Microsoft Copilot IBM NVIDIA AI artificial intelligence

IBM vs. Microsoft vs. NVIDIA and AI

Twenty years ago, NVIDIA and Microsoft were nowhere with AI, while IBM’s Watson was winning Jeopardy! and later taking on professional debaters. Conventional knowledge was that IBM was going to own AI, but no one seemed to care since projections at the time indicated that AI wouldn’t take off until sometime around 2040, when general-purpose AI was expected to become viable. Microsoft had a leading AI concept in Cortana but didn’t invest in it any more than Apple invested in Siri. Eventually, it was quietly phased out.

Then in March of last year, Microsoft announced Copilot and suddenly became the AI darling. Copilot was a workable AI that started generating revenue almost immediately. This year, NVIDIA, which had been developing AI quietly for over two decades, went vertical as the new King of AI, making Microsoft and NVIDIA two of the top three most valuable tech companies in the world.

What happened? IBM was way out front but is now almost a footnote. Maybe it’s a mismatch between strategy and leadership.

IBM’s Opportunity and Problem

IBM once had the longest-serving CEO in technology. Thomas Watson was CEO for a whopping 41 years. That consistency in leadership allowed the company to execute across decades. More recently, IBM leadership has turned over around every decade. Any effort with a payback out more than a decade (AI was in process since the early 2000s) just wasn’t going to be a top-level priority for the company. Why invest the firm’s capital in an effort that you’ll never get credit for? It shouldn’t be that way, but it is. How much would you invest in a project that you’ll never get credit for?

So, IBM under-invested in its technology, and now, instead of being ahead of the curve, it’s starting to look like it is behind (it isn’t, but perceptions are reality). It’s not because the IBM folks didn’t work hard or that watsonx is a bad product. It’s because IBM needed to invest at NVIDIA’s level from the start, but its organizational structure just didn’t allow for that. It should have either changed its policy with regard to CEO longevity or changed tactics to something more like what Microsoft has.

How Microsoft Moved on the Opportunity

Microsoft has made several tactical moves that benefited it strategically. It wasn’t the first out with a Windows product (that was Apple). It wasn’t the first with a major productivity product (that was Lotus). And it wasn’t the first with browsers (that was Netscape). However, Microsoft developed a process that allowed it to pivot quickly and take leadership from market founders. That process also worked with Xbox, though once it realized that the PS2 wasn’t the threat it thought it would become, it crippled that effort. Otherwise there likely would be no other gaming platform than Xbox today.

Microsoft failed with Zune, Cortana, and the Windows Phone because it didn’t follow the process it had for success. With AI, though, Microsoft effectively replicated what it did with Internet Explorer and bought into the market with OpenAI and ChatGPT. This allowed it to move around IBM, which was still, at the time of Co-Pilot’s launch, operating like AI wasn’t going to be a major thing for decades. Microsoft caught everyone else napping except NVIDIA, which was, as it turned out, the hardware power behind ChatGPT.


NVIDIA has the current longest serving CEO at 30 years. Jensen Huang, who admits what he did was insane, drove his company to create something that conventional wisdom said was decades away from being a market. As a result, when the market arrived, NVIDIA was, and is, considered one of the founders.

The reason NVIDIA was able to take the market so decisively was that its effort was a high priority from day one, while IBM didn’t get the focus it needed in time. It isn’t too late; with the right marketing, IBM could still get credit for leading in LLM integrity and security and being first with a viable solution. Sadly, IBM doesn’t have the marketing organization they once had under Gerstner, and recreating that, while it should be a high priority now, isn’t with any tech company I cover, including IBM.

Wrapping Up: Know Your Limitations

For a company with a CEO tenure of 10 years or less, the only viable path is to be a fast follower, which is what Microsoft did with its incredibly timely pivot to AI. In fact, given it hit first, this would have been like Microsoft bringing out Internet Explorer before Netscape was a power, which would have resulted in Netscape being stillborn. In a way, with Copilot, Satya Nadella out-executed Bill Gates, and that is an impressive accomplishment. And the concept of a CEO executing over decades puts Jensen Huang in Thomas Watson’s class in terms of empire-building, suggesting that NVIDIA’s valuation growth may be just starting. It wouldn’t surprise me at all to see NVIDIA pass Apple and Microsoft into the Number One slot, given its deeper AI foundation. I should note that at Huang’s GTC keynote, it was Apple’s Vision Pro, not a Windows PC, that was positioned as the best AI client device, suggesting that Apple has a potentially huge iPhone-like opportunity here.

In one of the Dirty Harry movies, one of Harry’s famous lines is, “You have to know your limitations.” IBM needs to understand that it can’t execute like Thomas Watson’s IBM did, but it can emulate what Microsoft did to accomplish the same result. But that means IBM needs to maintain a small core competence in future areas and hold back financial reserves so that when it sees a market about to break, it can buy in and ride the wave instead of building to it like NVIDIA did. IBM needs to reinvest in marketing so it can at least get the credit it’s earned correctly.

Long-term, NVIDIA needs to realize that if Huang steps down, its ability to execute will be crippled, and it will need to shift to a fast-follower model. If it doesn’t, it’ll find itself in the future where IBM is now. And unless IBM changes strategies, it’s likely to miss the quantum wave as well, even though; currently, it’s ahead of everyone else. According to Huang, AI is already doing many of the things we believed would require quantum computing (except security), suggesting there is a risk that the eventual quantum computing market may be far smaller than we’ve anticipated.

By the way, I was once a lead analyst in Market and Business Intelligence in IBM, and I likely would have reported this same thing internally if I still worked there. I loved that job.

Scroll to Top