Want to learn more?
Contact a data center team member today!
Will AI Eat the Cloud?
Whitepaper
Premium Content
Structure Research and Iron Mountain Data Centers outline the challenges, constraints and infrastructure impacts on both AI and cloud applications & platforms.
Exclusive Preview
Challenges, constraints and infrastructure impacts
As AI applications get a stronger foothold in the market, will they cannibalize cloud revenues and replace cloud services, or will they supercharge them? What will the AI-driven opportunities and challenges be for the global colocation industry and its customers, and how are data center planning and design likely to adapt to support AI’s appetite for processing power?
This short Iron Mountain Data Centers insight paper, based on the latest analyses and forecasts from Structure Research, looks at the likely practical impacts of AI on infrastructure over the coming five years.
While there is still plenty of scope for shifts, investment, development, and design trends for AI are becoming clearer, giving us strong indications of the potential shape of the market over the next three to five years.
Growth
Revenue versus investment
AI is never out of the news, and agentic AI features, chatbots and image generation are a fact of life for all. But how big is the AI business really, and is it a threat to the cloud?
Despite its remarkable growth, AI is still very much the junior partner in terms of hyperscale revenues. Microsoft Azure data gives a useful indicator of the current relationship. AI currently has a run-rate of about $11-12 billion of Azure revenue. However, crucially, AI is adding about 10-11% of quarterly growth and this figure is rising fast, underpinning overall growth of the Azure AI service stream of 35% as of Q1 2025.
Source: Structure Research
While core cloud revenues such as IaaS also continue to grow extremely fast, AI services are on a powerful trajectory, in what looks like a fast-forward version of the rise of the cloud itself. In fact, perhaps the most remarkable factor about AI services is their speed of adoption, with ChatGPT achieving 1 million users in just 5 days - something which took Netflix 5 years.
Goldman Sachs has predicted that Generative AI alone will account for 10-15% ($200-300 billion) of total cloud revenue of $2 trillion by 2030. This may be the largest AI revenue stream, but it will be one of many spread across multiple sectors. Chatbots like ChatGPT and Copilot will be the highest profile applications, but new industrial and commercial applications such as autonomous vehicles, automated manufacturing, medical diagnostics, defense applications, new financial services, and robotics will also disrupt and divert spend across whole sectors.
The consensus seems to be that AI will be generating about $500 billion overall by 2030 and will still be on a steep growth curve.
Levels of investment
Because of the physical compute density it requires, the critical factor for future AI growth is having the infrastructure in place to enable it. This is driving a boom in the data center and GPU markets.
Hyperscaler annual capital expenditures are projected to reach $375 billion this year, a 36% increase from $275 billion in 2024, and 2024 saw a 64% increase in spending compared to 2023. The top four hyperscalers alone are responsible for approximately 90% of this spend, and so far in H2 2025 these capex commitments are being built on rather than revised downwards.
Source: Structure Research
What is AI investment being spent on?
Typically, this money is being spent half on infrastructure needs, such as self-build and leased data center capacity, and half on servers, including CPUs, TPUs and GPUs, for both cloud and AI related needs.
The dominance of the cloud giants in this compute investment is remarkable. According to Structure Research:
- In 2024 the top four cloud service providers accounted for over 70% of H100 GPUs or equivalents
- The top eight hyperscalers now account for more than 70% of total AI compute
- This will amount to 63.4 million exaflops of new compute capacity this year.
Intelligence is getting cheaper
The unrelenting investment in AI hardware is driving a paradoxical dynamic concerning the cost of chips and the price of intelligence.
Although demand for high-performance GPUs is unprecedented, and there will continue to be fluctuations in the cost of training models (e.g. DeepSeek) while various approaches are tested, the averaged cost of artificial intelligence itself is declining rapidly.
Source: Structure Research based on: a16z Infrastructure. MMLU stands for Massive Multitask Language Understanding, a benchmark designed to test how well large language models (LLMs) — like ChatGPT, GPT-4, Claude, Gemini, etc. — can perform on a wide variety of tasks that resemble human academic and professional knowledge.
As far as the market implications are concerned, this phenomenon is a prime example of the Jevons Paradox. The cost of the cheapest Large Language Model (LLM) with a minimum Massive Multitask Language Understanding (MMLU) score of 83 has decreased by approximately 10x every year from 2022 (GPT-3, 60b) to 2024 (Llama 3.2, 3B). This reduction in the cost of AI training and inference does not, however, lead to reduced consumption. Instead, like improvements in automobile fuel efficiency that led to more car sales and larger cars, the declining price of intelligence is expected to fuel an exponential increase in the demand curve for AI.
The scale of investment is accelerating this cost-per-unit benefit. Hyperscalers are making enormous investments in GPUs and infrastructure because, in the race to deliver services and achieve Artificial General Intelligence (AGI) they understand that the risk of under-investing is dramatically greater than the risk of over-investing.
In the same way, the market is diversifying. LLMs are enabling the growth of a new generation of specialist Small Language Models (SML) which use just a few billion parameters to train, are more adaptable to mobile devices, and deliver niche services using exclusive data sets. This in turn drives up broader AI revenues, expands the market, and contributes to lower unit costs
The resulting explosion in demand will yield substantial returns, with increasing efficiency driving rather than limiting overall utilization and innovation.
Data center demand: through the roof
Roughly half of AI investment is going on data centers.
AI colocation revenue for training and inference is forecast to grow at an extraordinary 77% CAGR from 2025-2030, reaching $134 billion by 2030. By 2030, AI is expected to account for approximately 44% of total global data center colocation market revenue, all from a near standing start a few years ago.
The steepness of this curve has huge implications for the data center industry. Demand for data center capacity is projected to outstrip supply from 2027-2030, with annual global demand reaching nearly 90 GW by 2030, exceeding supply by as much as 500%.
Source: Structure Research
Want to continue exploring?
Enter your information to access the full content.