Micron Technology has seen its share price fall by about 32% from a June peak even as demand for its high-bandwidth memory (HBM) — a key component in artificial intelligence data-centre hardware — remains extremely high.
What has driven the sell-off?
The company is operating amid a reported shortage of HBM that has allowed suppliers to push prices higher. Yet the broader market reaction reflects mounting concerns about whether the ongoing surge in AI infrastructure spending is sustainable. The tension is between booming demand for specialised memory and the economic strain produced by sharply rising hardware costs.
- Stock movement: a recent fall of about 32% from the June record high, after a near-700% gain over the past year.
- Capacity forecasts: Bloomberg projects around 118 gigawatts of data-centre capacity will be installed across the US by 2030 to support AI.
- Capital intensity: Nvidia’s chief executive is cited saying a single gigawatt of capacity can require about $50 billion of investment, implying nearly $5.9 trillion in total spending if Bloomberg’s forecast holds.
Why it matters
Large-scale deployment of AI depends not just on software but on a continuous supply of expensive chips and memory modules. As hardware prices climb, the economics of running AI models and offering compute-as-a-service become more strained. Companies that must recoup vast capital expenditure may pass costs on to customers, creating pressure on adoption and pricing models.
The source notes that some AI providers have already raised prices for software services. Early evidence of pushback is visible: the example highlighted is that Uber reportedly exhausted its entire 2026 AI budget in four months, a sign that current cost trajectories can rapidly disrupt planning and operations.
"But here's why I won't be buying the recent dip."
That caution underlines investor anxiety: despite extraordinary gains over the year, elevated hardware costs and uncertainty about the return on multi‑trillion dollar infrastructure builds make future earnings paths harder to predict.
Implications and near-term risks
The situation presents several immediate risks for the technology ecosystem:
- Higher costs for AI services could slow adoption among price‑sensitive customers.
- Large capital outlays by hyperscalers and cloud operators require clear monetisation paths — if those prove elusive, investment plans could be scaled back.
- Suppliers like Micron, while benefitting from constrained supply today, face the longer‑term risk that excessively high prices force customers to seek alternatives or limit deployments.
| Metric | Figure |
|---|---|
| Recent peak-to-trough share fall | ~32% |
| One-year share gain | ~700% |
| US data-centre capacity to 2030 (Bloomberg) | 118 GW |
| Estimated capital per GW (Jensen Huang) | $50 billion |
| Implied total capex by 2030 | $5.9 trillion |
For UK and international technology stakeholders, the episode is a reminder that the AI boom depends on complex supply chains and heavy capital. A constrained memory market can lift vendor margins in the short term but also creates friction that may slow the rollout of compute capacity and elevate costs for end users.
Investors and industry watchers will be watching for signs that hardware price rises stabilise, for evidence that cloud providers can monetise their capex, and for whether software vendors and customers adjust pricing or usage to absorb escalating infrastructure bills.
