The Sherman Act theory DOJ just put on the record
A May 'Old Crime, New Code' speech tries to fit algorithmic pricing under the 1890 Sherman Act, with no cases filed and an 'agreement' theory the courts have not yet tested.
A May 'Old Crime, New Code' speech tries to fit algorithmic pricing under the 1890 Sherman Act, with no cases filed and an 'agreement' theory the courts have not yet tested.
On May 14 in San Francisco, Acting Deputy Assistant Attorney General for Criminal Enforcement Daniel Glad told a West Coast antitrust conference that the DOJ intends to treat algorithmic pricing tools the same way it treated the price-fixing cartels of the 1990s. The speech, titled "Old Crime, New Code", did not announce a single case or name a single company. Its actual contribution was setting out the legal theory the division wants to bring when it does.
That theory is the part worth reading carefully. Glad's argument is that no new statute is needed. The Sherman Act, the 1890 federal antitrust law that criminalized price fixing, already prohibits "agreements" to fix prices. Algorithmic pricing, in the DOJ's emerging view, can satisfy the agreement element without any human communication. On this reading, two firms whose pricing models learn from the same data feeds, or share a common third-party tool, can constitute the kind of coordinated behavior Sherman Act §1 was written to reach.
This is where the legal theory runs into trouble. Sherman Act §1 has historically required proof of "contract, combination… or conspiracy," meaning some form of human communication or coordinated intent. Algorithmic pricing tools rarely produce that kind of evidence. They tend to produce parallel pricing patterns that look, in retrospect, like coordination. Federal appellate courts have wrestled with the gap, including in the Masonite and Text Messaging Antitrust cases, where they have been skeptical of inferring agreement from parallel behavior alone.
Glad's speech tries to close that gap by treating shared inputs as proxies for agreement. If two competing energy traders license the same third-party pricing model trained on the same historical capacity data, the DOJ's emerging position is that the shared tool itself can supply the connective tissue the courts have demanded. The same logic applies to firms that subscribe to common SaaS pricing platforms or run large language models against the same competitive feeds.
Major law firms read the speech the same way: A&O Shearman, DLA Piper, Hogan Lovells, Mayer Brown, and Mondaq all treat it as a real escalation, not a routine talking point. The fact that every major competition practice reached the same read in the same week is itself a signal: firms are telling clients to assume the speech will be litigated, not forgotten.
The energy sector is the most exposed not because the DOJ singled it out, but because the sector matches every feature the speech identifies. Wholesale power markets are concentrated. Capacity and fuel procurement repeat on tight cycles. Pricing models in this space share inputs through the regional grid operators' public market data feeds and a small number of dominant trading platforms. Mayer Brown's analysis reads the speech against Enron-era precedent, where the same concentration plus coordinated behavior produced actual prison time.
The Enron comparison goes beyond what the speech itself offers. Glad's remarks announced no investigations and no grand jury activity. The Sherman Act theory on offer is one the DOJ has wanted to assert for years and has yet to win in court. What the May speech changes is the signaling environment: companies that run algorithmic pricing tools in concentrated markets now have a public, on-the-record statement from a senior DOJ official of how the division wants to bring the next case.
The cross-firm consensus tells compliance teams, pricing-software vendors, and procurement leaders to treat the speech as the opening move of an enforcement campaign. For any company whose pricing touches shared data, common platforms, or competitor-observable signals, the practical question is no longer whether the DOJ is interested. It is whether the firm's pricing stack can document, in advance, that each input was independently sourced and that any human review has the authority to override the algorithm. That is the documentation regime the speech implies, even where it does not yet say so.
The next concrete signal will be whether the Antitrust Division files a case, sends a grand jury subpoena to a pricing-software vendor, or issues formal guidance to companies using algorithmic tools. None of that has happened yet.