AI advances faster than corporations can possibly adapt, executives warn

Executives at Semafor’s World Economy Summit highlighted a growing disconnect between AI’s rapid technological advancement and corporate America’s ability to adapt organizationally. While AI is enabling breakthrough innovations like Formula 1 teams using digital twins instead of physical testing, legacy companies are struggling to rewire decades-old business processes quickly enough to keep pace with the technology.

What you should know: AI is delivering on ambitious promises faster than many anticipated, creating practical solutions that were impossible just years ago.

• Sassine Ghazi, CEO of Synopsys, a chip design software company, pointed to Formula 1 as an example where teams now create “digital twins of entire systems” to optimize performance through computer simulations rather than costly physical track testing.
• “From a cost point of view, it’s much cheaper,” Ghazi said. “From an ability to optimize at the system level — it’s possible. None of this was possible without accelerated compute.”

The cultural challenge: Goldman Sachs President John Waldron identified organizational adaptation as the real bottleneck, not the technology itself.

• “The technology will move much faster than firms like ours can adopt it, because we have to go rewire the way we’ve been doing things for decades,” Waldron explained.
• This cultural lag represents a significant challenge for established companies trying to integrate AI into existing workflows and decision-making processes.

The bubble debate: Despite rapid innovation, concerns about an AI investment bubble dominated discussions at the two-day Washington, DC event.

• Waldron pushed back against bubble characterizations, noting that current data center investments are “two, three, four years out, in terms of getting it finished and trying to determine how it all gets used.”
• The timeline suggests the true test of AI’s economic value may still be years away as infrastructure investments come online.