JASM’s first Kumamoto fab posted a $29 million net profit in the first quarter of 2026, up from a $43 million loss the quarter before and a $101 million loss the year before that. That number isn’t new; TSMC filed it back in May. What’s new is that Taiwan’s Minister of Economic Affairs stood up at Monday’s Taiwan-Japan Semiconductor Technology Forum and said it out loud again, six days after a magnitude-7.1 earthquake rattled the same prefecture and forced Toyota, Sony, and Renesas to halt nearby production. The timing is the actual story here, not the profit.

I keep coming back to how deliberately boring the good news is. Fab 1 makes automotive and industrial chips at 12, 16, 22, and 28 nanometer nodes for customers like Sony Semiconductor Solutions, Denso, and Toyota, all of whom sat through the two-to-three-year qualification cycle automotive parts require before anyone signs a real contract. Clear that bar and demand stops behaving like consumer electronics and starts behaving like a utility bill: slow, contractual, largely immune to the boom-bust cycle wrecking everyone selling into phones or GPUs right now, the same cycle that has TSMC raising prices across its own lineup this year. That’s a genuinely reassuring signal to send Japan. It’s also the easy half of the story.

Fab 2 is the hard half. It was supposed to run 6 to 12 nanometer processes, itself still deep ultraviolet territory, until Taiwan’s Economy Ministry approved an upgrade to 3 nanometer in March, on the back of a direct commitment C.C. Wei made to Prime Minister Sanae Takaichi to chase AI-chip demand instead of settling for another automotive fab. 3 nanometers means extreme ultraviolet tools, the kind ASML effectively holds a monopoly on and that SMIC has been hitting a hard ceiling trying to route around, priced at roughly $200 million a machine and bolted to foundations built to tolerances that make them acutely sensitive to ground movement. That’s not a footnote. That’s the whole risk profile of the upgrade TSMC just committed to.

Which is why TSMC’s actual post-earthquake statement is worth reading slowly instead of skimming past as boilerplate. The company said the quake’s “relatively high seismic force” means “a certain amount of time will be needed for related inspections and calibrations.” Calibration is a specific word in this industry: the alignment verification every lithography tool needs after meaningful vibration, and at 3 nanometers the tolerances are tight enough that a clean structural inspection doesn’t guarantee the machines are still aligned. Fab 1 shrugged the quake off because DUV tools aren’t nearly as fussy. Fab 2, still under construction, dodged the worst of it mostly by luck, and Taiwan’s own Economy Ministry is already hedging on whether the 2028 production target survives the aftershocks and a disrupted local supply chain intact.

None of that stopped Minister Kung from calling the Taiwan-Japan relationship “lips and teeth,” an idiom for interdependence so tight neither side can function without the other, and pairing it with news that Japan’s Nitto Denko is putting roughly $140 million into a new materials and process-chemistry facility in Taiwan. Taiwan sends fabrication capacity north, Japan sends materials expertise back, and I don’t think either government volunteers this kind of granular financial detail at a bilateral forum unless there’s a narrative that needs managing. I’m not accusing TSMC of hiding anything. I think a months-old profit figure got redeployed as a reassurance blanket the same week an earthquake made the far more ambitious bet in the same city genuinely uncertain, and that’s a better story than either the earthquake or the profit on its own.

The 2028 target for Fab 2 was already tight before July 28th. Whether Taiwan’s Economy Ministry is right that the quake shaves off weeks rather than months isn’t something I can verify from a bilateral forum transcript, and I’d be skeptical of anyone who tells you they already know.

Sources