How long before the Matrix is possible?

By Charles Cunningham. 2025-03-09 London

The rapid advance of AI has made me ask: How long will it take before it is possible to run a full scale Matrix style simulation that has 10 billion simulated people in it with the same fidelity as the real world?

I need to define what I mean by Matrix style simulation. In my view a Matrix simulation would be indistinguishable from the real world to a human observer who enters it. But how can the human observer enter this simulated world? I imagine we can create a full body sensory suit that has full haptic feedback for every part of the anatomy to such an extent that the observer can experience enough sensation so as to feel they are truly present in the virtual world.

The human brain has about 20 petaflops of processing power. 20 petaflops of processing power is c. $60K to buy today (see Nvidia's Project DIGITS,  one petaflops for about $3000). Every aspect of the universe doesn’t need to be simulated to create a simulated universe, just the parts that the observers interact with it. Based on these observations I assume that we’ll need 10x the human brain's compute power to provide a simulated experience for an individual.

Hence,  my estimate is that the hardware needed  to run this simulation will cost about $600K  for a single individual, so a simulation with 10 billion people in it would cost $6,000,000,000,000,000  or 6 Quadrillion  (6x10^15) at today's prices.

The global GDP in 2022 was estimated at about 100,000,000,000,000 USD (10^14).

So the price of a Matrix world is about 60 times the size of the global economy.

This is a lot.

But Moore’s law is still working at an economic level  even though the semi conductor industry has almost reached the point at which components can not be reduced in size. The price of compute is still halving every 18 months.

This is about a one thousand fold cost reduction every 15 years.

This means in 15 years the cost of the Matrix world will be $6 Trillion. Or the current size of the global fossil fuel industry. Still a lot.

But in 30 years time the cost of this simulation will be $6 Billion.

This is well within the range of affordability for large corporations.

So the Matrix is possible in about 30 years at current rates of progress.

But you ask, how will the software be developed? This is going to require a lot of software to simulate every aspect of the world and the observable universe. Surely this is beyond the ability of human programmers?

And you're right. But people will not be needed to program most of the Matrix software. We have AI.

How is this Matrix world going to happen?

I think the development of a synthetic computed universe will be driven by the following industries:

• Computer gaming, especially immersive first person role playing games

• The Metaverse, especially virtual tourism. How else will everyone on the planet be able to visit Venice, Stonehenge and the Pyramids?

• Physical sciences will increasingly want to simulate the universe

• Perhaps even the social sciences will find value in hyperrealistic social simulations using AI agents existing in simulated environments

The big question is, why would anyone bother? I think the answer is definitely entertainment. The value will be in the fact that a Matrix World experience will seem to be as real as a real world experience. Just imagine the sort of holiday destinations and first person game experiences that could be created. The possibilities are truly out of this world! Want to go to Mars? Want to visit the Star Wars universe? Want to be a senator in ancient Rome? Meet Julia Caesar? Talk with the Budha?

There will even be jobs in the Matrix. Real people will be able to interact with each other via their Matrix avatars. This will potentially create a demand for paid for roles in the Matrix universe. Earnings and money exchanged in the Matrix may well be instantiated in the real world as crypto currency.

It’s quite a thought that a Matrix world may be a possibility within less than 30 years.

#Metaverse #Crypto #AI #supercomputing #future

Charles Cunningham

2025-03-09 London