When Tech Giants Claim Nuclear Energy to Power Their AI Chatbots

When a query about how to make Rogan Josh asked to an AI chatbot consumes more energy than making the dish itself, it’s time we start thinking about alternative ways to power the AIs of the world.

A simple query asked to ChatGPT consumes 15 times more energy than a Google search. Last year, the data centres that enable AI models to work, were responsible for using up 4% of the total electricity consumption in the US. As more and more AI models come up, the demand for data centres – the heart and soul of AI models – would see a surge like never before.

With the growing popularity of OpenAI’s ChatGPT that broke the Internet last year, global tech giants like Microsoft, Google, Amazon, and Meta have joined the race to build the smartest AI-powered chatbot. And all these AI assistants are gobbling up energy like a bear about to go into hibernation. Moreover, it’s not just about more electricity but better quality of electricity as data centres need greater computational power to train more advanced AI models.

All this doesn’t bode well for companies like Microsoft that have pledged to reduce their carbon footprint. But it appears that they have kind of found the solution, or at least working on it.

The answer to feeding the ever-hungry AI chatbot models with clean energy seems to lie in using nuclear power. According to the International Energy Agency, nuclear power accounts for about 10% of global electricity generation today. However, new nuclear technologies are still unproven when deployed at scale.

In this article, co-founder of The Content House , Avanish Tiwary writes about the recent developments that have happened at the intersection of AI, data centres, and nuclear power as well as the role Southeast Asia will play in hosting data centres for the tech giants of the world.

Tech Giants and the Renaissance of Nuclear Energy

The US alone has roughly 5,388 data centres – mostly run by big tech firms like Google, Amazon, Microsoft, Meta and Apple.? According to the Electric Power Research Institute, these data centres are projected to more than double their electricity consumption to 9% by 2030.

The chatter in the tech space for a better and alternative source of energy – one that doesn’t speed up the death of an already dying planet – has been going on for some time now.

But it’s only recently that things have started to move in the right direction. Tech giants are scripting deals that could make nuclear energy ready for docile purposes at a large scale.

It’s no joke to get hold of nuclear power and that’s why tech giants are moving fast and making sure they take steps in the right direction. Let’s quickly see who’s doing what and how.

• Google

Google has signed an agreement with engineering company Kairos Power in its bid to gain access to carbon-free power that can run its AI systems. In what Google terms as the “first of its kind” deal, Google will back the construction of seven small nuclear-power reactors, also known as small modular reactors (SMRs), to power its data centres. SMRs are advanced reactors that generate about a third of the power produced by traditional nuclear reactors.

These SMRs will bring up to 500 megawatts of carbon-free power to US electricity grids. The first reactor is expected to be functional by 2030, followed by additional deployments through 2035.

This is in line with Google’s commitment to run on carbon-free energy by 2030. In a recent interview with Nikkei, Google’s CEO Sundar Pichai said, “Obviously, the trajectory of AI investments has added to the scale of the task needed. We are now looking at additional investments, be it solar, and evaluating technologies like small modular nuclear reactors, etc.”

• Amazon

Amazon has signed three agreements to support the development of nuclear energy projects. This month it announced its agreement with Energy Northwest, a consortium of state public utilities. As a part of this agreement, Amazon will help in the development of four advanced SMRs.

These small nuclear reactors are expected to generate roughly 320 megawatts of power in the first phase of the project. It has the capacity to go up to a total of 960 megawatts, which is enough to power more than 7,70,000 American homes.

Additionally, it has partnered with Talon Energy in Pennsylvania to purchase its data centre campus powered by a nuclear plant; invested in X-energy, a leading developer of next-generation SMR reactors and fuel; and signed an agreement with utility company Dominion Energy to explore the development of an SMR project.

It is expected that these three partnerships alone would generate more than 5,000 megawatts of power by the late 2030s.

• Microsoft

Microsoft’s target of being carbon negative by 2030 was dampened when its carbon emissions shot up by almost 30% in 2023 as it opened multiple new data centres to chase its AI dream.

The run to get hold of nuclear reactors to generate clean energy for running the data centres is so severe that Microsoft has gotten into an agreement that would revive the shuttered Three Mile Island nuclear plant.

Microsoft has signed a 20-year agreement with Constellation Energy Corp., the owner of the decommissioned power plant. The nuclear reactors are expected to come online by 2028, pending regulatory approval.

All Data Centres Stop in Southeast Asia?

While the tech giants of the world are investing heavily in nuclear power plants, they are also looking for alternative homes for data centres that are expected to increase multifold as the demand for AI goes up.

Considering that the US already accounts for 70% more data centres than the 10 largest markets combined, it makes sense for the likes of Amazon, Google, and Microsoft to set up new data centres at locations that could bring down the cost without compromising on the quality.

Despite China being the next big market for data centres, it is not an option for US companies due to the geo-political tussle between the two. Similarly, India’s privacy laws are a bit difficult for the tech giants to navigate.

This leaves Southeast Asia as the next best choice for tech conglomerates. Fast internet connection, adoption of cutting-edge technology, and favourable government policies of major Southeast Asian countries are a few reasons why the region makes for a great market to set up data centres.

Google Cloud, Amazon Web Services (AWS), Microsoft Azure, and NVidia have already made their moves to establish their data centres in Southeast Asia.

AWS plans to double its investment in Singapore to more than USD 17 billion by 2028, with additional commitments of USD 6 billion in Malaysia by 2037 and USD 5 billion each in Thailand and Indonesia over 15 years. Microsoft CEO Satya Nadella announced investments of USD 1.7 billion in Indonesia and USD 2.2 billion in Malaysia for cloud and AI infrastructure.

Similarly, Google has announced to invest USD 1.6 billion in Thailand to build new data centres. The data centres in Thailand will support Google’s AI-driven services such as search, maps and workspace. It has already invested billions of dollars in Singapore and Malaysia.

NVidia on the other hand, has taken the partnership route to bring its AI capabilities to Southeast Asian countries. It has joined hands with Singapore-based telecommunication giant Singtel where the latter will provide remote access to the US chip maker’s GPUs through its local data centres. It’s also planning to build a USD 200 million AI centre in Indonesia in partnership with local telco giant Indosat Ooredoo Hutchison. There are talks between NVidia and the Thailand government as well for a possible investment in the country to set up data centres.

All this to say that Southeast Asia is becoming the data centre hub that will power the AI capabilities of global tech giants. In 2023 alone, Southeast Asia attracted USD 10.23 billion in investments for the construction of data centres, with projections reaching USD 17.73 billion by 2029.

So What Now?

With queries on AI-led chatbots requiring at least four to five times more processing capacity compared with traditional internet searches, data centre demand in the region is expected to grow at around 20% a year for the next five to seven years. And the need for more processing capacity means not just more energy but alternate and clean energy, which has resulted in the resurgence of nuclear power at a large scale.

As AI's influence grows, so too will the demand for powerful data centres, driving a quest for sustainable energy solutions. It will be interesting to see how the deep connection and dependence between artificial intelligence, data centres, and nuclear energy shakes things up.

Like Hedwig? Subscribe to it here to get insights on the Asian startup ecosystem and building a sustainable business.

At The Content House, we offer research-based, analytical content that has a strong narrative quality to it. What makes us different in the crowded content market is our ability to convert institutional knowledge and expertise locked inside organizations into content that companies can use to increase brand reach.