In the futuristic microchips, every microchip has its own power source.

Above: AI-image that portrays the microprocessor that has its own fuel cell. 

Why it's hard to make a computer that mimics brains in energy use? People say that brains will not need so much energy as microchips. But our brains are more effective than any computer. The reason our brains are so energy-efficient is that every cell has its own power source, called mitochondria. That means our cells begin their energy production in them. Energy travels in cells in the chemical form. 

And that decreases waste energy. The ion pumps on the cell's membrane control the need for nutrients and when cells do not have very much to do, it turns mitochondria in the low level. That decreases temperature. And keeps the waste temperature minimum. The brains are energy-effective. Because every single cell in the neural structure can increase or decrease their energy production by controlling the nutrients that mitochondria use for energy production. 

That keeps the energy level in billions of neurons as low as possible. The human brain has two types of data transmission. The electric and chemical. The neurotransmitters are chemical qubits that travel in axons. The electric impulses are things that the neurons use very fast communication. 

And the neurotransmitters collect that information and put it in packets. The electric impulses prepare the neurons to route those neurotransmitters into the right cells. In human brains, every neuron in neuron groups is one state of qubit. The thing that makes human brains so effective is that neurons always act in groups. They have so-called speculative computing modes. 

The reflex is the natural version of speculative data processing when certain cells send a message to the common address that electric impulse activates the reflex that is movement in certain cells. Electric signals are enough to activate reflexes. This means that brains must not store all information in them. Brains also share their responsibilities with larger areas in the nervous system that make things lighter for them. 

But then we can start to think about computers that architecture is far from systems that we know. In regular computers the power source is centralized. The system gets its electricity from the electric net. And that makes the system a little bit ineffective. The wires between those microchips form heat that decreases the data systems' effectiveness because heat causes oscillation in those wires. 

The solution can be the hydrogen or hydrocarbon miniature fuel cells. Those fuel cells can used as the artificial mitochondria for every microchip.

That fuel cell is installed in every single microprocessor. The microprocessor controls the fuel, like hydrogen (or hydrocarbon-like alcohol) flow in those fuel cells. The hydrogen system can replace electric systems in power input. Another thing is that the two-layer multicore microchips can mimic human brains. Those processor groups can be used as morphing neural networks. 

The system can control every fuel cell independently. And those systems can be the solution to thermal problems. 

Another solution is the photonic computer. The multicore processor with those fuel cells or artificial mitochondria can communicate with other processors by using laser beams. The microprocessor groups can also have internal photonic architecture. The main goal is that the data travels in computers as little in copper, or electric wires as possible. 

https://www.quantamagazine.org/how-much-energy-does-it-take-to-think-20250604/