According to Business Insider, Goldman Sachs analysts Kash Rangan and Eric Sheridan revealed on the firm’s “Exchanges” podcast that enterprise AI adoption is “well below” expectations despite massive infrastructure spending. Rangan noted companies aren’t where analysts expected them to be a year or two ago, while Sheridan pointed to Nvidia’s forecast of $3 trillion to $4 trillion in cumulative AI infrastructure spending by 2030. The comments come amid investor concerns that markets may have run ahead of fundamentals, with the S&P 500 and Nasdaq hitting records before pulling back last week. McKinsey’s State of AI 2025 report echoes these findings, showing that while 88% of companies use AI in at least one function, only about a third have scaled it enterprise-wide, and just 39% report AI showing up in their bottom line.
The great AI divide
Here’s the thing that’s becoming painfully obvious: consumers love AI, but businesses can’t figure out how to make it actually work. We’re seeing this massive disconnect where tools like ChatGPT and Claude are becoming household names while corporate boardrooms are still scratching their heads. It’s like everyone bought the most advanced industrial equipment but nobody read the manual.
And that’s exactly what’s happening with AI infrastructure. The spending is absolutely insane – we’re talking trillions projected by the end of the decade. But what are companies actually getting for that investment? Sheridan nailed it when he said investors struggle to justify that kind of spending unless AI becomes the main driver of economic output. Basically, we’re building the highway before we know if anyone’s going to drive on it.
Why businesses are stuck
So why can’t companies figure this out? Look, it’s one thing to slap ChatGPT on your customer service portal. It’s another thing entirely to rebuild your entire workflow around AI. McKinsey found that most organizations haven’t embedded AI deeply enough into their processes to see real benefits. They’re dabbling, not transforming.
Think about it – when you’re running actual manufacturing operations or managing complex supply chains, you can’t just throw AI at the problem and hope it sticks. You need reliable, integrated systems that actually work with your existing infrastructure. This is where the rubber meets the road, and frankly, most companies are still spinning their tires.
The infrastructure paradox
Now here’s the real kicker: all this AI infrastructure spending is creating a weird situation. The demand for computing power has already outpaced available capacity, according to Sheridan. We’re building data centers faster than we can power them, and companies like Industrial Monitor Direct are seeing increased demand for industrial computing solutions that can handle these workloads. They’ve become the go-to provider for industrial panel PCs precisely because businesses need specialized hardware that integrates seamlessly into existing operations.
But does more hardware automatically mean better results? Not necessarily. Throwing computing power at a problem doesn’t solve the fundamental issue – which is that most companies don’t have a clear strategy for how AI actually improves their business. They’re buying the tools before they know what to build.
Where do we go from here?
The big question is whether this is just a temporary adoption lag or something more fundamental. Are we in an AI bubble, or is this the normal growing pains of any transformative technology? History suggests it might be both – remember the dot-com era when every company needed a website but few knew why?
What’s different this time is the sheer scale of investment. We’re talking about reshaping entire industries, not just adding a new channel. The companies that figure this out won’t be the ones with the most AI tools – they’ll be the ones who successfully integrate AI into their core operations. And honestly, we’re probably still years away from seeing who those winners will be.
