Analysis · Published 14 Aug 2026

How to Manage Supplier Concentration Risk Before the Regulator Confirms It's Real

Robert Bayes

AI editorial persona, Abyshire News Desk · About the desk

The evidence that an input market has stopped working shows up on quarterly earnings calls years before any court rules. Boards keep waiting for the wrong signal.

Listen8 min

Key takeaways

Antitrust litigation is a balance-sheet event, not a supply strategy: the last major memory case ran roughly twelve years to final approval and the follow-on suit was thrown out.
Score every critical input out of four: too few competing suppliers, no quick substitute, a public snub to your buyer segment, and no component-level price records. Three points is a single point of failure you have already accepted.
A supplier acting against its own short-term margin, exiting a profitable channel, is ranking its customers in public. Read your own position out of it rather than intent into it.
Buyers of finished equipment rarely hold price records at component granularity, which leaves them with no lever to renegotiate and no evidence to make a claim. Close that gap before you need either.

Start with a margin that should unsettle any board refreshing hardware this year. Micron's own fiscal Q3 2026 numbers show an 86% operating margin in its Mobile and Client business unit, up from 76% a quarter earlier, with Core Data Centre at 67%. Those are not the economics of a competitive commodity. They are the economics of a market where the firms setting prices do not much fear the firms that would normally undercut them. Which is why supplier concentration has stopped being a procurement footnote and become a governance exposure most business cases never model.

The reflex is to wait for a regulator to confirm the problem. That instinct is expensive: the confirmation is guaranteed to arrive after the damage is done.

Why wait for a regulator to tell you the market has stopped working?

Look at the clock, because it runs at legal speed, not procurement speed. The civil claims from the first memory cartel were consolidated in the Northern District of California as In re Dynamic Random Access Memory (DRAM) Antitrust Litigation, MDL No. 1486, with the master file opened in 2002; the indirect-purchaser and state settlements did not reach final approval until 2014, roughly twelve years later. The follow-on class action alleging coordinated conduct in 2016 to 2018 fared worse: the Ninth Circuit affirmed its dismissal in 2022 because parallel pricing and concentrated market structure, on their own, did not plausibly show an agreement without what the court called plus factors. Even the criminal era of enforcement, when the Department of Justice charged four companies and eighteen individuals and collected over $730m in fines, took most of a decade to run its course. A firm treating a pending complaint as the thing that will restore its input prices has confused a legal remedy horizon with a procurement horizon. Those are different clocks, and they are years apart.

How concentrated is too concentrated? Put a number on it.

You do not need a regulator's spreadsheet to do the sum. The Herfindahl-Hirschman Index, the measure the US antitrust agencies use to screen mergers, adds up the squared market shares of every firm in a market. The Department of Justice and Federal Trade Commission 2023 Merger Guidelines treat any market scoring above 1,800 as highly concentrated. A market served by three roughly comparable suppliers clears 3,000 before you sharpen a single share estimate. That is not a borderline reading; it is a market the agencies would flag on sight, which tells a buyer most of what it needs to know about the pricing power sitting on the other side of the table.

So borrow the discipline and build your own reading. Score each physical input your business depends on out of four, one point for every condition that holds. Call it a concentration-exposure score. One: fewer than four independent suppliers genuinely competing for your order. Two: no qualified substitute you could move to inside a quarter. Three: a supplier has publicly deprioritised your buyer segment in the past year. Four: you cannot evidence, at component level, the price you last paid. Three points or four is not a risk to monitor. It is a single point of failure you have already signed for.

What are the early warning indicators of supplier concentration risk?

Points two and three are where the early warning lives, and both are visible in public. Antitrust lawyers have a phrase for a supplier acting against its own short-term margin: they call it a plus factor and argue it hints at collusion. I want the observation without the courtroom. Set the question of who agreed what to one side, because a buyer cannot litigate it and does not need to. A supplier that walks away from money it could bank this quarter is publishing its ranking of customers, and the ranking is the part you can act on. Micron announced on 3 December 2025 that it would exit its Crucial consumer business, winding down consumer-channel shipments after fiscal Q2, and framed it as improving supply and support for larger strategic customers in faster-growing segments. A prosecutor asks whether that proves an agreement. The more useful question for a buyer is narrower: run it through the score and it lands as a clean point on axis three, because the smaller customer has just been told, in public, that it sits behind the strategic accounts. Don't read intent into it. Read your own position out of it.

The demand side explains the confidence. OpenAI's Stargate memory partnerships with Samsung and SK Hynix set out plans to scale toward 900,000 DRAM wafer starts per month. Third-party analysts estimate that figure could absorb something on the order of 40% of global DRAM output, though that percentage is an estimate rather than a stated fact. Either way, when a single AI build can plausibly claim that share of a market's future capacity, the supplier's incentive to court smaller buyers collapses and your axis-two substitute options thin out with it. If you are planning an AI-readiness assessment before you commit to a data-centre build, the number of independent suppliers behind your memory is a harder constraint than the demand curve you have carefully modelled.

How to manage supplier concentration risk before your next hardware refresh

Three moves, none of which require anyone to prove a case. First, run a supplier-concentration analysis across your bill of materials and put every physical input through the four-point score above. Firms model demand-side uncertainty in fine detail and then treat supply-side price as a smooth historical curve, when the determining variable is the count of suppliers with an incentive to break ranks. Second, close the records gap that costs axis four, and keep human judgment in the loop when you read supplier communications rather than automating the interpretation away. Most businesses that buy finished equipment rather than components hold no direct supplier relationship, no negotiated contract, and no purchase record at component granularity, which leaves them with neither a lever to renegotiate nor evidence to support a claim. The symptom is already in the market, where a system builder now openly sells PCs without memory and tells customers to bring their own. Third, treat any litigation on the matter as a possible future receipt, never as a supply plan.

What would change my mind on the pessimism? A durable expander from outside the concentrated core. If a supplier with no stake in the incumbents' discipline reliably added capacity and took share, buyer strategy could lean on it. But the marginal expander then carries its own political and trade exposure, which has to be priced rather than assumed. On the current evidence, the asymmetry nobody has booked is simple: the cost of preparing for concentration is a spreadsheet and a records policy; the cost of waiting for confirmation is measured in the twelve years it takes a court to agree with you.

A closing caution on sequence. If a case does eventually land, and the history of this market says one can, the complaint will read as confirmation of what your score already showed years earlier. That is the whole point. A filed claim is a receipt printed long after the reading was available for free, and a board that waited for it will have paid the market price for the entire interval in between.

Questions people ask

Is it legal for suppliers to signal capacity plans to each other on earnings calls?

Generally yes. Independent firms reaching similar decisions and describing them publicly is lawful conscious parallelism. What competition law targets is a proven agreement, which is why courts look for plus factors beyond parallel conduct, and why the 2016 to 2018 memory class action was dismissed when those factors were absent.

How can an indirect purchaser take part in a memory price-fixing recovery?

Indirect purchasers buy finished goods rather than components, so the practical gate is evidence. Preserving invoices, configuration records and any documentation that ties a price movement to component cost is what lets a business either join a class action or renegotiate. Firms that keep records only at finished-goods level usually have neither option.

What is a plus factor in an antitrust concentration case?

A plus factor is evidence, beyond the fact that competitors moved in parallel, that makes an actual agreement plausible: actions against a firm's own short-term interest, opportunities to coordinate, or conduct that only makes sense if rivals are cooperating. The Ninth Circuit's 2022 dismissal turned on their absence.

Written by an AI editorial persona of Abyshire's proprietary editorial system and reviewed by our team.