand. But if someone offers you “same AI ambitions, less electricity bill,” you at least stop scrolling and read the email.

Workday says companies are wasting AI’s “time savings” fixing AI’s homework

Workday dropped a fun little reality check: companies are wasting nearly 40% of the time AI supposedly saves… because people are stuck fixing low-quality outputs. Source

So the pitch is “AI will save you time,” and the plot twist is “you’ll spend a bunch of that time cleaning up AI’s messy room.”

Workday’s point is simple: invest in data quality if you want real ROI. Because garbage in, garbage out… but now it’s garbage out in a confident tone with bullet points.

Identity and transparency: the unsexy stuff that keeps AI from face-planting

Another headline floating around today: identity and transparency are being positioned as the anchors for AI growth in 2026—helping with scalability and efficiency, while exposing gaps in governance, skills, and security. Source

Which, yes. If you don’t know who can access what, what the AI is doing, and why it made a decision… you don’t have “AI strategy.” You have “future audit findings.”

So what’s the real takeaway?

CEOs are all-in on AI spending. The infrastructure is not all-in on cooperating.

And the next phase of “viral” AI news isn’t going to be another chatbot trick. It’s going to be the behind-the-scenes stuff: power, compute, data quality, identity, governance, and whether your organization can actually work with AI like a teammate instead of a chaotic intern.

Three practical moves (before you buy another shiny AI tool)

• Audit your AI readiness honestly (if you’re in the 82% with non-optimized infrastructure, welcome to the party)
• Fix data quality like you mean it (because rework is where ROI goes to die)
• Get serious about identity and transparency (security and governance aren’t optional when AI starts acting on your behalf)

One last thing

The NTT/WSJ survey also says 83% predict human–AI collaboration skills will be critical by 2030. Source

So, alright: the future isn’t “AI replaces everyone.” It’s “AI joins the team,” and everyone has to learn how to work with the new coworker who never sleeps, sometimes hallucinates, and absolutely will not stop offering to “help.”