The figure is only for FY25 and comes at a time when the public is finally noticing the industry’s impact
Microsoft vice chair and president Brad Smith kicked off the New Year by revealing the company is on track to invest approximately $80 billion to build out AI-enabled datacentres to train AI models and deploy AI and cloud-based applications around the world – in FY25 alone. In a flag-waving blog post – with one eye on the incoming president – he highlighted that more than half of this spend will be in the United States.
The timing is fascinated given, as Richard Windsor at Radio Free Mobile recently pointed out, there are more signs that the “scaling laws” that have underpinned the AI explosion (and all of the hype attached to super-intelligent machines) are coming to an end meaning that the real potential of LLMs is now visible and is falling way short of the craziest of forecasts.
The inevitable consolidation round is already underway and Microsoft has backed Open AI so hard it needs this LLM to succeed. According to Windsor, foundation models (GPTs, Gemini, Llama, Claude etc) are rapidly becoming the operating systems of AI and, as such, they are likely to be control points making them strategically vital to making money from AI. And this money will not be coming from training new models, it will come from interesting ways inference can be applied to solve enterprise business cases.
Gridlock for DCs
At the same time, 2025 looks like the year that the public will finally notice the stupendous impact the spread of data centres will have on the global grid, climate change and inevitably, consumer power bills.
According to Bloomberg, the enormous demand for electricity by data centres is now leading to a phenomenon known as bad harmonics. Electricity travels across high-voltage lines in waves of alternating current and when those wave patterns deviate from what’s considered ideal, it distorts the power that flows into homes. Bad harmonics can force home electronics to run hot or damage the motors in refrigerators and air conditioners, resulting in billions of dollars of damage, according to Bloomberg.
A Bloomberg analysis of exclusive sensor data coupled with data from DC Byte, a market intelligence firm, showed a strong link between proximity to data centers and worsening harmonics. More than half of the tracked households showing the worst distortions of power quality are located within 20 miles of significant data center activity, according to the analysis, which covered readings from February through October, 2024.
Consumer price impact
There needs to be further research on the expanding power needs of data centres and the direct impact on consumer energy bills this year. There is an inevitable cost to upgrade the grid due to rising energy demand and that cost is highly likely to be passed on to consumers. This is because, expanding grid capacity to handle more highways requires large investments in generation, transmission and distribution infrastructure. Energy firms recover these costs through rate adjustments.
Many countries have regulatory systems that allow utilities to recoup costs for infrastructure projects deemed necessary to ensure reliable supply, typically through higher prices. As data centres consume more energy, peak demand periods may require additional capacity. To incentivise efficient usage or cover higher costs, energy providers can increase rates for all consumers. In many countries, grid upgrades are viewed as shared benefits, meaning even consumers who don’t directly use energy-intensive services like data centres will see a portion of their costs on bills.
At the same time, efforts to transition to renewable energy sources may compound the investment burden, further driving up rates. Energy prices have significantly influenced politics since the pandemic, making energy security, affordability and transition pivotal issues across the globe. It’s difficult to discern how this will play out but it isn’t going to resolve itself and once this is a political issue, the data centre industry will almost be too late to change any opinion, regardless of how environmentally aware their public positions are.
American AI
In his blog, Smith (above, right) mentions “American AI” no fewer than 10 times and emphasises how Microsoft is part of a tech movement that is “investing heavily to spread American AI platforms around the world” and “advancing international adoption of American AI”. In the same blog he argues the most important US public policy priority should be to ensure that the US private sector can continue to advance “with the wind at its back”.
“The United States cannot afford to slow its own private sector with heavy-handed regulations. The country instead needs a pragmatic export control policy that balances strong security protection for AI components in trusted datacentres with an ability for US companies to expand rapidly and provide a reliable source of supply to the many countries that are American allies and friends,” he said.
Such potential light-touch “American AI” regulations will, of course, be viewed carefully in Brussels – also keen on sovereignty – but Microsoft’s flag-waving post instead reminds readers the company intends to invest more than $35 billion in 14 countries within three years to build trusted and secure AI and cloud data centre infrastructure.