The American company Normal Computing has reached a historic milestone in hardware development by announcing its CN101, which we can already consider the world's first thermodynamic computing chip. This technology proposes a radical change from the traditional silicon-based computing paradigm. To do this, it harnesses thermal noise and the natural randomness of physical systems as a computational resource instead of treating it as an error to be eliminated.
In this model, the chip components start from a partially random state and, after receiving input data, evolve to reach a thermodynamic equilibrium. The stable state that results from this process represents the solution to the posed problem. This approach is especially useful for non-deterministic algorithms such as those used in AI training, in generating images through diffusion models, and in complex linear algebra operations. Well, now we explain it to you in Spanish.
How does the world's first thermodynamic computing chip work?
In traditional computing, chips are designed to be precise and orderly: any signal that is not perfect is considered an error. The Normal Computing CN101 breaks this norm. When it starts working, its components are placed in a somewhat chaotic middle state, like disordered dominoes. When receiving a problem, the system gradually reorganizes itself until it finds a natural balance. That stable final state is the answer. In even more concise terms. While current chips try to operate in a perfectly ordered environment, the CN101 dives into chaos to find the answer more quickly. And all of this with the milestone of consuming less energy.
This approach is called thermodynamic computing and is inspired by how physical systems tend to stabilize over time. The result is a chip capable of efficiently addressing calculations where uncertainty and probability are part of the problem itself. According to Normal Computing, in specific tasks of artificial intelligence training, the CN101 could be up to a thousand times more efficient than a conventional chip.
In short. While current chips try to operate in a perfectly ordered environment, the CN101 dives into chaos to find the answer faster and with less energy. Best of all, it can be combined with traditional processors or AI chips. Thus allowing the creation of a kind of supercomputer capable of solving any type of calculation by resorting to the best solution on the market in terms of performance and efficiency.
