The US 10-Year Treasury yield is approaching 5%, rising from 4.58, while Japan's 10-Year Government Bond yield has jumped from 2.478 to 2.807.1 Together they signal the end of a decade of artificially cheap global capital.
The synchronized rise is cleaving the AI sector along one fault line: cash flow timing.
Higher discount rates compress the present value of distant earnings. Pre-revenue AI infrastructure companies — data center buildouts, chip-dependent compute platforms, model developers projecting 2028 revenue — absorb that compression disproportionately.1 A 40-basis-point increase in the risk-free rate produces a structurally larger haircut when most projected returns sit years away.
Profitable AI enterprise software firms face a different calculus. Microsoft and peers generate near-term cash flows that discount at shorter effective durations.1 Rising rates shrink their valuations less, because less of their intrinsic value depends on distant projections.
Japan's rate move matters beyond Tokyo. For decades, ultra-low Japanese yields made US growth stocks comparatively attractive. As that spread narrows, the relative premium that global capital assigned to American tech equities erodes.1 Investors from Frankfurt to Singapore are recalibrating exposure accordingly.
Rate uncertainty compounds the pressure. The 10-Year US Treasury benchmark oscillated between 4.5 and 4.6 even as the note benchmark pushed toward 5.0, reflecting market disagreement on the terminal rate.1 That dispersion itself is a valuation input — wider uncertainty raises risk premiums across all long-duration assets.
The implied portfolio response is rotation within AI, not exit from it. Reducing weight in unprofitable infrastructure while adding profitable enterprise software with existing contracts represents the rate-resilient position. AI ETF flows over the next 60 days will show whether that rotation is already underway.1
