Tech giants from Nvidia, OpenAI, Microsoft, and Intel shared bold Cisco AI Summit takeaways on AI’s transformation of software, infrastructure, security, and society. Key insights inside. Credit: Cisco
For a few intense days in Silicon Valley, the people inventing the AI economy gathered in one room.
Cisco’s second annual AI Summit brought together an unusually concentrated lineup of technologists and executives—Jensen Huang of Nvidia, Sam Altman of OpenAI, Kevin Scott of Microsoft, Dr. Fei-Fei Li of World Labs, Marc Andreessen, leaders from Intel, AWS, Google, Anthropic, Box, and senior U.S. national security officials.
The setting was corporate. The implications were global.
Over 120 chief executives and technology leaders attended in person. More than 18 million watched online. The conversations ranged from software development and chips to geopolitics and the future of work. The collective message was unmistakable: AI is no longer a tool. It is becoming the operating system of business.
Here are the summit’s most important takeaways.
Software Isn’t Dying—It’s Being Rewritten
No topic drew more attention than the future of software development.
There is a growing anxiety that AI will make traditional coding irrelevant. Jensen Huang, Nvidia’s Chief Executive Officer, dismissed that idea bluntly.
“Remember what software is. Software is a tool,” Huang said. The suggestion that AI will end the software industry, he added, “is the most illogical thing in the world.”
His argument was philosophical as much as technical: innovation cannot be tightly controlled.
“If you want your company to succeed, you can’t control it. You want to influence it,” Huang said. “Innovation is not always in control.”
Rather than obsessing over return on investment, Huang urged companies to focus on what truly matters. “What is the essence of my company? What’s the most impactful work that we do?”
At Nvidia, he said, the company lets experimentation run wild. “We just let 1,000 flowers bloom.”
The takeaway: AI will not replace software—it will change how software is conceived and built.
More Activity Does Not Equal More Progress
Microsoft’s Chief Technology Officer, Kevin Scott, offered a cautionary counterweight to the excitement.
Coding agents can generate enormous amounts of code, he noted. That doesn’t mean the output is useful.
“You can produce a lot of code with these coding agents right now,” Scott said, “but there’s nothing to say that it’s good code.”
The software world is in a frenzy, he argued, with even senior engineers struggling to keep pace. The danger is confusing motion with advancement.
“People need to really make sure that they’re not getting confused between activity and progress,” Scott warned.
In an age of AI abundance, judgment matters more than ever.
Software May Soon Be Built for AI First
Sam Altman, Chief Executive Officer of OpenAI, pushed the conversation further.
What if software itself is no longer designed primarily for people?
“Does that change the architecture of the software itself, where you’re going to optimize it for agents more so than humans?” Altman asked. “It fundamentally changes how you build software.”
Altman imagines a future of “always-on computing,” where AI systems continuously observe and assist—listening to meetings, monitoring workflows, executing tasks.
But he acknowledged the hurdles are not just technical. They are legal, ethical and structural.
“Our permissioning system, and how we think about what an AI gets to see and do stuff with… is not really meant for that,” he said.
Cisco’s President Jeetu Patel offered a real-world example: soon, he said, 100% of Cisco’s AI Defense package will be written by OpenAI’s Codex platform.
“We are moving toward what Altman called ‘full AI companies,’” Patel said.
AI Is Hitting a Hard Hardware Wall
While software discussions dominated, Intel’s Chief Executive Officer, Lip-Bu Tan, brought the room back to physics.
“The biggest challenge for a lot of my customers is memory,” Tan said. “And there’s no relief as far as I know.”
The industry is facing its worst memory shortage in history. Prices are rising sharply. Demand for AI compute is exploding faster than supply chains can adjust.
If AI feels infinite in theory, it is still painfully finite in practice.
Security Must Be Reinvented for AI
Altman described security as one of AI’s least-understood problems.
“How are we going to balance security and data access versus the utility of all of these models?” he asked. “It feels to me like there is a new kind of security or data access paradigm that needs to be invented.”
Patel agreed, calling AI the first technology where security is not optional.
“This is the first time that security is actually becoming a prerequisite for adoption,” he said.
In the past, companies traded productivity for protection. With AI, they must achieve both.
The Next Frontier Is Spatial Intelligence
Dr. Fei-Fei Li, Founder of World Labs, argued that language models are only the beginning.
The next great leap will be spatial intelligence—the ability of AI to understand and interact with the three-dimensional world.
“The ability to understand, to reason, to interact with, and to navigate the real 3D, 4D physical world is as foundational as language intelligence,” she said.
Robotics, healthcare, and immersive computing will be reshaped by this shift.
AI Is an Economic Necessity
Marc Andreessen framed AI not as a luxury but as an economic imperative.
Productivity growth has stagnated for decades, he said. AI is the engine that can restart it.
“We’re in a baby and the bathwater moment right now,” Andreessen said. As AI makes software cheaper to build, entire industries will be reorganized.
Kevin Scott echoed that sentiment, linking AI to demographics.
With aging populations and labor shortages, he said, AI may be the only tool capable of sustaining modern standards of living.
Taste Will Matter More Than Ever
Dylan Field, Chief Executive Officer of Figma, predicted that AI will flood the world with options. The scarce resource will be human discernment.
“Instead of thinking agents will adapt to how we work, we will have to adapt to how agents work,” Field said.
In a future of infinite creative possibilities, taste becomes the ultimate differentiator.
From Experiments to Production
Matt Garman of Amazon Web Services argued that many AI projects fail for a simple reason: unclear goals.
The era of pilots is ending. Companies must move from “trying AI” to running on AI.
The winners will be those who turn experimentation into operational infrastructure.
AI Turns Software Into Living Systems
Mike Krieger of Anthropic described a profound shift: software is no longer static.
“Software is now a living, breathing system,” he said—non-deterministic, adaptive, sometimes unpredictable.
The challenge for enterprises is learning to manage systems that behave more like organisms than machines.
Science Will Be Transformed Next
Kevin Weil of OpenAI made perhaps the boldest prediction of all.
“If 2025 was the year AI transformed coding, 2026 will be the year AI transforms science,” he said.
AI could compress decades of research into years, acting as a “metal detector for hypotheses.”
12. Infrastructure, Not Models, Will Decide Winners
Google’s Amin Vahdat reminded the audience that AI breakthroughs depend on hardware and networks.
The pace of model innovation is so fast that chip design cycles must shrink from years to months.
In AI, he said, infrastructure is destiny.
Context Is the New Currency
Aaron Levie, Chief Executive Officer of Box, argued that the defining challenge ahead is context.
Enterprises will soon have hundreds or thousands of AI agents. Feeding them accurate, secure information will be the central problem.
“Use agents to be more ambitious,” Levie said—not just to automate small tasks but to rethink what organizations can accomplish.
Geopolitics Is Now Part of the AI Stack
National security officials Brett McGurk and Anne Neuberger closed with a sobering reminder: AI is a geopolitical technology.
Cyber defense is increasingly a battle between machines. Over-regulation in one country could become a strategic disadvantage.
The Final Lesson: Leadership Is the Real Transformation
Francine Katsoudas, Cisco’s Chief People Officer, summarized the human dimension.
AI adoption, she said, is ultimately a leadership challenge. Nearly 80% of technology roles will require AI skills. Companies must invest as much in people as in algorithms.
The Bigger Picture
The Cisco AI Summit was not a showcase of gadgets. It was a glimpse into a new economic order.
AI is shifting from assistant to agent, from experiment to infrastructure, from novelty to necessity.
The companies that thrive will not be those that adopt AI fastest, but those that govern it best—balancing innovation with security, ambition with responsibility.
As the summit concluded, one reality felt inescapable:
The future of business is being written in code and models.
And the writing has already begun.
