Last week we covered fuel cells. The week before MLCC. This week the bottleneck is about PCB drill bits.
AI server boards are getting harder to drill.
They are thicker, the holes are smaller, and the materials are tougher. That means the drill bits used to make those holes wear out faster, sometimes after only a few hundred hits.
The effect is easy to miss. Even if PCB volumes grow normally, each board can consume more drill bit life than the one before it.
That is already showing up in the numbers of high-end drill makers such as Topoint and Union Tool. Revenue is accelerating, margins are expanding and both companies are adding capacity.
But there is an important distinction.
PCB drill bit constraint is getting tight. They are not stopping PCB production…yet.
The question is whether drill consumption keeps rising fast enough to absorb the new capacity now being built.
What Is a PCB Drill Bit?
A printed circuit board (PCB) is a stack of copper layers separated by insulating material. Signals travel across those copper layers, but they also need to move vertically through the board.
That requires holes.
Once drilled and plated with copper, those holes become vias, which are the connections between different layers of the PCB.
Lasers handle many of the smallest, shallow holes near the surface. But deeper holes through thick multilayer boards still need to be mechanically drilled. An advanced server board can require tens of thousands of them.
CNC micro-drilling a thick multilayer AI server PCB — extreme aspect ratios make specialized carbide bits a chokepoint.
The tool doing that work is remarkably small. It’s a tungsten-carbide drill bit spinning at roughly 200,000 RPM. More importantly, it is a consumable.
Every drill has a hit count, the number of holes it can make before the cutting edge becomes too worn to reliably produce a clean hole. Once that happens, the bit is replaced or resharpened. At the smallest diameter drill bits, useful life can fall to only a few hundred hits.
Almost every change happening in AI hardware pushes that hit count lower.
Boards are getting thicker: More layers mean more material for the bit to cut through.
Holes are getting smaller: Higher component density requires finer drilling, but thinner bits are also more fragile.
Materials are getting harder: High-speed boards increasingly use lower-loss laminates that preserve signal quality but are tougher on the cutting edge.
This creates an unusual demand multiplier.
AI does not simply require more PCBs. Each PCB can consume more drill bit life.
So drill bit demand can grow faster than board volumes themselves.
Tightening, Not a Bottleneck Yet
The broader High-Speed PCB market still reads Balanced on Tessara, with new high-layer and high-precision capacity arriving fast enough to absorb much of the AI demand.
Drill bits are showing a tighter setup underneath that.
The mix is moving rapidly upmarket: At Topoint, high-end coated drills increased from 48% of volume in 2025 to 56% in H1 2026, already exceeding the company's original 55% target for the full year.
The mix shift is showing up directly in earnings: Topoint's revenue surged in the latest quarter, but profits grew even faster, pushing operating margins to roughly 28%. That suggests the company is not just selling more drills but also a richer mix of higher-value products, used for AI boards.
Topoint’s Revenue Chart on Tessara
Supply is responding quickly: Topoint's drill capacity reached roughly 35 million units per month from Q2 and is expected to reach 45 million by year-end. It is now planning for 70 million by the end of 2027 and 90 million by the end of 2028.
The demand signal is clearly strong. But suppliers are still able to add capacity. So the read today is not that PCB manufacturers cannot get enough drills.
It is that high-end drill consumption is rising fast enough to improve pricing, product mix and supplier profitability even while capacity expands.
That is a tight market.
Whether it becomes an actual chokepoint depends on what happens next.
Who Captures the Economics?
The drill bit market is not equally exposed to AI.
Two things matter:
Commodity drills benefit from more PCB volume: More boards mean more tools consumed, but competition is broader and pricing power is limited.
High-end drills benefit from more PCB complexity: Thicker boards, smaller holes and harder laminates shorten tool life while pushing customers toward finer-diameter and coated products.
That puts the manufacturers strongest at the high-end in the better position. Here’s a quick comparison:
Topoint (8021 TT · Taiwan) | Union Tool (6278 JP · Japan) | |
|---|---|---|
Market cap | ~NT$63bn / US$2.0bn | ~¥273bn / US$1.7bn |
2026 share price | +128% YTD, ~25% off Jun high | +55% YTD, ~51% off Jun high |
Valuation | ~86x trailing, ~52x on FY26E | ~29x on FY26 company guidance |
H1 revenue | NT$3.17bn, +66% | ¥27.1bn, +48% |
H1 operating profit | NT$782m | ¥7.6bn, +81% |
H1 operating margin | 24.7% (Q1 20.1% → Q2 28.1%) | 28.1% |
Topoint is the clearest example of PCB complexity translating into earnings.
The stock is up roughly 142% this year and trades around 52x forward earnings, even after pulling back from its June high. That valuation leaves little room for the current growth rate to slow.The next test is whether demand can absorb the capacity now being added.
Topoint expects monthly capacity to rise from roughly 35 million units today to 70 million by end-2027.
If utilisation and pricing hold as that supply comes online, it would suggest the market is absorbing new capacity almost as quickly as Topoint can build it. If either rolls over, today's tightness may simply have pulled supply forward.Union Tool: the high-end capacity play
Union Tool offers a slightly different setup.
It is the global leader in PCB drills, with more than 30% share, and has direct exposure to high-layer-count PCBs and package substrates used in AI servers and data centres.
The latest results show how strong that exposure has become. First-half operating profit rose more than 80% year on year. Management attributed the increase to demand for server package substrates and high-layer-count PCBs, and raised its full-year forecast again.
The valuation setup is also different from Topoint.Union Tool is still up roughly 55% this year, but the stock has fallen about 50% from its June peak and now trades around 25x earnings.
So Topoint offers the more explosive earnings conversion. Union Tool offers the larger incumbent with a much less demanding valuation after a sharp correction.
For both, the same question matters:
Can AI driven drill consumption continue absorbing the capacity now being added?
The Upstream Risk: Tungsten
There is another constraint sitting further upstream.
PCB micro-drills are made from tungsten carbide, and that supply chain is heavily concentrated in China.
Tungsten ore is first refined into ammonium paratungstate (APT), then converted into carbide powder before reaching drill manufacturers. China produces more than 80% of the world's tungsten and added several tungsten products, including APT and tungsten carbide, to export controls in 2025.
Prices have already moved sharply higher.
So far, that has been an economics problem rather than a production problem. High-end drill makers have raised prices, shifted toward richer products and continued expanding margins.
The risk is what happens if access, rather than price, becomes the constraint.
Paying more for tungsten supports pricing power. Not getting enough tungsten limits how many drills can be made.
And unlike drill bit capacity, new tungsten supply cannot be added quickly.
What Becomes Binding Next?
We asked Tessara: if AI demand for high-end PCB drill bits keeps rising, what is most likely to tighten first? Precision manufacturing capacity or tungsten supply?
Research Tab on Tessara
Precision manufacturing capacity: the more immediate risk. High-end grinding and coating equipment is specialised, the supplier base is narrow, and new lines take years rather than quarters to scale.
Tungsten: the bigger tail risk. So far, higher prices have been manageable. A harder Chinese export restriction would change that quickly.
The clearest confirmation would be lead times. If high-end PCB drill waits push beyond six months, the market starts looking like a real production bottleneck rather than just a tight one
What Would Change the Read?
For now, drill bits look like a tightening input inside a still-Balanced High-Speed PCB market. It’s not a demonstrated production bottleneck yet.
The read tightens if:
New fine-diameter capacity arrives and pricing or lead times still stretch.
Drill makers start reporting capacity limits rather than simply stronger demand.
Tungsten controls begin restricting physical availability of APT or carbide powder.
Tessara's broader High-Speed PCB demand signal strengthens alongside it.
The read eases if:
Topoint, Union Tool and peers add capacity faster than drill consumption grows.
Better coatings or drilling technology extend tool life.
Tungsten supply normalises without affecting production.
Until then, the interesting divergence remains: PCB capacity is keeping up, but one of the consumables needed to use that capacity is getting tighter.
The Week Ahead
Earnings Tab on Tessara
Thursday, September 3
CIEN (Ciena) — Optical Components. Watch: Whether backlog prints above $7.7B. Rising backlog alongside 40% revenue growth means orders are outrunning the ability to ship.
Read our post and pre-call briefs here and stay prepared.
See where the AI buildout goes next
This issue, we named PCB drill bits as the chokepoint and Chinese tungsten as the thing that actually controls them.
In Tessara terminal, you can track the high-speed PCB constraint and the tungsten chain running beneath it, and see which of 400+ public names are most exposed to the squeeze.
See you next week,
Teng & Arvind
This article is for informational and research purposes only. It is not financial advice, investment advice, or a recommendation to buy or sell any security. Tessara Research does not publish price targets. The views expressed here reflect our analysis at the time of publication and may change as new evidence arrives. Readers should do their own research and consult a qualified financial adviser before making investment decisions.