This week, we witnessed something rare: the builders of the AI economy, all in one room, all at once, defining what comes next.
Cisco AI Summit wasn’t just an event—it was a moment. Hosted by Chuck Robbins and Jeetu Patel, it brought together the visionaries, technologists, and leaders shaping the trillion-dollar AI economy. Over 120+ CXO’s from leading companies across industries filled the room. And beyond those walls? Over 18 million people and counting joined via livestream, making this one of the most-watched technology events in recent history.
The energy was electric. The conversations were unfiltered. And the insights? They will shape how we think about AI for years to come.
What made this summit different was the caliber of speakers and the honesty of the dialogue. This wasn’t about hype—it was about reality. These are the people building the models, designing the chips, securing the infrastructure, and reimagining what’s possible. They came to share not just what AI can do, but what it will do, and what we must do to shape it responsibly.
Here’s what unfolded.
Jensen Huang (NVIDIA): The world of abundance
Jensen Huang reframed AI as more than a new capability—it’s a fundamental reinvention of the 60-year-old computing stack. We’re moving from explicit programming (writing code) to implicit programming (defining intent). The result? A world where intelligence becomes abundant and engineers can treat compute constraints as effectively infinite.
Huang argued that coding itself is becoming a commodity. What matters now is domain expertise and understanding customer problems. Every company will eventually have “AI in the loop” to capture life experiences and turn them into intellectual property. And companies that deal in electrons rather than atoms will see their value explode.
“When I think about a problem these days, I just assume my technology, my tool, my instrument, my spaceship is infinitely vast… If you’re not applying that sensibility, you’re doing it wrong.”
Sam Altman (OpenAI): From tool to teammate
Sam Altman opened with a vision that felt both thrilling and inevitable: AI is no longer just a tool we use—it’s becoming a collaborator we work alongside. AI has evolved from answering discrete questions to taking full control of a computer to execute complex, multi-step tasks. The implications are staggering.
Altman’s most striking insight? The biggest constraints aren’t technical. They’re architectural. We need new security paradigms for data access, and we need to rewrite software so it’s equally usable by humans and AI agents. The “upper limit” of this technology, he suggested, is companies run entirely by AI. By the end of 2026, he predicts a 10x improvement in the problems AI can solve.
“I think the companies that are not set up to be able to adopt AI coworkers very quickly, will be at a huge disadvantage.”
Dr. Fei-Fei Li (World Labs): Spatial intelligence as the next frontier
Dr. Fei-Fei Li introduced a compelling shift: the next phase is spatial intelligence—AI’s ability to understand and navigate the 3D physical world. Perception, she argued, is more foundational than language. This opens possibilities from revolutionizing psychiatric care through immersive environments to enabling AI to interact with the physical world in transformative ways. Dr. Li was candid: building generalized robots will be a decades-long journey, much like self-driving cars. But the work begins now.
“The ability to understand, to reason, to interact with, and to navigate the real 3D, 4D physical world is the foundation—as foundational as language intelligence.”
Marc Andreessen: AI as the productivity engine
Marc Andreessen brought historical perspective: economic growth has stagnated since the 1970s, and AI is our primary engine to restart massive productivity gains. He introduced the metaphor of open source AI as an “asteroid strike”—it eliminates profit margins for proprietary labs but turbo-boosts the rest of the industry.
“We’re in a baby and the bathwater moment right now… software will be cheaper to build, which means you’re going to get more SaaS.”
Kevin Scott (Microsoft): The demographic imperative
Kevin Scott from Microsoft framed AI as a necessity, not a choice. With global population decline and labor shortages, AI is the only technological intervention that can maintain our quality of life. In coding, the bottleneck has shifted from creation to review and taste. Computer science education will evolve back to algorithmic thinking, and as inference becomes cheaper, demand for compute will never go down—humans will simply find more ambitious ways to use it.
“I hope that we can resist the temptation to make the whole narrative about AI… and make it more about: what does society really need from this technology?”
Dylan Field (Figma): Taste is the new scarcity
Dylan Field of Figma explored how AI is blurring the lines between designers, product managers, and engineers. While AI can generate infinite options, human taste remains the scarce commodity. The current “prompt box” is primitive—the future is a multiplayer canvas where humans and agents work side-by-side. Field’s most provocative prediction? Within 12 months, designers will likely be able to update existing production codebases directly through design canvases. The challenge? Adapting our workflows to how agents work, not the other way around.
“Instead of thinking agents will adapt to how we work, we will have to adapt to how agents work.”
Matt Garman (AWS): From experimentation to production
Matt Garman from AWS focused on the transition from AI pilots to AI at scale. Many AI projects fail because companies don’t define success criteria upfront. Success at scale requires choice, security, and seamless integration. AWS is betting that inference will be built into every application and that personalized AI experiences will emerge from longitudinal data. The companies that win will move from “let’s try AI” to “AI is how we operate.”
Mike Krieger (Anthropic): Software as a living system
Mike Krieger of Anthropic described AI as transforming software from a static artifact into a living, breathing system. Instead of filling out forms, users will simply have conversations. AI will learn preferences through observation rather than configuration. One of the most practical applications? Paying down technical debt in large enterprises.
“Software is now a living, breathing system with this non-deterministic, wonderful, but also sometimes infuriating engine at its core.”
Kevin Weil (OpenAI): 2026 is the year of science
Kevin Weil from OpenAI made a bold declaration: if 2025 was the year AI transformed coding, 2026 will be the year AI transforms science. AI can act as a “metal detector for hypotheses,” allowing scientists to skip failed experiments and compress research cycles. The goal? Achieve 25 years of scientific progress in just five years. Weil envisions robotic labs where AI designs experiments, sends them to robotic arms to execute, and iterates on results in a closed loop.
“There is no excuse not to be creating whatever you can think of.”
Lip-Bu Tan (Intel): The silicon bottleneck
Lip-Bu Tan, CEO of Intel, brought the conversation back to hardware. Every AI breakthrough rests on silicon, and the biggest bottleneck today is high-bandwidth memory. Moving from a product company to a foundry requires a culture shift—earning customer trust through yield and reliability. New materials like glass substrates and artificial diamonds will be necessary to solve thermal and power management challenges.
Tareq Amin (HUMAIN): The power of intent
Tareq Amin delivered one of the summit’s most radical visions: the current lack of enterprise productivity gains stems from layering AI on top of legacy platforms. HUMAIN is creating a “genetic operating system” built from scratch where applications become “second-class citizens.” Instead of navigating separate apps with distinct logins, users simply “declare intent” (e.g., “hire this person,” “pay this invoice”), and a multi-agent orchestration platform executes workflows invisibly.
Brett McGurk & Anne Neuberger: Geopolitics and security
Brett McGurk and Anne Neuberger addressed the intersection of AI, national security, and geopolitics. Cyber defense is now a machine-on-machine fight—human defenders can no longer keep up with AI-driven attacks. The risk of over-regulation is real: if the U.S. constrains AI development while adversaries scale without guardrails, we create a national security vulnerability.
“Partnerships is America’s competitive advantage in the world across every aspect of diplomacy, national security, commerce, and trade.”
Amin Vahdat (Google): Infrastructure decides winners
Amin Vahdat from Google explained that while models get headlines, infrastructure decides winners. Moving away from one-size-fits-all architectures enables 10x gains in power efficiency. The challenge? The current three-year cycle for hardware design needs to compress to three months to keep pace with AI model evolution.
Aaron Levie (Box): The context problem
Aaron Levie of Box highlighted the defining challenge of the next decade: the context problem—feeding context to agents in the most efficient, accurate, and comprehensive way. Levie predicts we’ll see 100x to 1,000x more agents in an enterprise than people, requiring systems of record to act as “traffic cops.” The companies that succeed will use agents not just for small tasks, but to be more ambitious—to do way more as an organization.
“Use agents to be more ambitious and to do way more as an organization—not just the smaller things, but the bigger things.”
Francine Katsoudas: Leadership is the transformation
Francine Katsoudas closed the summit with findings from the AI Workforce Consortium. AI adoption isn’t just a technical transformation—it’s a leadership transformation. The data is clear: 78% of technology roles will require AI skills, necessitating a massive industry-wide commitment to free, accessible learning. The companies that win will invest in their people as aggressively as they invest in their technology.
“The map didn’t end at the boundary of knowledge. It ended at the boundary of risk… The future belongs to those who fearlessly walk with the lions.”
What we learned
Cisco AI Summit delivered on its promise. It wasn’t about what AI can do—it was about what AI will do. The conversations were direct, the insights were profound, and the implications are unmistakable.
AI is moving from tool to infrastructure. From assistant to agent. From experimentation to dependency. The companies that succeed will be those that move fast, but move responsibly. Those that invest not just in models, but in systems, security, and the people who will shape this technology.
At Cisco, we sit at the intersection of connectivity, security, collaboration, and the platforms that make digital work possible. That gives us a unique vantage point into how AI is being operationalized inside enterprises—and where the friction, risk, and opportunity really live.
The future isn’t being built by those who wait. It’s being built by those who act. This week, we saw the builders of the AI economy define what comes next.
Now it’s our turn to build it.
