AI energy demand seems to be insatiable and is growing faster than renewables and batteries can grow. Will AI become another thing that humans become addicted to at any cost? this is also going to increase gas demand which drives up price which decreases our advantage in powering industrials like steel AI execs who urgently need more energy to power their tech revolution are turning to fossil fuels (msn.com) No one knows how much electricity will be needed to power the AI boom. AI requires massive computing power and energy loads and has triggered an explosion of data centers. Microsoft cofounder Bill Gates said at the conference that the amount of power AI is bound to consume is mind-blowing. "You go, 'Oh, my God, this is going to be incredible,'" Gates reportedly said. ChatGPT alone uses more than 17,000 times more electricity than the average US household daily. The concern among tech execs is that the energy demands of AI will surpass the capacities of clean sources. Wind and solar power aren't reliable because they're subject to changes in weather. While nuclear facilities take years to build, so they aren't a viable option when companies need power immediately. "Tech is not going to wait 7 to 10 years to get this infrastructure built," Toby Rice, the CEO of natural gas producer EQT, said in an interview with the Journal. "That leaves you with natural gas." At the conference, Rice said he was repeatedly asked two questions: "How fast can you guys move? How much gas can we get?"
Amazing how the tech people are just turning into the 19th/20th century robber barons who wrecked the environment in pursuit of profit. Maybe capitalism, in its pursuit of endless growth, is just bad for the planet.
The power demands are high right now because most AI platforms are running on Video Cards due to the high amount of calculations needed. New HW is being designed to accommodate AI SW. NVIDIA, Intel, AMD etc have HW that is in work to be more efficient for AI applications.
what do you know about Cerebras and the chip they are having built just for themselves? private company hired taiwan faactory to build dinner plate sized chip using 5 nm layering. None available for sale Chip startup Cerebras launches new AI processor | Reuters "So the largest chip that we made was our first generation. People said we couldn't make it," Cerebras CEO Andrew Feldman said to reporters on Tuesday. "Eighteen months later we did it in seven nanometer. Eighteen months (after that), we've announced a five-nanometer part. This is the largest part by more than three and a half trillion transistors." CONTEXT Power consumption is a critical problem for AI processing. Cerebras' third-generation chip uses the same amount of energy to achieve superior performance, when power costs to build and run AI applications have soared. Cerebras does not sell the chips by themselves, but says the systems constructed around them are a more efficient method of building AI applications, a process called training. The new Wafer-Scale Engine 3 (WSE-3) has 4 trillion transistors capable of performing 125 petaflops of computing. It was built on Taiwan Semiconductor Manufacturing Co's 5nm manufacturing process.
There are quite a few of these startups designing AI chips. Personally, I think its very simular to the 80s when the 286 started having "co-processors" to do math outside the CPU. Right now, GPUs are doing that for AI and Crypto since thats what they really are. At some point, they will likely be absorbed in the big chip manufacturers with unique instructions for AI applications. Like the 80s/90s, lots of changes in HW and SW are likely. If you havent invested in tech, you probably should. Lots of $ to be made, we are just scratching the surface of AI applications.
already being applied to find better magnetic materials and to tune the magnetic fields to prevent the fluctuations that cause the fields to break down. fission is coming sooner than most think but not fast enough
I had AI format all my associate review comments this year. I listed their good and bad points, goals etc in bullet points and AI produced well written comment sections for their reviews. Took way less time than normal cuz all I had to do was personalize a bit of it. Something to be said about not starting from scratch. Probably saved me a few hours.
As AI booms, land near nuclear power plants becomes hot real estate (msn.com) According to the Hartford Courant, NE Energy has secured $1.6 billion to construct the switching station and bit barns, which will span 1.2 million square feet in total. NE Energy will reportedly spend an equivalent sum on between 25,000 and 35,000 servers. Considering the price of GPU systems from Nvidia, AMD, and Intel, we suspect that those figures probably refer to the number of GPUs. We've asked NE Edge for more information. NE Energy has faced local challenges getting the project approved because residents are concerned the project would end up increasing the cost of electricity. The facilities will reportedly consume as much as 13 percent of the plant's output. The project's president Thomas Quinn attempted to quell concerns, arguing that by connecting directly to the plants, NE Energy will be able to negotiate prices that make building such a power hungry facility viable in Connecticut. NE Energy has also committed to paying a 12.08 percent premium to the town on top of what it pays Dominion for power, along with other payments said to total more than $1 billion over the next 30 years. But after initially denying the sale of land to NE Edge back in January over a lack of information regarding the datacenter project, it's reported that the town council has yet to tell the company what information it is after. While NE Energy vies for approval, other datacenter operators have already furthered nuclear site ambitions or are laying the groundwork to do so. Earlier this month, Amazon Web Services (AWS) agreed to purchase Cumulus Data's atomic datacenters from $650 million.
You hear this about clean energy reliability as the problem, but really the bigger problem is grid capacity: https://www.reuters.com/sustainability/us-power-regulator-weigh-plans-speed-up-green-energy-connection-2023-07-27/#:~:text=Today there is more than,conference following the unanimous vote. Today there is more than 2,000 gigawatts of renewable power waiting to be connected to the grid -- nearly double the amount of current U.S. generation capacity, Federal Energy Regulatory Commission Acting (FERC) Chairman Willie Phillips said at a press conference following the unanimous vote. I supposed it’s a chicken and egg scenario, if you have a limited grid capacity, sources that can consistently supply the limited grid capacity, or better yet flex up and down on demand, are more valuable. But it puts into perspective where the real problem is - not renewable energy but our inadequate grid. Texas has the most clean energy of any state waiting to be hooked up to the grid. More clean energy projects are planned in the US than its grid can handle. A recent study from the Lawrence Berkeley National Lab shows the US has a clean energy backlog of around 2 terawatts of power waiting to go online. The total capacity of existing power plants currently connected to the grid is 1,250 gigawatts. While transmission capacity is lacking, a key hindrance to making renewable energy a reality, the data show that developers are clearly ready to build renewable energy projects.
