The Bank of Japan left its policy rate at 1%, and one board member wanted an immediate increase to 1.25%. Calling the decision a hawkish hold is fair, but the more consequential change was inside the inflation outlook: the BOJ now says rising semiconductor prices associated with global AI demand are likely to make Japanese durable goods more expensive, especially while the yen remains weak.
The bank expects inflation excluding fresh food to move clearly above 2% during the second half of fiscal 2026, sees price risks tilted upward and still intends to raise rates if conditions develop as forecast. Oil, wages and the currency remain more established sources of pressure, but AI demand has now joined them as something the BOJ may have to answer.
Amazon and Apple help explain how that pressure travels. Amazon reported 37% growth at AWS, raised its annual capital-spending plan to roughly $220 billion and said demand for computing capacity still exceeds available infrastructure. Apple also reported strong sales, but said limited advanced-chipmaking capacity is constraining its outlook.
Those results don’t tell us how much Japanese inflation will rise, and they shouldn’t be used that way. However, they do show that semiconductor scarcity is affecting real decisions about capacity, production and which customers get supplied first.
The largest cloud companies can justify paying for scarce advanced chips because those chips support rapidly growing businesses. Semiconductor manufacturers then have every reason to direct capacity toward the highest-value demand, while companies making phones, computers, vehicles, appliances and industrial equipment compete for what remains.
Japan adds a currency multiplier to that competition. Many of these inputs are priced globally, so a weak yen raises their domestic cost before manufacturers decide how much to absorb and how much to pass through. The same AI demand supporting Japanese exports and investment can therefore work against households through more expensive consumer goods.
If that pressure becomes persistent, the BOJ may have to tighten partly in response to an investment boom taking place outside Japan. Higher rates could support the yen and reduce imported inflation, but they would also raise financing costs for Japanese businesses and households. Nearly half of the Japanese companies surveyed by Reuters earlier this month said previous BOJ increases had already hurt their operations, and some said a move to 1.25% would cause them to reduce capital investment.
The yen’s response suggests that this tradeoff remains unresolved. It weakened to around 160.66 per dollar after the BOJ held, giving back part of the sharp gain attributed to suspected intervention during New York trading. That doesn’t mean intervention failed, but it does show that currency purchases can’t fully replace a change in the underlying rate structure.
There are good reasons not to overstate the risk. The BOJ lowered its fiscal 2026 inflation forecast because government energy subsidies will temporarily restrain reported prices, oil has fallen below $90, and semiconductor producers are adding capacity. Manufacturers may also absorb part of the increase, while AI demand continues to support Japanese production, exports and investment.
The next evidence should appear in Japan’s import-price index and in durable-goods inflation for computers, communications equipment, appliances and vehicles. It will also matter whether Governor Kazuo Ueda begins connecting semiconductor prices to the timing of the next increase rather than treating them as a longer-term risk.
If new capacity lowers chip prices, the yen strengthens and durable-goods inflation remains contained, the pressure will remain marginal. But if those costs continue reaching consumers, the BOJ may have to keep tightening against an inflation channel created by an investment boom centered elsewhere.
The Second Order Brief will now publish no more than twice a week, on weekdays, with additional editions reserved for developments that can’t reasonably wait. I’d rather write when there’s a real consequence worth tracing than publish because the calendar says I should.