Why the AI bubble is good for business

Recent events have brought about a flurry of discussion about the AI bubble. Around mid-August, OpenAI CEO Sam Altman said that while AI is incredibly important, the current investor excitement is overblown, and “someone’s gonna get burned” by the frenzy of investment.

Around a week later, we saw MIT release its report, “The GenAI Divide: State of AI in Business 2025,” which stated that 95% of corporate gen AI pilot projects were failing to deliver measurable returns. What’s undeniable is we’re entering an AI era of exponential, unpredictable change.

I’ve been warning of the AI hype since early 2024, when AI was in more of a gaslamp moment. This year, we also looked at the hype around agentic AI and discussed the AI reset that needed to occur.

While tech vendors work hard to control the narrative that there’s no AI bubble, here’s why those outside it, such as most enterprise CIOs, can proceed with rational exuberance, and why AI is good for business, bubble or not.

Parallels with the dot-com bubble

While investor excitement is overblown, there’s a powerful aspect to the AI vendor landscape that will endure post-bubble. As a CIO or CAIO, you can take advantage of this by discarding the fluff and looking for the core AI vendors that’ll help enable your organization. The MIT report cited that purchasing AI tools from specialized vendors and building partnerships succeeded about 67% of the time, and internal builds only one-third as often.

Therefore, by working with specialized vendors, you’ll be able to exponentially magnify your success rate. And while your internal teams may want to do a lot of work in-house, it’ll be important to recognize not invented here syndrome within these teams, and to encourage them to explore and utilize external tools wherever it makes sense.  

Most AI strategies need a reset for GRC

Bubble or not, most enterprise AI strategies need a reset to ensure adequate governance, risk, and compliance are built in from the start, otherwise it’s just one step forward and two steps back. This is especially true as organizations move from specific AI/ML projects, to gen AI, and then to agentic AI. The risk profile for the organization increases with every step as AI finds its way deeper into core enterprise applications and workflows, and becomes more automated and autonomous.

As I discussed last month about who should manage AI agents, the risk is particularly high the more we automate and orchestrate with agentic AI. Spending time on AI orchestration and governance right now, and thinking about how AI agents will impact your technology platforms, centers of excellence, and your governance approach, will enable your organization to scale more rapidly, and advance your AI maturity in the years ahead.

Enterprise AI projects need IT after all

With the hype around AI disrupting the consulting and advisory industry, and even replacing entire IT teams with vibe coding, one might think consulting, systems integration services, and internal IT teams are no longer required in an AI world. But for anyone who’s tried some of the AI-enabled website builder applications, despite looking great on the surface with lines of code flashing by on the screen, the results are often far less impressive.

As a CIO or CAIO, this means your IT team is more important than ever. You can leverage their expertise to utilize AI where it makes sense for productivity gains and to incorporate the necessary enhancements to ensure AI projects are sufficiently robust and reliable. It’s also okay for AI pilots to fail as part of the learning curve, which can help inform and progress the AI technical stack suitable to your organization.

Pointing the blame of AI failure

AI vendors are quick to attribute the cause of failed AI pilots and projects to inadequate vision, leadership, and change management at companies deploying AI themselves. They cite the importance of looking for new business models, products, and services that can be enabled via AI, stressing the importance of not simply applying AI to an existing flawed or inefficient process, or just looking for cost savings.

While this is true, it’s only half the story. In many cases, the issues are attributable to the AI technology as well. The well-known issues of accuracy, hallucinations, privacy, safety, and security, for instance, have meant that gen AI has now slipped into Gartner’s Trough of Disillusionment.

For CIOs and CAIOs, however, this is actually good news, since there’s now nowhere to go but up in terms of future improvements and productivity gains. Be sure to review your organization’s vision, leadership, and change management around AI implementations, but keep an eye on the technology as well and retain your due diligence.  

For anyone who lived through the dot-com era, we’ve all seen this story before. We tend to overestimate the potential of emerging technology in the near-term, while underestimating its transformative potential in the long-term. AI technologies are transformational for sure, and there’s going to be continuous innovation in the market to make them even more robust and reliable over time. Composite AI, for example, is already being applied to utilize multiple AI technologies so the sum is greater than the parts, and they can address the limitations of individual elements and techniques.

So whether there’s an AI bubble or not is irrelevant. Like any bubble, however, it’s important to focus on the kernel of truth, maintain careful due diligence, utilize the wisdom of your partners and teams, and continually test, learn from, and monitor the technology as it evolves. This isn’t about taking a step back, it’s about continually innovating with rational exuberance.