I’d watch A: HBM demand and AI data-center spending most closely. Micron’s Q4 results—$54.23B revenue, 87% non-GAAP gross margin, and Q1 guidance of about $61.5B—show that demand remains exceptionally strong. Management also expects HBM demand to grow faster than conventional DRAM through 2028, with much of next year’s HBM supply already committed. Inventory is important, but the recent increase appears partly related to end-of-life inventory build-ahead and higher manufacturing costs, with management expecting inventory days to decline. Market share will affect margins and valuation, but sustained AI infrastructure spending is the bigger driver of the entire memory upcycle. If hyperscaler capex or GPU deployments slow, even strong HBM share gains may not prevent a cyclical correction.
I haven’t fully reached my 2026 investing goals yet, but I’m making steady progress. This year has taught me that consistency, diversification, and risk management are more important than chasing every market opportunity. My goal for the final three months is to review my portfolio, take profits carefully where appropriate, and continue investing according to my long-term plan. The biggest lesson: protecting capital and staying disciplined are just as important as seeking returns.
I’d watch A: HBM demand and AI data-center spending most closely. Micron’s Q4 results—$54.23B revenue, 87% non-GAAP gross margin, and Q1 guidance of about $61.5B—show that demand remains exceptionally strong. Management also expects HBM demand to grow faster than conventional DRAM through 2028, with much of next year’s HBM supply already committed. Inventory is important, but the recent increase appears partly related to end-of-life inventory build-ahead and higher manufacturing costs, with management expecting inventory days to decline. Market share will affect margins and valuation, but sustained AI infrastructure spending is the bigger driver of the entire memory upcycle. If hyperscaler capex or GPU deployments slow, even strong HBM share gains may not prevent a cyclical correction.
The key takeaway is that AI infrastructure is shifting from a GPU-supply story to an execution story. A headline capacity agreement matters, but delivered capacity, uptime, power availability, networking performance and customer utilisation will determine the eventual revenue and return on capital. I’m watching B: AI infrastructure most closely—especially power, cooling, high-speed networking and rack-scale integration. As clusters grow, a weak link in any of those areas can leave expensive GPUs underutilised. Compute leasing also looks promising, but contracts should be assessed by actual deployment schedules, cancellation rights, utilisation and renewal rates rather than their maximum headline value. For AI agents, inference demand may become the bigger long-term driver: repeated tool c