Estimated reading time: 7 minutes

The Big Picture: Not One, But Three Grey Rhinos

Copper just touched a record high, and laminate suppliers have raised prices in double-digit increments across separate rounds. Oil spiked, collapsed, and is climbing again. Tariffs keep shifting under everyone's feet. If you say all of that in one breath, it sounds like a single undifferentiated mess: Everything costs more, and nobody's sure why or when it stops.

That's the wrong way to hold it in your head, and I think it's an expensive mistake. There are three kinds of problems. The first is cyclical: It goes up, it comes down, and it has already done both this year. The second one is structural: global, slow, and not something a new supplier can price away. The third isn't a market force at all; it's a policy choice with a date attached. Each problem behaves differently, resolves on a different timeline, and calls for a different response from you.

Author Michele Wucker termed this pattern a “gray rhino.” It’s a big, obvious threat coming straight at you, and still manages to run people over anyway, because obvious isn't the same as acted-on. These three problems qualify as gray rhinos. It didn’t require a crystal ball to see them coming; the signals were public well before this year. But what separates them isn't whether you saw them coming. It's how fast each one is moving, and whether it can be turned.

Oil Only Reaches You Going Up

I’ll start with brent crude. It started in the mid-$60s in January, passed $100 within weeks of the Iran war's Feb. 28 start, reached a peak near $104 in May, fell back to the low $70s by early summer, and reached near $100 again in September as the fighting moved into the shipping lanes. That’s a full round trip in eight months, and nothing else on your cost sheet has done that.

What should get your attention isn’t the barrel price, though. That round trip only went one direction to your board. The spring run-up moved fast down the chain, with two Saudi petrochemical strikes—Ras Tanura in March, and the Jubail complex weeks later—adding to the pressure. Bisphenol A and epichlorohydrin, the feedstocks behind epoxy, jumped 19% and 6% inside a single week, and epoxy went up 12% behind them. Then crude fell by a third over the summer, and none of it came back. Resin contracts ratchet. They reset upward on force majeure and allocation, and they do not reset downward because a barrel got cheaper for six weeks.

Epoxy does track crude, and it’s why the spring run-up arrived as fast as it did. But it doesn't track it back down. That asymmetry means the commodity only tells you half of what you need. Our own supplier data has epoxy climbing through the second half regardless of where crude goes next, and that reflects every driver—feedstock, capacity, allocation, AI-driven demand for the high-end grades—not just the number on the evening news. You can still keep an eye on crude, but don't build a fourth-quarter quote on the belief that a falling barrel will show up in your laminate.

Copper and Laminate
What about these two? Copper is up roughly 40% year over year, triggered by a September 2025 mudslide that forced Freeport-McMoRan to declare force majeure at its Grasberg mine in Indonesia, one of the largest copper deposits on earth. Two rounds of revisions have cut that mine's 2026 output by nearly half. When will it come back? Freeport tells regulators it will be at the end of 2027. Its own Indonesian unit head has floated early 2028.

Chinese smelters spent the year absorbing it, to the point that sulfuric acid—a smelting byproduct—became worth more than the copper itself, and some now buy pyrite, a sulfur ore with almost no copper in it, purely to keep their acid plants running. That is what a structural shortage looks like from the inside. By spring, it reached us: Mitsui raised copper foil pricing 12% in a single move.

Laminate has the same problem, compounding rather than spiking. CCL suppliers have raised prices at least 10 times since 2025, with Kingboard and Panasonic both moving inside the same 48 hours in early September. Most customers will see only a 4–5% increase because board shops will absorb the costs by consolidating. Standard grades are up more than half for the year, and fiberglass cloth has climbed faster still. Note that not everything is moving at once. We’re tracking flat drill and routing bit pricing through most of this year, though suppliers are flagging increases there, too. These are specific inputs under specific pressure, and the list is lengthening.

The effect on a quote is not what most people expect. Rather than exotic grades being the problem, and standard FR-4 being fine, it is closer to the reverse. AI-grade laminate is far more profitable to produce than standard FR-4, so capacity has shifted toward it, and standard FR-4 has borne the brunt. The advanced grades are expensive and allocated. The ordinary ones are simply being squeezed off the schedule. It’s not a capacity issue for fabricators, but having the material to run through it.

What’s surprised me is that grade barely predicts lead time—the supplier does. Across our own sourcing, two mid-Tg FR-4s of equivalent specification run about a month from one maker and close to a half-year from another, and I have seen a high frequency laminate quote faster than plain FR-4 from several makers on the same sheet. That’s a six-fold spread inside one grade, wider than the spread between grades. Ask by part number, supplier by supplier. The grade on the drawing will not tell you what to expect.

Be precise about what that does and doesn't buy you. Changing vendors does not change the price; each is buying the same copper, glass cloth, and resin from the same allocation at the same moment you are. But it places you in a queue and the grade you're asking for. That can be the difference of months. A board requalified onto a more available laminate is a genuine solution. The same scarce material with a different logo on the invoice is not.

I don’t see this turning around soon. The binding constraint on fiberglass cloth is not the cloth plants. It is the looms. The air-jet looms that weave high-end electronic cloth come almost entirely from a handful of Japanese machine builders, delivery has stretched from roughly six to 18 months, order books run past 2027, and from ordering a loom to stable production could be up to two years. That puts you in 2028, no matter what anyone spends.

Which is about where the laminate producers land, too: EMC, Shengyi, Doosan, and Panasonic have all committed real money, but Doosan's Thailand line, the most concrete addition on the table, isn't targeting mass production until the second half of 2028. EMC’s and TUC’s 2027 and 2028 additions are reported fully booked by AI server, satellite, and high-end switch customers, which means relief, when it arrives, may not arrive on standard grades first.

This squeeze doesn't trace to a broad economy bidding up materials, but to the capital spending of a very small number of hyperscalers, and that buildout isn't on rails. U.S. grid interconnection queues have passed two terawatts, beyond total installed capacity, and Omdia expects 30–50% of planned 2026 data center capacity to slip to 2028. Slip, not cancel, but on this year's order book, those look identical. If relief arrives before the plants do, that's how it arrives. The first sign won't be a price cut; it will be availability returning on standard grades, which is why I'd watch lead times on standard material rather than the price line.

Gold is up roughly two-thirds of its entire 10-year climb in the past two years, but it's also down about a fifth from its January peak, so it behaves more like oil than like laminate. Fabricators care about the level, not the direction. It is a real and growing cost on gold-finger and mil-spec work. On standard ENIG, the gold layer runs two to four microinches, so it moves a quote by less than people expect, but it does move it, and on a board using ENIG, it is now a line worth watching on its own rather than noise.

Tariffs Aren't a Commodity at All

Tariffs are worth separating out cleanly, because they don't behave like a commodity. This year alone:

• The Supreme Court threw out the IEEPA tariffs in February
• The administration replaced them inside a week with a 10% global surcharge under Section 122
• A trade court struck that down in May

It expired on its own statutory clock in July, replaced the same day by a Section 301 forced-labor tier covering 60 economies that, when stacked on top of the 25% rate PCBs already carry, works out to roughly 37.5% on Chinese-origin boards and 10 to 12.5% almost everywhere else.

Two more dates are already set: A Section 301 overcapacity investigation, sitting on USTR's public docket since March, by Sept. 24 summit, and Nov. 9, when a batch of product exclusions expires and reverts to the underlying rate.

None of this shows up in our own cost-driver data for raw materials, which is entirely about copper, resin, and laminate; tariffs have no effect. Tariffs move the landed cost by an amount that depends entirely on the origin. Many people miss that the forced-labor tier spans 60 economies, so leaving China doesn't mean leaving tariffs behind; it just means choosing which ones you pay. On a $25,000 shipment, a Chinese-origin board carries roughly $9,375 in duty. The identical board from Taiwan costs about $2,500. A different number on the invoice, purely because of where it was made.

That's also why it's the one lever you can actually pull, on your own timeline, without waiting for a mine to reopen or a plant to finish construction. Qualify a non-Chinese source, and move enough of the work to change where substantial transformation legally happens, or accept the duty and price it in on purpose. But don’t wait until Nov. 9 to find out.

Know Which Rhino You're Looking At
So, three problems, three rhinos. Treating them the same is exactly how a rhino gets close enough to charge. If you wait out the structural pressure hoping it behaves like oil, you'll be waiting until 2028. Price a quote off the crude chart, and you'll be wrong in whichever direction it moves next. Ignore the tariff dates because they feel like just another commodity headline, and you'll pay a penalty that was entirely yours to avoid.

I'll admit this worries me less than it did a year ago, not because the numbers got smaller, but because the shape of the problem finally makes sense to me, and that’s something I can actually plan around. But act isn't the same as acted-on; that's what makes it a gray rhino, and it's the part that no amount of data solves for you. You can name all three now. However, naming them isn't the same as moving.

Mehul J. Davé is chairman of Linkage Technologies, Inc.