Two Numbers, One Government Each, Same Day
TSMC disclosed July revenue of NT$467.58 billion, about $14.5 billion, on 10 August - up 44.7% year on year and 5.6% from June, a new monthly record. Cumulative revenue through July reached NT$2.872 trillion, 37% ahead of the same period in 2025, and TSMC used the print to raise its full-year 2026 revenue guidance to slightly above 40% growth in dollar terms, alongside a 2026 capital expenditure budget now set at $60-64 billion, up from the $52-56 billion range it had guided earlier in the year. Chairman C.C. Wei described AI-related demand as extremely robust, which for a company that rarely editorializes on an earnings call is as close as TSMC gets to saying its own forecasts keep underselling reality.
Hours later, South Korea's Presidential Chief of Staff Kang Hoon-sik announced a new 5 trillion won, about $3.52 billion, government fund. In his words, the government will create a new semiconductor fund worth about 5 trillion won focused on promising materials, parts and equipment firms and fabless companies. Alongside it: another 5 trillion won in trade finance for suppliers, a 10-year, 1 trillion won program to bind large manufacturers to smaller suppliers across development, testing and production, and a Mega Special Zone Act intended to fast-track permits and infrastructure. A military airbase near Gwangju is being relocated by mid-2028 to clear 8.3 million square meters of land for the resulting chip cluster.
The Layer Seoul Chose Is Not the Layer It Chose in June
It matters what this fund is not. In June, Seoul unveiled a fab megaproject in which Samsung and SK Hynix, with suppliers and local governments, committed roughly 550 trillion won to new memory capacity - the deal this desk covered on 4 July, which will not ease memory prices before 2028 because it deepens output in the same two companies that already produce about two thirds of the world's memory. That was a bet on capacity. The 5 trillion won fund announced on 10 August is a bet on a different layer entirely: materials, parts, equipment and fabless design, the tier that sits between raw silicon and a finished fab, and the tier a fab operator has to buy from rather than build itself.
That is also, not incidentally, the layer where Europe currently holds real pricing power. ASML supplies lithography no other company can match; German and Dutch materials and equipment suppliers sit deep in TSMC's and Samsung's own bill of materials. Seoul's new fund is not aimed at out-building Taiwan's fabs. It is aimed at growing a domestic bench of materials, equipment and fabless companies that today mostly do not exist at scale in Korea, precisely because Samsung and SK Hynix have historically bought that tier from outside suppliers.
Why This Lands the Week ASML Rallied on TSMC's Print
The same day TSMC's numbers landed, European chip-equipment stocks moved on them: ASML rose more than 2%, with Infineon and STMicro also higher. That is the market pricing exactly the tier South Korea just decided to subsidize - proof, in the same news cycle, that owning the materials-and-equipment layer of chipmaking is currently worth a premium, because a company like ASML captures value regardless of which fab or which country wins the AI capacity race downstream.
Put the two events next to each other and the read is not that Korea is chasing TSMC's fab crown. It is that Korea watched the same earnings call the market did, saw where the premium sits, and moved state money into that layer rather than only into more fab concrete. A fabless-and-equipment fund does not compete with TSMC next quarter. It is a multi-year attempt to build a Korean supplier tier that, if it works, competes for the same purchase orders European materials and equipment firms currently win by default.
What EU and UK Procurement Teams Should Do Before 2028
None of this changes a 2026 or 2027 sourcing decision. The fund is 5 trillion won against an industry where a single leading-edge fab line can cost tens of billions, and building a materials-and-equipment supplier base from a near-standing start takes years of qualification cycles before a Korean fabless or materials firm can credibly bid against an incumbent European supplier. But mid-2028 is now a real date on two separate Korean timelines at once: it is when the airbase relocation clears land for the new chip cluster, and it is close to when the June fab megaproject's new memory capacity was already expected to come online.
For EU and UK buyers running multi-year procurement or supplier-diversification reviews - hardware makers, industrial buyers of chip-adjacent equipment, anyone budgeting a 2028 refresh - the useful move now is to ask each incumbent European materials or equipment supplier one question: what happens to your quote if a state-subsidized Korean alternative exists by the time you place the next order. That is a planning input worth having on file well before 2028 arrives, not a reason to change a supplier today.
Read next: Intel's Foundry Grew 31%, and You Still Depend on TSMC | Sony And TSMC Bet 1 Trillion Yen On The Fault Line That Just Shut Them Down



