The next AI chokepoint: power, memory, storage, or cooling?
As artificial intelligence scales, the next constraint may not be chips, but the infrastructure around them. Power, memory, storage and cooling are emerging as the new battlegrounds of the AI economy.
Aako Ugbabe · 24 July 2026
The first wave of AI investing was relatively simple: buy the companies supplying the compute. Nvidia became the clearest winner, while memory and storage names such as Micron and SanDisk benefited as investors realised AI workloads require far more than just GPUs. The question now is where the next bottleneck appears.
Insight research
Continue reading with a subscription
Insight members get deeper company research, sector analysis and premium market commentary. Or buy just this report.
More in Technology & Infrastructure
Nvidia Q2 Earnings Show the AI Infrastructure Boom Is Still Accelerating
Nvidia delivered another substantial earnings beat in its fiscal second quarter, with revenue more than doubling year-on-year as demand for AI infrastructure continued to accelerate. Strong Data Center growth, the ramp-up of Vera Rubin and an unusually bullish outlook suggest the AI investment cycle still has considerable room to run, although supply constraints, rising memory costs and China remain important risks.
ET&A Research · 27 August 2026
The Hidden Importance of Semiconductor Packaging
As AI chips become larger, faster and more memory-intensive, advanced packaging is emerging as a critical performance bottleneck, reshaping semiconductor design, supply chains and the investment opportunity across the industry.
ET&A Research · 25 August 2026
The Infrastructure Underneath the Internet
The digital economy may feel weightless, but it depends on thousands of kilometres of fibre-optic cable, landing stations and network infrastructure. As cloud computing, AI and geopolitical tensions grow, the physical foundations of the internet are becoming increasingly valuable.
ET&A Research · 12 August 2026