Stop Building More Towers. Let’s Just Program The Walls.

The brute force problem in modern networks

We are pushing 5G to its limits and starting to lay the groundwork for 6G. But let’s be real about the physics here. Higher frequencies like mmWave and sub-THz give us insane bandwidth, but they are incredibly fragile. A wall, a tree, or even a person standing in the way can kill the signal.

Traditionally, telecom’s answer to this has been brute force: just throw more power at the problem and build more expensive, power-hungry base stations everywhere to fill in the dead zones. But that’s not sustainable, and it’s a nightmare for infrastructure costs. This is where Reconfigurable Intelligent Surfaces (RIS) come in. Instead of constantly fighting the environment to get a signal through, we are moving toward a paradigm where we simply program the environment itself.

Making walls smart

Think of an RIS as a software-defined mirror for radio waves. It’s a thin sheet of material embedded with thousands of microscopic elements. These aren’t active radios; they don’t generate their own signal or chew up heavy power. They just manipulate the radio waves that hit them.

By applying a tiny bit of voltage via a simple controller chip, we can change how these elements react to incoming waves in real-time. We can tell the surface to reflect a beam in a specific direction, focus it, or even absorb it. For the first time in wireless history, the physical environment isn’t just a barrier – it’s part of the network architecture.

Solving the dead-zone problem

The real magic here is that it solves the dead-zone problem without breaking the bank or blowing up the power grid. If a direct line-of-sight between a cell tower and a user is blocked by a massive concrete building, an RIS mounted on a nearby billboard or window can literally bend the signal around the corner.

Because these surfaces are nearly passive, their power consumption is incredibly low. We could realistically power them with small solar cells or even by harvesting ambient radio energy. They allow us to boost the signal quality at the receiver without increasing the transmit power of the base station.

The software hurdle

Standards groups like ETSI are already working on this, and we’re finally moving out of the lab and into real-world pilot tests. The engineering challenge now isn’t making the materials-we know how to do that. The challenge is on the software side: building control algorithms fast enough to coordinate thousands of these smart surfaces in real-time as users move down the street.

It’s a massive shift in how we think about network design. We’re moving away from brute-force hardware and toward a world where the physical spaces we live in are actively helping us stay connected.

Why Your Next CPU Might Be Alive

For decades, we’ve tried to make computers act like brains. Now, we’re starting to use actual biological neurons to act like computers. This isn’t science fiction; it’s the exploding field of Organoid Intelligence (OI). While our current AI models require massive data centers and enough electricity to power a small country, the human brain operates on about 20W of power - barely enough to light a dim bulb. By integrating lab-grown “mini-brains” with traditional silicon, we are entering the era of the Bio-computer.

Efficiency Beyond Silicon

The problem with the current AI boom is the “energy wall.” A massive LLM (Large Language Model) might take weeks and megawatts to train. A biological neural network, however, learns through experience and synaptic adaptation with incredible efficiency. We are seeing the first prototypes of “Brain-on-a-Chip” systems where clusters of neurons are trained to play video games or process sensory data in real-time.

The “Green” Data Center

Imagine a data center that doesn’t hum with giant fans and air conditioners, but instead looks like a high-tech botanical garden. Bio-computers don’t just offer speed; they offer a path to sustainable intelligence. Because these systems are biological, they don’t produce the same heat signature as a GPU cluster, potentially saving billions in cooling costs and carbon emissions.

The Ethical Frontier

“Programming” biological matter raises questions. At what point does a bio-computer deserve rights? If we use human-derived neurons, is the computer part “us”? As we move from artificial intelligence to synthetic intelligence, the line between “it” and “them” is going to get very blurry, very fast.

The future of tech isn’t just about building better machines; it’s about growing them.

The Day AI Stops Waiting for Your Input

We’ve spent the last decade building AI that reacts. You type a prompt, it responds. You ask a question, it answers. It’s fast, impressive, and sometimes even feels magical. But at its core, today’s AI is still passive. It waits. That’s about to change.

From Tools to Teammates

Right now, AI is a tool. A very powerful one, but still a tool. You have to know what to ask, how to ask it, and when. You decide the direction. Proactive AI flips that model. Imagine a system that monitors your workflows, understands your goals, and quietly optimizes everything in the background. It drafts emails before you think to send them. Flags risks in your projects before they escalate. Surfaces insights you didn’t know to look for. We’re moving from “AI as a hammer” to “AI as a junior partner.”

The Context Engine

The key breakthrough isn’t just better models-it’s better context. Future systems will continuously ingest signals: your calendar, communications, documents, even behavioral patterns. Over time, they build a dynamic understanding of how you work, what you prioritize, and how your decisions get made. This persistent context turns AI from stateless to stateful. From isolated interactions to continuous awareness. And once an AI understands context, it can act.

The Invisible Interface

Here’s the strange part: the best proactive AI might feel like it disappears. No prompts. No chat windows. No dashboards. Just outcomes. Your schedule rearranges itself to avoid conflicts. A report is ready before the meeting starts. A problem gets fixed before it’s reported. The interface becomes the real world, and AI becomes the invisible layer shaping it. When it works well, you won’t notice it at all.

The Control Problem

Of course, this raises a serious question: how much autonomy is too much? An AI that acts on your behalf needs boundaries. It needs to know when to act, when to ask, and when to stay silent. Too passive, and it’s useless. Too aggressive, and it’s dangerous. Designing that balance-between initiative and control-will define the next generation of AI systems. We’ve trained AI to listen. Now we’re teaching it to act. And once it does, the biggest shift won’t be what AI can do-it’s what you no longer have to.