AI Check Point - Nadella, Zuckerberg, Huang

Introduction

In recent days, the CEOs of Microsoft, Meta, and Nvidia shared their thoughts on the future of AI. These are three conversations I’ve watched that I found incredibly insightful — and worth revisiting.

Satya Nadella & Mark Zuckerberg: Engineering and Economic Transformation

The first is an interview where Mark Zuckerberg speaks with Satya Nadella. They cover a wide range of topics, including the "hyperacceleration" of Moore’s Law, with exponential improvements every 6–12 months across chips, model architectures, caching, software, and more. They also discuss multimodal applications orchestrated by agents, and emerging protocols like MCP and A2A to standardize these interactions — among many other ideas. What I’d like to highlight from this conversation are two points:

• Reconceptualizing the engineer’s role: According to Satya, engineers will become agent leaders — think of a tech lead surrounded by "developer agents" executing specific tasks.

• Economic growth: For Microsoft’s CEO, the real promise of AI doesn’t lie in the current hype, but in its ability to drive economic growth. To do so, AI must transform processes and workflows across sectors like healthcare, retail, and finance. He believes it's not just a tech shift — it's a management and cultural one, much like electricity, which only unlocked value once factories were redesigned around it.

📅 April 29, 2025
🔗 Welcome to LlamaCon 2025 - Closing Session!

Mark Zuckerberg: Infrastructure and Code Generation by AI

The second interview features Mark Zuckerberg reflecting on Llama 4, Meta’s latest model; the role of open-source models; AI monetization; governance, regulation, and values; and the use of AI as a cultural engine, among other themes. But there are two aspects I’d like to call out:

• Infrastructure: Zuckerberg notes that physical infrastructure is an inevitable bottleneck. Building gigawatt-scale data centers, securing networks, permits, energy — it all takes time. Even when AI is ready, the human, regulatory, and logistical environment must evolve in parallel.

• AI programming agents: Meta’s CEO predicts that within 12 to 18 months, most of the code written to develop models like Llama will be generated by AI — not via autocomplete, but through agents that understand goals, run tests, refactor code, and improve it.

📅 April 29, 2025
🔗 Mark Zuckerberg – AI Will Write Most Meta Code in 18 Months

Jensen Huang: The Intelligence Infrastructure and the Rise of Physical AI

The last conversation is an interview with Jensen Huang. Personally, I find his interviews always fascinating — and this one was no exception. Topics include AI as the new industrial revolution, token factories, chips, systems and ecosystems, geopolitical competition, and global standards. From all that, here are two concepts I want to highlight:

• The intelligence infrastructure: Huang argues that we are entering the era of intelligence infrastructure — a layer that cuts across all economic sectors: healthcare, finance, logistics, manufacturing, education, entertainment. He also claims AI factories will become pillars of national competitiveness.

• Physical AI and robotics: Huang envisions factories where robots build robots, which in turn build more robots — all coordinated by AI agents. He believes this autonomous production ecosystem, combined with sensors, computer vision, and action models, will completely transform manufacturing and logistics, creating trillion-dollar markets.

📅 May 6, 2025
🔗 A Conversation with Nvidia CEO Jensen Huang | Global Conference 2025

Common Threads Across the Three Leaders

Despite the different focuses, there are key points on which all three leaders converge. I’d like to highlight three of them:

• The economic growth potential of AI and its impact on the job market.
• The need for infrastructure — especially energy and data — to power and scale models.
• The role of governments and the urgency to create effective regulatory frameworks around AI use.

Three interviews, three CEOs, three visions… and a checkpoint to look back on when the future becomes the present.