What Is Irrefutable

LLMs are changing the way we think about intelligence.

Over the last couple of decades, we have turned almost every technical and intellectually demanding job into something that can be transferred to a computer — and therefore digitally recreated.

• An architect uses a computer to work.
• An engineer uses a computer to work.
• A salesperson uses a computer to work.
• A teacher uses a computer to work.
• A doctor uses a computer to work.

Almost everyone whose work is not directly only tied to the physical world relies on a computer to do their job. And even the people who work directly in the physical world are increasingly leveraged by the digital world.

The digital world is accelerating at an extraordinary pace. The physical world is not.

There are a few things we believe are increasingly difficult to dispute:

• Large language models will become more accurate.
• Large language models will become more computationally affordable.
• Agentic harnesses will improve reasoning capabilities by intelligently delegating sub-tasks across multiple agents.

This will lead to the following:

Large language models will turn the creation of code into a commodity. The digital world will increasingly be able to create itself. Software can replicate software. Digital infrastructure can be created by digital intelligence. Knowledge work can increasingly be performed, accelerated, and coordinated by machines.

Note: We should not conflate software engineering with the creation of code. Code is only one component of software engineering. But as the cost of creating code approaches zero, the cost of building and modifying digital systems will fall dramatically.

The physical world is different

Atoms do not move because a model generated better code.

• Buildings still need to be built.
• Factories still need to be operated.
• Goods still need to be manufactured and moved.
• Energy still needs to be generated and distributed.
• Infrastructure still needs to be maintained.

This creates a growing gap between the rate at which intelligence can improve and the rate at which the physical world can respond.

In a perfect world, every person would be able to leverage AI as much as possible. We would become more productive, create more abundance, and work less.

But if that happens, the physical world will struggle to accommodate our new capabilities and our new needs.

Over time, it will become increasingly difficult to justify having people perform physical jobs when the alternative is to make those jobs easier, safer, more productive, and ultimately automated.

In other words, we need to enable the physical world to benefit from the same advances that are transforming the digital world.

• Better robotics.
• Better manufacturing.
• Better automation.
• Better infrastructure.

It means bringing intelligence into the physical world.

The next transition

We are at the beginning of what I believe will be the next major technological transition.

The first part of this transition is happening in the digital world: intelligence is becoming abundant, cheap, and increasingly autonomous.

The next part will happen in the physical world: intelligence will need to be embodied.

We are absolutely in the middle of a major technological revolution around AI, and the next decade will very much be the time and place for Physical AI.

This is coming from someone who has spent most of his career building in the digital world, working against non-physical assets in financial services, and only theoretically talked about the physical world.

And yet I am completely drawn to being part of the next technological revolution — which might be considered the same as the AI revolution.

The mission

B2A’s mission is to turn Bits into Atoms.

To take the abundance of intelligence being created in the digital world and bring it into the physical one. To be part of the army of people pushing the incredibly heavy boulder up the mountain.

I believe that this is absolutely necessary if we want to continue to uplift society and bring more prosperity where it is needed.

There is also another reason. I alluded to the best case scenario with AI — making the world more abundant. There is also a worst case scenario, where the physical world we have built around us decays.

• Buildings deteriorate.
• Machines break.
• Infrastructure ages.
• Factories become obsolete.
• Systems become too complicated for fewer and fewer people to maintain.

Eventually, we could find ourselves in a world where we have incredible digital intelligence but increasingly lack the physical capacity to maintain the world around us.

B2A will be part of the journey to preserve it.

---

A world where you can design and plan anything will be constrained by a physical world that cannot create and deliver goods at the same pace.