By AJ Thompson, CCO at IT Consultancy Northdoor plc
Why the AI sector needs a more balanced approach with realistic infrastructure planning to ensure long-term success
The rapid rise of Artificial Intelligence (AI) has driven both investment and regulatory debates in recent years. However, the latest developments suggest that technology companies and creative industries are re-assessing the long-term sustainability of AI from an economic, infrastructure and regulatory perspective, leaving the industry at a crossroads.
According to new information from investment bank TD Cowen, Microsoft is scaling back its data centre investments. The company has cancelled leases with two datacentre operators and held back on completing contracts on a number of others. Data centre spending in international markets has also been pulled back and re-allocated to the US.
Although the picture isn’t entirely clear, Microsoft’s scale-back indicates that the company has oversupplied following its investment in the data centre space after the success of ChatGPT in 2022. However, the move also suggests that there may be a growing scepticism around AI’s economic returns, both short and mid-term.
Cancelling leases may also indicate that it overestimated the immediate need for an AI-based infrastructure and is now taking a more cautious and measured approach. We also need to consider that this may indicate that the wider tech industry is concerned that the demand for AI justifies the large-scale investment.
This potential slowdown in international markets also shows a shift in Microsoft’s global strategy, impacting cloud and AI service availability and slowing digital transformation efforts in emerging markets. Datacentres also require significant power and cooling facilities and securing the necessary infrastructure in areas where the power supply in an issue may also be another reason for the lease cancellations.
Microsoft’s scale-back could also signal a broader trend among cloud and AI leaders, affecting suppliers, datacentre operators, and enterprises that rely on these services. This could lead to a slowdown in AI infrastructure investment across the industry.
Ultimately, this move signals that AI is facing practical and economic challenges, forcing companies to rethink their infrastructure strategies.
As well as the issues from an economic and infrastructure perspective in the technology sector, there are also issues from a regulatory perspective in the creative industries. AI uses a vast amount of creative content, including video, audio and text. Currently AI developers gather this content by ‘scraping’ data from millions of websites which is then used to train new AI models. In UK law, AI developers can only scrape websites without permission for non-commercial research only.
The UK Government has proposed a change to the current copyright laws to allow technology companies to train their AI models using creative works from films, TV shows, audio works and original journalism without the permission of the creators, unless the creator has opted out.
The creative industries have pushed back on AI training without permission, signalling the increased concern over content protection and fair compensation.
AJ Thompson, CCO at Northdoor plc, explains: “AI’s rapid rise has driven both investment and regulatory debates. With Microsoft’s retreat on datacentres due to overcommitment and infrastructure challenges and copyright concerns from the creative sectors, the AI industry is certainly experiencing growing pains.
“Demands on datacentres are huge and they require high-powered computing and energy resources, leading to concerns over sustainability. Similarly, AI’s reliance on vast amounts of creative content for training raises questions around protecting intellectual property rights.
“Microsoft’s slowdown in datacentre investment reflects a more cautious approach after the initial AI surge. This suggests AI firms are reconsidering their growth strategies amid infrastructure and cost challenges.
“When a company like Microsoft makes significant changes to its AI programme because they aren’t convinced that the business model works, then arguably we all need to take a breather and review our investments before just assuming that the answer is AI whatever the question. Arguably, AI should be used in association with other technology and human interactions.
“The AI sector needs a more balanced approach, with fairer partnerships, licensing models, and realistic infrastructure planning to ensure long-term success. Microsoft’s datacentre cuts and the copyright concerns highlight the need for a more sustainable, regulated approach to AI growth, one that balances innovation with economic and ethical considerations,” concluded Thompson.
