A single statistic looms over the artificial intelligence (AI) sector: 95%. This figure comes from a prominent, though some say exaggerated, MIT study from August 2025. When surveyed about six months later in Davos, Switzerland, a significant number of CEOs—56%—reported receiving “nothing” from their AI initiatives.
According to Nvidia CEO Jensen Huang, the remedy involves peace, love, understanding, and effective parenting. The leader of the $4 trillion company came to the Cisco AI Summit with a message that echoed 1960s counterculture and contemporary child-rearing more than Wall Street discipline: “Let a thousand flowers bloom.”
In a conversation with Cisco’s Chuck Robbins, Huang tackled the dilemma corporate leaders face, pressured to implement AI but worried about the absence of quick, measurable returns. When Robbins inquired about the initial steps for a business, Huang rejected an immediate focus on financial calculations.
“I get questions like … ROI,” Huang stated. “I wouldn’t go there”.
He instead promoted a mindset of plentiful, unstructured experimentation, directly likening business innovation to parenting. Insisting on proof of profitability before letting an engineer test a new AI tool, he contended, is as restrictive as requiring a child to draft a business plan for a pastime.
“I want the same thing for my company that I want for my kids: go explore life,” Huang elaborated. He added that when children express interest in trying something, the answer should be yes. At home, we never pose questions such as: What’s the return on investment? How will this lead to monetary gain? Can you prove its value? “We never do that at home. But we do it at work.”
Innovation needs therapy, not control
Huang acknowledged that this method demands leaders give up a level of oversight that may be unsettling, but he insisted the resulting creativity and innovation justify it. “The number of different AI projects in our company is, it’s out of control and it’s great,” he said, noting that breakthrough ideas often don’t emerge under tight control. “If you want to be in control, first of all, you’ve got to seek therapy. But second, it’s an illusion. You’re not in control. If you want your company to succeed, you can’t control it.
Huang asserted that successful leaders should aim to influence their organizations, not control them. The reasoning for allowing “a thousand flowers bloom” is to manage risk by diversifying efforts. This strategy “makes for a messy garden,” he admitted, but it avoids the mistake of concentrating all resources—”putting all your wood behind one arrow”—prematurely in a tech transition where the best tools remain unclear.
Lift the hood
While supporting a flexible stance on ROI, Huang was firm on the need for hands-on, practical knowledge. He advised executives against depending exclusively on cloud services or off-the-shelf solutions.
Computers are ubiquitous now, he observed, but building one yourself yields deeper insight, similar to how a conscientious car owner wouldn’t only use ride-sharing but would also inspect their own engine. “Lift the hood, change the oil, understand all the components,” he urged. “Build something. You might discover you’re actually insanely good at it. You might discover that you need that skill.”
He emphasized that since AI is crucial for the future, firms need to develop some on-site infrastructure to fully grasp how the “components” function. This connects to data security and what Huang identifies as the most precious intellectual property: the questions. “The most valuable IP to me is not my answers… they’re my questions,” Huang said, explaining that answers are common, but insightful questions are unique.
He highlighted the approach at the accounting firm KPMG in Orlando, Florida, which was implementing its AI training program, starting with interns and expanding company-wide. Their method for coaching staff to use AI was “Think, prompt, check,” underscoring that the initial and final steps should not be overlooked.
From explicit to implicit
The need for this widespread testing, Huang said, arises from a core “reinvention of computing.” The field is transitioning from “explicit programming”—crafting precise code—to “implicit programming,” where people declare their goal, and the AI determines the method.
In this emerging paradigm, “typing is a commodity,” Huang pointed out. Real value resides in the specialized knowledge needed to direct the AI. “You now tell the computer what your intent is, and it goes off and figures out how to solve your problem.”
Huang concluded by inverting the common ethical discussion of “humans in the loop.” The objective, he declared, should be “AI in the loop.” Embedding AI into all operations allows companies to record the “life experience” of their workforce, converting everyday tasks into lasting corporate knowledge. Essentially, they will enable a thousand flowers to bloom, provided they foster the appropriate curiosity, ask the right questions, and receive the necessary top-down encouragement to explore without constraints.
For this story, journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.