AI Is Leaving the Screen

When the Tractor Works After Midnight

How Physical AI Is Broadening the Investment Landscape

Key takeaways

• AI is shifting from generating information to performing real work in the physical economy.
• The opportunity extends beyond chipmakers and AI platforms to power, cooling, security, and everyday businesses that put AI to use.
• Different parts of the AI supply chain may lead at different times as bottlenecks shift.
• Transformative technology does not guarantee profitable investments. Valuation and discipline still matter.

In a northern California orchard, a farmer has been testing a tractor that travels between rows of almond and walnut trees without anyone in the cab. Guided by cameras, sensors, and artificial intelligence, it can apply pesticides and fertilizer while he attends to other work or gets some sleep. For a farmer racing to finish before the weather turns, autonomy is not a novelty. It can protect a year's crop at the moment it matters most.

This is where the next chapter of AI begins. It is moving beyond the screen and into farms, factories, hospitals, vehicles, warehouses, and electrical grids, crossing from generating information to doing useful work. As that happens, the investment opportunity may broaden well beyond the companies building AI models and semiconductors.

From answering questions to completing jobs

The first wave of generative AI taught machines to produce text, images, code, and analysis. The emerging wave connects that intelligence to sensors, business systems, and physical equipment.

Retailers show how this works. Walmart uses AI and automation to forecast demand, reposition merchandise, and reduce waste. Imagine a shipment of strawberries delayed by a storm. An intelligent system could identify which stores still need the fruit, find an alternate route, and move it before it spoils. The benefit is less waste for the retailer, fresher food for the customer, and less repetitive work for employees.

Amazon's warehouse robotics offer another example. By combining AI, robotics, and computer vision, its systems bring inventory to workers instead of sending workers hunting through a vast facility. The customer sees a package arrive the next morning. The employee walks fewer miles in a physically demanding shift.

The value of AI will increasingly be measured not by how impressive an answer sounds, but by whether it saves time, prevents waste, reduces risk, or helps someone finish a difficult job.

Intelligence is moving closer to us

Not every AI task needs a round trip to a giant data center. Some of the work can happen directly on a phone, vehicle, camera, or piece of industrial equipment. This is called edge AI, and it can deliver faster responses, better privacy, and reliability when connectivity is poor.

Picture a grandmother speaking Spanish on a video call with a grandchild who is more comfortable in English. Apple's Live Translation can translate their conversation as they speak, using models that run on the device. The technology fades into the background, and two people simply understand each other.

The industrial version is less personal but just as meaningful. A camera on a production line might spot a hairline crack in an automotive part and halt the process before thousands of defective units are made. Siemens has been expanding its Industrial Copilot to help engineers generate automation code and diagnose malfunctioning lines. In that moment, AI is not writing an essay. It is helping a skilled employee get a factory moving again.

Machines that can see, decide, and move

Agriculture may become one of the clearest demonstrations of AI in the physical economy. John Deere has introduced autonomous equipment for farms, orchards, and construction sites that uses cameras and sensors to navigate and operate attachments. Spraying often must happen in a narrow window of calm wind and available labor, and an autonomous tractor can work through the night while being monitored remotely.

Autonomous transportation follows a similar path. Waymo now provides hundreds of thousands of paid driverless rides each week in several U.S. cities. For someone who can no longer drive, that can mean getting to a medical appointment without asking a family member. For a nurse finishing a night shift, it can mean a safe ride home.

Behind every autonomous machine sits a broader system: cameras, processors, mapping, communications networks, charging, maintenance, insurance, and manufacturing capacity. What looks like "an AI driver" is the work of many industries.

Healthcare may be where AI becomes most personal

A physician reviewing a medical image is sometimes searching for a change too small for the human eye to catch. AI can act as an extra set of eyes, highlighting an unusual area and directing the physician's attention to it. The FDA has authorized well over a thousand AI-enabled medical devices, including tools that help detect diabetic eye disease, support skin-cancer diagnosis, and automate insulin dosing from continuous glucose monitors.

Consider a patient with diabetes who believes her eyesight is fine. A retinal photograph at a community clinic could reveal early damage before she notices symptoms, and AI could flag the image for professional review. The meaningful outcome is not that an algorithm recognized a pattern, but that she may have a better chance of preserving her sight.

Healthcare also shows why trust matters. A bad movie recommendation costs little. A flawed diagnostic tool costs a great deal. Adoption therefore favors companies that can demonstrate clinical reliability, including device makers, imaging firms, secure data platforms, and specialized chipmakers.

The physical foundation, and the security around it

AI may seem weightless on a screen, but the infrastructure behind it is intensely physical: servers, transformers, switchgear, cooling systems, and fiber networks. The International Energy Agency projects that data-center electricity use could roughly double by 2030. A new facility may draw as much power as a small city, and it cannot operate until the utility connection, transformers, and cooling equipment are in place. If a project must wait two years for a transformer, the transformer, not the AI chip, may be the most valuable link in the chain.

Energy is already shaping decisions far from Silicon Valley. Microsoft's long-term power agreement with Constellation supports the planned restart of Pennsylvania's Three Mile Island Unit 1 as the Crane Clean Energy Center. A technology most people experience through a laptop is influencing whether a nuclear plant returns to service.

Security becomes part of the machinery too. A chatbot that drafts a memo creates one kind of risk. An AI agent that can change bank details or send payments creates another. Microsoft's Security Copilot includes agents that triage suspicious-email alerts so analysts can focus on the few that matter. As AI gains permission to act, identity, access, and trust become increasingly valuable.

The opportunity is broader than the obvious winners

One way to think about it is to follow an AI application from idea to real-world result.

Building the intelligence. Semiconductor designers, memory makers, cloud platforms, model developers, and specialized software providers. An autonomous tractor needs processors that can interpret multiple camera feeds instantly, so it recognizes a person or fallen branch in time to stop.

Supplying the foundation. Utilities, electrical-equipment makers, data-center operators, cooling specialists, networking firms, construction companies, and cybersecurity providers. A hospital can't deploy a new imaging system without secure storage, reliable networks, and trained technicians.

Applying AI to existing businesses. A retailer that cuts spoilage, a manufacturer that catches defects, or a utility that spots a failing transformer before it causes an outage may create real value without ever being called an "AI company."

What this means for investors

A transformation this broad is unlikely to reward every industry at the same time. In one phase, the constraint may be advanced processors. In another, it may be memory, networking, transformers, grid connections, or skilled labor. If developers already have their chips but face a long wait for power, more computing equipment won't solve the problem, and attention and capital may shift toward power generation, electrical equipment, cooling, and engineering. Later, the advantage may move to the businesses that use the infrastructure most productively.

That is why Veda continues to pay close attention to sector and industry rotation, particularly during periods of market volatility. AI and semiconductors remain important long-term themes, and we also look at the companies that power, connect, cool, secure, and apply AI.

Progress will not travel in a straight line. A retailer's impressive AI pilot may never save enough to justify its cost. A data center may be announced with fanfare and then stall for years because the grid can't support it. An algorithm that performs well in a controlled study may struggle with different patients and equipment in the real world.

The same discipline applies to investing. A company can participate in a fast-growing industry and still be a poor investment if its valuation already assumes perfect execution. Railroads, telecommunications, and the internet each created enormous long-term value, and each also brought overbuilding, excessive debt, and companies that disappeared.

The larger picture

For most people, the importance of AI won't be measured in model parameters. It will show up in small, human moments: the farmer who finishes spraying before the storm, the patient whose disease is caught before symptoms appear, the grandmother who speaks naturally with her grandchild, the warehouse employee who goes home with less pain, the older adult who stays independent after giving up a driver's license.

Each of those moments rests on an economic system of chips, sensors, equipment, electricity, networks, software, security, and skilled people. The most valuable artificial intelligence may not be the one that sounds the smartest. It may be the one that quietly helps someone live a better day.

Curious how themes like this fit into your own financial plan? [Schedule a conversation with Veda Financial]

Veda Financial is a registered investment adviser. This commentary is for informational and educational purposes only and does not constitute individualized investment advice or a recommendation to buy or sell any security. References to particular companies and technologies are illustrative. Veda Financial and its clients may hold positions in securities mentioned. This material contains forward-looking statements and opinions that are subject to change and may not come to pass. All investing involves risk, including the possible loss of principal, and past performance does not guarantee future results. Please see our Form ADV Part 2A for important disclosures.