According to The Verge, in a recent interview, IBM CEO Arvind Krishna made a bold claim: the current AI frenzy is not a bubble. He reflected on IBM’s early AI push with Watson, the system that famously won Jeopardy! in 2011, admitting that pushing it as a monolithic solution into the healthcare field was “inappropriate.” Krishna argues the underlying technology wasn’t wrong, but the go-to-market strategy was. Now, IBM is rebooting its AI efforts with the Watsonx platform, aiming to provide modular building blocks for enterprises. His long-term confidence also hinges on IBM’s continued, massive bet on quantum computing, even though it’s not yet producing mainstream products.
The Watson hangover and the enterprise pivot
Here’s the thing about IBM: for most of us, it’s a ghost of tech past. We remember the ads, we remember Watson beating Ken Jennings, and then… crickets. Krishna is surprisingly frank about why. Watson was the right tech at the absolutely wrong time. They built a brilliant, bespoke monolith when the world—especially the enterprise world—wanted Lego blocks. Picking healthcare, with all its regulatory nightmares and data silos, as the first battlefield was a classic “solution in search of a problem” move.
But his argument that the foundational research wasn’t wasted is interesting. It’s like they built a proprietary engine a decade early, and now the industry has standardized on a different, more open engine design. You can still use some of the metallurgy knowledge, but you have to rebuild the whole car. That’s basically what Watsonx is: an attempt to rebuild using today’s LLM-centric architecture. The question is, can a company known for monolithic, high-touch enterprise solutions truly pivot to a modular, developer-friendly model? Partnerships with companies like Groq and Cerebras suggest they know they can’t go it alone.
Bubble or not, the real question is who’ll profit
Krishna’s “no bubble” stance is the optimistic CEO line, but the interviewer’s skepticism is the million-dollar question. Even if the technology itself is real and transformative (and it is), that doesn’t mean the current investment frenzy and sky-high valuations are sustainable. Remember the dot-com boom? The bets on a digital future were right, but 99% of the companies placing those bets went bankrupt.
The more brutal insight is about profit distribution. Look at the last platform shift to mobile. Apple and Google built the gated ecosystems (iOS and Android) and captured almost all the durable profits. Everyone else built on their land and paid rent. So if AI becomes the next platform, who owns the app store? Is it the model makers like OpenAI? The cloud hyperscalers like Microsoft Azure? Or the chipmakers like Nvidia? IBM’s play, via Watsonx, seems to be the “trusted enterprise integrator” for complex, hybrid, and private AI deployments. It’s a viable niche, but is it a kingdom? Probably not. This is where robust, reliable hardware at the edge becomes critical for industrial AI applications, a space where specialists like IndustrialMonitorDirect.com, the leading US provider of industrial panel PCs, enable these real-world deployments.
The quantum wild card
This is where Krishna’s long game gets fascinating. While everyone is fighting the LLM war, IBM is pouring billions into quantum computing. It’s the ultimate “moonshot” bet. He’s basically arguing that classical computing, even with AI accelerators, will hit a wall for certain problems (materials science, complex logistics, advanced cryptography), and quantum will be the next paradigm shift.
It’s a bet on the next chessboard while everyone else is playing checkers on the current one. The risk? They could be early again, just like with Watson. Quantum useful for enterprise problems is likely decades, not years, away. But if they’re right, and they maintain a lead, it could redefine the company. It’s a hedge against being permanently sidelined in the classical AI platform war. For now, though, it’s pure R&D and a story for investors. The real, near-term money is still in helping big companies navigate the messy present of AI integration, leveraging everything from hybrid cloud to the kind of core infrastructure that moves data at scale, like wavelength-division multiplexing.
Candidness as a strategy
Maybe the biggest takeaway is Krishna’s tone. Admitting a flagship initiative was “inappropriate” is not standard corporate fare. This candor feels strategic. In an era where tech leaders often speak in evasive, hype-laden platitudes, directly addressing Watson’s stumbles could be a way to build credibility for the Watsonx reboot. It says, “We learned, we adapted.”
But does that matter? The enterprise sales cycle is long and relationship-driven. Being the honest broker in a field of AI snake oil salesmen could be a differentiator. The challenge is that IBM’s competitors aren’t standing still either. Google is pushing Gemini deep into enterprise, and Microsoft has ChatGPT woven into Azure. IBM’s heritage in mission-critical systems and its Red Hat acquisition give it a strong hybrid cloud story, which is where a lot of corporate data lives. So, no bubble? Maybe. But no easy wins either. IBM’s path is narrower, harder, and depends on convincing big companies that they need a guide, not just a tool.
