As massive amounts of power are needed to run AI models, Anthropic has signed a $3.4 billion contract to secure generators for a stable power supply.
Imagine this: the AI services we use every day are actually the product of machines running non-stop in massive factories. The virtual space we call the “cloud” is actually rooted in physical locations called “data centers” that consume an enormous amount of electricity. As AI models have become smarter recently, the amount of power these data centers require has increased to an unimaginable degree.
In this situation, AI company Anthropic recently made a surprising decision. They decided to secure gas-fired generators by investing an astronomical sum of $3.4 billion [Anthropic Reportedly Strikes $3.4B Deal to Generate Gas-Fired Power, Anthropic Strikes $3.4bn Deal to Generate Gas-Fired Power for Data Centres]. Why did a software company suddenly jump into the “power generation” business?
Why is this important?
The role of AI in our daily lives is growing every day. However, the work of making AI smart is an extremely energy-intensive process. While it was common for AI companies to rent servers from cloud service providers in the past, companies have now begun to get involved in energy infrastructure directly [Anthropic Strikes $3.4bn Deal to Generate Gas-Fired Power for Data Centres].
Anthropic’s move is evidence that AI companies are seriously grappling with “energy security.” Beyond just being good at making software, there is a risk that AI services themselves could stop if they cannot control the power required to run them. This shows that the AI era is moving beyond the simple “era of coding” into an “era of energy security.”
Making it easy to understand
Let’s use a simple analogy. If a chef is someone who makes delicious food (training AI models), the stove in the kitchen (the data center) is essential equipment. But as the chef’s skills improve, they need more powerful heat, and the neighborhood gas supply is no longer sufficient. It’s as if the chef has chosen a strategy of connecting gas pipelines and buying large-scale generators themselves instead of just researching recipes.
At the end of September, Anthropic signed a contract to supply high-performance, large-scale generators with a capacity of 1.7 gigawatts (GW) from OnePWR Solutions, an equipment supplier located in San Antonio, Texas [Anthropic Strikes $3.4bn Deal to Generate Gas-Fired Power for Data Centres]. Does the number 1.7 gigawatts not give you a sense of scale? It is a massive amount equivalent to the power produced by one to two typical nuclear power plants. This gives you an idea of just how much electricity AI data centers require.
Current situation
The competition in AI technology is becoming fiercer by the day. Anthropic, as well as many other AI companies, is staking everything on expanding data centers to create larger and more powerful models. However, operating a data center is not solved by simply building a structure.
Some media outlets are concerned about the impact of AI growth on environmental issues. It is pointed out that not only are there greenhouse gas issues generated by data centers, but the costs and power consumption to cool data center servers could increase further due to climate change [Anthropic is Burning Gas for the AI Boom]. Anthropic itself has officially mentioned that climate change could be a threat to its profitability. In other words, protecting the environment has become both a survival strategy and a way to save costs for AI companies.
What happens next?
It is highly likely that AI companies will no longer rely solely on “cloud platforms” in the future. As with Anthropic’s case, attempts to invest directly in energy sources or secure stable power grids will continue. The ecosystem of the AI industry is becoming more deeply intertwined with the energy industry, moving beyond software development.
What readers should watch for is just how “sustainable” this energy procurement strategy is. While they secured gas-fired generators this time, finding a solution for how AI companies can increase energy efficiency and use cleaner power amidst the global trend of eco-friendly energy transition will be the next challenge.
Perspective of AI
Behind the rapid development of technology, there is ultimately a need for physical energy input. AI companies stepping directly into infrastructure and energy issues is an inevitable choice, but how they reduce carbon emissions and environmental costs in the process must be dealt with as importantly as the technical perfection itself. It is time to think about what kind of footprint that AI leaves on the planet, as well as how to make it smart.
References
- AI chipsets
- Gas-fired generators
- Eco-friendly solar panels
- 3.4 Gigawatts
- 1.7 Gigawatts
- 5 Gigawatts
- To generate more profit
- To stably secure the power required for data center operations
- To expand into the energy business