Listen to an AI audio recording of this newsletter

The energy in ​South by Southwest​ (SXSW) is hard to describe if you haven’t experienced it.

Every March, Austin, Texas, turns into a living lab for ideas. Founders, technologists, filmmakers, investors, and musicians spill out of panel rooms into coffee shops and sidewalks, continuing conversations that started on stage. New technologies get introduced. Big bets get debated. And occasionally, you can feel a shift beginning before the rest of the world notices.

This year I presented at SXSW alongside ​Shantnu Sharma​, ​Nadia V. Gil​ and ​Roopa Unnikrishnan​ to explore a question that has been occupying a lot of our thinking lately.

Shantnu brings a deep perspective from the semiconductor ecosystem, where the physical limits of computing, such as chips, packaging, and supply chains, are becoming central strategic questions.

Nadia operates at the intersection of AI, data center infrastructure, cloud, cybersecurity, and corporate strategy, leading work across partnerships, M&A, and large-scale technology integration.

Roopa brings her expertise as a global business leader and author who advises companies on strategy, leadership, and innovation in rapidly changing markets.

Together, we explored what happens when AI stops living only on our screens and starts operating in the physical world.

I sometimes describe it this way:

AI is about to get hands.

In preparation for the session, Shantnu and I recently sat down to record an upcoming episode of the ​Outthinker CSO Podcast​ to unpack the ideas we would be discussing on stage. One theme stood out: many of the most important developments in AI aren’t happening in chat windows or coding environments. They’re happening in factories, supply chains, and infrastructure.

The AI We Know Today

Over the past two years, AI has moved into daily life at remarkable speed. It drafts emails, summarizes documents, writes code, and recommends what to watch and what to buy.

It adjusts thermostats, predicts traffic patterns, manages customer service chats, and increasingly helps run digital workflows inside companies.

For most people, AI shows up through text, voice, and screens.

But that is only the first phase.

The next phase will be far more tangible. Instead of quietly suggesting what we should do from our screens, AI will begin acting directly inside factories, warehouses, energy systems, and supply chains.

And that shift changes the entire strategic landscape.

The Physical Reality Behind AI

One of the themes Shantnu and I kept returning to is how much the AI story is constrained by very physical realities.

When people talk about AI today, the conversation often focuses on models and algorithms.

But beneath those models sits a stack of infrastructure:

• Semiconductor fabrication plants
• Advanced chip packaging facilities
• Massive data centers
• Electrical grids
• Cooling systems
• Critical minerals

Every layer introduces constraints.

Take advanced packaging. For decades, the semiconductor industry improved performance by shrinking transistors and packing more of them onto a single chip. But we are approaching physical limits.

Instead of one giant chip, the industry is shifting toward systems built from multiple specialized chips connected together.

The performance gains increasingly come from how chips are combined — not just how they are designed.

Packaging used to be an afterthought.

Now it’s becoming one of the most strategic battlegrounds in computing.

The Bottleneck

There is another constraint that comes up in almost every serious conversation about AI infrastructure.

It’s not chips.

It’s power.

Modern AI data centers consume staggering amounts of electricity — sometimes enough to rival a small city. And unlike software, energy infrastructure cannot scale instantly. You cannot simply wish a data center into existence.

You need land, permits, cooling systems, grid connections, transmission lines, and enormous capital.

Which means the growth of AI is governed not just by technological progress, but by the speed at which the physical world can keep up.

As Shantnu often explains, the AI supply chain tightens sequentially:

• First packaging.
• Then memory.
• Then computing.
• Then energy.

Each layer exposes the next constraint.

Where Things Get Interesting

All of that might sound like a story about limits. But those constraints are also pointing to something much bigger.

Shantnu calls it the convergence frontier: the moment when AI bridges the gap between the digital and physical worlds.

Imagine a semiconductor factory. Today, vision systems detect defects on wafers.

Now imagine AI tracing those defects back to the process conditions that caused them, like temperature, chemical concentrations, or equipment calibration.

Then running thousands of simulations to determine which adjustments improve yield. Then implementing those changes in the production line.

At that point, the AI is no longer just monitoring the factory. It is actively shaping how the factory operates.

Every production cycle becomes a learning loop.

The Next Gen Warehouse

We see similar shifts emerging in logistics.

During our podcast conversation, Shantnu described visiting a distribution center where a human picker raced a robot to retrieve an item.

The robot won easily.

But the interesting part is that it didn’t win because of speed. It had already anticipated the next several picks based on order patterns and inventory positions.

It wasn’t simply reacting faster. It was predicting the future better.

Now imagine that intelligence coordinating the entire warehouse:

• Anticipating inbound shipments
• Repositioning inventory dynamically
• Routing robots across the floor
• Adjusting staffing schedules in real time
• Responding instantly to disruptions

The warehouse stops being just a building full of equipment.

It becomes a living control system.

Much More Than Software

When AI crosses into the physical world, the scale of opportunity changes dramatically. Today, many people frame AI as a software market.

But if intelligence becomes embedded in manufacturing, logistics, energy, and healthcare, the addressable market expands far beyond software spending.

It begins touching labor, capital equipment, and operating systems across the entire economy. Global GDP is roughly $120 trillion.

Even modest improvements in productivity across physical industries represent enormous value. That’s why companies pursuing this frontier are not thinking in incremental growth. They’re thinking about becoming the platform layer that helps the physical economy run more efficiently.

The Conversation in Austin

That’s the conversation Shantnu, Nadia, Roopa, and I had at SXSW.

Because the most important AI questions over the next decade may not be about prompts or models. They may be about factories, power grids, robotics, and supply chains.

In other words: Not just how machines think. But how they act.

And if that future unfolds the way many of us suspect, the story of AI will not just be about software.

It will be about the moment intelligence finally got hands.

Get ready for the future by visiting ​Outthinker.com​ today.