Will energy usage doom AI technology’s expansion?
?
?
Power outages could undermine the growth of artificial intelligence (AI) technology unless a solution to the massive energy needs is found.
?
To learn more:
Training significant language and generative AI models takes energy and large amounts of it.
?
The greenhouse gas emissions at Google have increased nearly 50% in the past five years.
Why?
Gemini, Google's AI, is hungry for knowledge, and knowledge takes electrical power to feed it.
They are not the only ones dealing with energy upticks. Although not everyone has been as open about revealing their energy usage, Microsoft has doubled its usage in the past four years.
GPT-3, OpenAI's 175 billion parameter model, needed a reportedly 1,287 MWh (MegaWatts per hour)
DeepMind, Google's 280 billion parameter model, used 1,066 MWh.
To put the energy usage into perspective, it equals about 100 years of energy for a single U.S. household.
Data center energy usage by 2026 will equal all of Japan's electricity for a year.
Goldman Sachs believes that by 20230, data center energy usage could grow by 160%.
Analysts at Bernstein Financial believe AI training could "outstrip current supply [of electricity] in the next two years."
These are staggering numbers not seen since the turn of the 20th century when the industrial revolution began.
?
Is there an answer?
The short answer is to limit AI training, but a digital world will not allow that, so the next best answer is to find ways to meet the need.
My recent story, "What's a miner to do?" offers one solution: tapping large-operation Bitcoin miners.
One of the (many) challenges is to limit the carbon footprint by using renewable energy sources like carbon-free electricity.
Amazon is looking for an old solution to utilize nuclear reactors through a U.S. firm called X-energy. They also look at over 500 solar and wind projects to power their data centers.
OpenAI seems to be turning to Exowatt, a startup company seeking to combine solar thermal technology into thermal batteries.
Microsoft signed a deal with Brookfield, a renewable energy company, to develop wind, solar, and "new or impactful carbon-free energy generation technologies" between 2026 and 2030.
Izzy Woolgar, director of external affairs at Octopus Energy Group, said, "The race to invest in vast new data centers—and the green energy to power them—is on."
Other innovative ideas
While creating renewable energy sources is essential, it isn't the only idea swirling around those creative minds.
Celestial AI is a startup micro-chip company that builds chips using energy-efficient light.
Etched, another startup microchip company, seeks to enhance an AI model known as transformers, which are specific AI singular-purpose chips.
The chips are 20 times faster than Nvidia GPUs (Graphic Processing Units), using less power, which means less energy usage.
?Not resting on past successes, NVIDIA's (the technology company and world leader in AI computing components) next-generation GPU will have a 30% increased power.
A lower-tech innovation is using liquid cooling—as opposed to conventional air conditioners—to cool data centers.
Andrew Schaap, CEO of Aligned Data Centers, suggests liquid cooling could lower energy costs by up to 20%.
Phaidra is an AI-enhanced Industrial control system that helps lower energy costs by maximizing energy efficiency.
?
Is there an answer?
The short answer is yes.
Consider how essential electricity was during the Industrial Revolution. Innovative minds and forward-thinking engineers worked tirelessly to create a solution.
The challenge is the same in the 21st century, although the technology is different. Innovative minds and forward-thinking (electrical) engineers will work tirelessly to create a solution.
From my perspective, watching as challenges arise and innovation rises to the occasion is exciting.
How it impacts everyday people
Humanity-changing shifts are on the way and already here.
The question is, what will you do about it?
You can either fear it or learn how to prosper from it.
So, I'd like you to learn about it, from it, and how to profit using it.
It will be a challenge that involves educating (or re-educating) yourself, but it will be essential for your financial future.
I focus on helping everyday people find financial freedom through the cryptoverse. However, AI development and innovation intrigue me, and I see similar possibilities to financial freedom.
Join the journey. Maybe the things I learn will help you.