I don’t follow, what’s the connection between EV batteries and data centers? Are you thinking of battery storage to complement renewable energy at data centers?
There is nothing like demand to create capacity and overcapacity. In China that seems to be the modus operandi, they overbuilt housing only to demolish some of it.
In the West it’s a different story, fear of China policies and EV adoption forecasts drove the new factories only for ICE incumbents to reduce their EV plans.
The other use of these batteries is energy storage. I wonder if all formats can be used. Cylindrical cells certainly but how about pouches?
Very simply, he is saying that irrational exuberance in EV’s caused an over-capacity in battery supply, and that irrational exuberance in AI will cause an over-capacity in data centers. We had the same with dark fiber 25 years ago.
As long as you guys keep thinking that AI is some kind of algorithm on steroids you keep missing the paradigm shift. Traditional computing, algorithmic computing, is a boolean process, If A then B. Intelligence is a statistical pattern matching process, that bunch of pixels look like a dog.
I started writing algorithms in February 1960. I was assigned to write a program to tally the toll road cards for the Caracas Valencia highway on an IBM 650. The problem I had was that the code was too big to fit in the machine. On three occasions I went to my boss to tell him the code was too big. His reply, 'It fIts!" Total frustration, I could not do it. Then one morning at around 4 AM I woke up with the solution as clear as daylight. I could not wait for the office to open to try it out. It worked.
The solution was to replace a lengthy algorithm with pattern matching, also know as a Table Lookup. For years I tried to explain “smart algorithms” (boolean) vs. “dumb algorithms” (pattern matching) but I could not explain it. Years of writing code also kept me thinking about how the brain solves problems. Dumb vs. smart algorithms didn’t fly. Algorithmic AI (expert systems) was such a flop that AI was declared DOA.
Finally this century technology was advanced enough to implement neural network based AI and it works. Neural networks mimic how the brain does pattern matching. It’s not algorithmic but statistical.
The human brain has billions of neurons and trillions of synapses. Don’t take my word for it, look it up:
Google AI
The statement is correct: the human brain contains approximately 86 to 100 billion neurons, which are interconnected by over 100 trillion synaptic connections. These synapses are the points where neurons transmit signals to one another, and their vast number allows for complex thinking, learning, and memory.
Current AI data centers are puny by comparison.
The Captain
o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o
8 billion human brains are how many neurons and synapses? Just add nine zeros to the above numbers
