Commerce’s proposed supply chain rule reached into old contracts

Commerce’s November 2019 proposal would let it block or unwind technology deals case by case, including updates and managed services under contracts signed years earlier. For regulated buyers, the exposure lived in the installed estate and existing contracts, so four preparations could not wait.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

Commerce’s November 2019 proposal would let it block or unwind technology deals case by case, including updates and managed services under contracts signed years earlier. For regulated buyers, the exposure lived in the installed estate and existing contracts, so four preparations could not wait.

  1. The week On November 27, 2019, the Department of Commerce published a proposed rule in the Federal Register.
  2. How the proposal reached a contract already signed Some supply chain rules attach to a product or a vendor.
  3. The estate: exposure lived in what was already running Procurement processes are built around new purchases.
  4. Provenance: knowing who stands behind each supplier The proposal’s second test turned on whether technology was designed, developed, manufactured or supplied by persons owned by, controlled by, or subject to the jurisdiction or direction of a foreign adversary.
  5. Contracts: who pays when a deal must be unwound A technology contract written around outages, breaches and price changes may not contemplate a government order to stop using a product mid-term.
  6. Replacement: the operational test The proposal’s final step was the hardest for operations.
  7. Where the argument is weakest The strongest objection is scale.
  8. What to put in place first Sequencing matters because the four parts depend on each other.
  9. With hindsight Since 2019, the proposal’s procedures have been replaced twice, and its core reach has survived.

The week

On November 27, 2019, the Department of Commerce published a proposed rule in the Federal Register. Its title was “Securing the Information and Communications Technology and Services Supply Chain.” It proposed a procedure for carrying out Executive Order 13873 of May 15, 2019. Written comments were due by December 27, 2019.

The executive order rested on the International Emergency Economic Powers Act and the National Emergencies Act. It gave the Secretary of Commerce power to prohibit any acquisition, importation, transfer, installation, dealing in or use of information and communications technology or services. The proposal called each of these a “transaction,” and it used the shorthand ICTS for the technology and services involved.

A transaction could be prohibited if it met three tests. It involved property in which a foreign country or foreign national had an interest. It involved technology designed, developed, manufactured or supplied by persons owned by, controlled by, or subject to the jurisdiction or direction of a foreign adversary. And it posed an undue risk to critical infrastructure or the digital economy, or an unacceptable risk to national security or the safety of U.S. persons.

The proposal set out how the Secretary would decide. Reviews would be case by case, started on the Secretary’s own motion or on referral from another agency. Parties would get written notice of a preliminary determination. They would then have 30 days to oppose it or to propose mitigation. The Secretary would issue a final determination within 30 days of receiving that information. The outcome could be a prohibition, a finding that the transaction was not prohibited, or permission subject to mitigation. A prohibition could require the parties to stop using the technology immediately, even if it had been installed or running before the decision.

Three design choices stood out. First, the rule reached backward. It covered transactions initiated, pending or completed after May 15, 2019, regardless of when the contract was signed. Ongoing activities, including managed services, software updates and repairs, would count as transactions completed after that date even under a contract signed before it.

Second, there was no list. The proposal did not name the foreign adversaries, calling that determination a matter of executive branch discretion. It also declined to treat any technology or market participant as categorically in or out of scope. Third, there was no way to get comfort in advance. The Secretary would not issue advisory opinions or declaratory rulings on particular transactions.

The proposed penalties followed the emergency powers statute. A person who violated the rule could face a civil penalty of up to $302,584, adjusted annually for inflation, or twice the value of the transaction. The proposal expected parties to keep records of covered transactions in the ordinary course of business. Anyone notified of a review would have to retain all related records at once.

The department asked commenters several pointed questions. It asked whether there should be categorical exclusions, and for whom. It asked what mitigation could look like and how the government would learn if it stopped working. It also asked how “dealing in” and “use of” should be read. The proposal also named the agencies the Secretary would consult on who counts as a foreign adversary. They were Treasury, State, Defense, Justice, Homeland Security, the Trade Representative, the Director of National Intelligence, the General Services Administration and the Federal Communications Commission. On December 23, 2019, Commerce extended the comment period to January 10, 2020.

The same autumn, federal contracting rules were tightening on a narrower front. On December 13, 2019, the Department of Defense, the General Services Administration and NASA issued a second interim rule to the Federal Acquisition Regulation under section 889(a)(1)(A) of the fiscal 2019 defense authorization act. It required offerors to represent annually whether they offer the government equipment, systems or services that include covered telecommunications equipment or services. That rule applied to companies selling to federal agencies. The Commerce proposal applied to almost everyone subject to U.S. jurisdiction.

Read quickly, the Commerce proposal looked like a rule about future purchases. A buyer would avoid certain suppliers and carry on. The text said something different. It reached transactions completed after May 15, 2019 under contracts signed before that date, and it named managed services, software updates and repairs as transactions in their own right. A bank renewing a support agreement, a hospital accepting a firmware patch or a county paying for a hosted service was, on this reading, entering a new transaction.

Our argument is that the proposal moved supply chain risk out of the purchasing decision and into the installed estate and the contracts already signed. Commerce offered no list of adversaries, no categorical exemptions and no advisory opinions. A regulated buyer could not wait for clarity before acting, because the clarity would arrive as a determination with a deadline. The organizations best placed were those that could answer four questions about what they already ran.

The four parts are distinct and, together, cover what a buyer controls. The first is the estate: which running systems and services the rule could reach. The second is provenance: who owns, controls or directs each supplier and its subcontractors. The third is the contract: who bears the cost if a deal must be stopped or unwound. The fourth is replacement: whether operations can swap a component on a deadline set by someone else.

A first section explains how the proposal reached contracts already signed. The four parts follow in order. A sixth section tests the argument against the strongest objections and against different kinds of organization. The last sets out what to put in place first. This analysis is not legal advice, and organizations should confirm their reading of the current rule with counsel.

How the proposal reached a contract already signed

The transaction as the unit of review

Some supply chain rules attach to a product or a vendor. A list names the item, and buyers stop buying it. The Commerce proposal attached to a transaction instead. It defined that broadly, using the executive order’s verbs: acquisition, importation, transfer, installation, dealing in or use. Any one of those acts, involving covered technology, could be reviewed.

The choice matters because a single relationship with a supplier produces many transactions over its life. The original purchase is one. Each renewal is another. Each software update pushed to a device, each repair, each month of a managed service is arguably another. The proposal said so directly for managed services, software updates and repairs.

The May 15, 2019 line

The proposal covered transactions initiated, pending or completed after May 15, 2019, the date of the executive order. It added that this applied regardless of when any contract, license, permit or authorization was signed or granted. A contract from 2015 did not shelter the update shipped under it in 2020.

For a buyer, the effect was to turn the installed base into live exposure. A network appliance bought years earlier was not itself a new transaction. The vendor’s next support call was. If the Secretary later found that the vendor met the foreign adversary test and posed undue risk, the continuing relationship could be stopped.

Case by case, with no advance comfort

The proposal described a case-by-case process. Commerce explained this as a way to target risky transactions without catching similar ones that posed no undue risk. It also said the approach would avoid precluding innovation or access to technology.

The price of that flexibility fell on the buyer. Without a list, no one could check a supplier against a published standard. Without categorical exemptions, no sector could assume it was outside scope. Without advisory opinions, no buyer could ask in advance whether a specific deal was acceptable. The first formal signal a buyer might receive was a notice of a preliminary determination, with 30 days to respond.

List rules and transaction rules compared

The contrast with federal contracting rules shows why the design mattered. Section 889 of the fiscal 2019 defense authorization act worked from named companies and categories of telecommunications and video surveillance equipment. A contractor could read the list, check its products and make a representation. The FAR interim rule of December 13, 2019 asked offerors to do exactly that once a year.

The Commerce proposal offered no such checklist. A buyer could not certify compliance in advance, because compliance depended on a determination that had not yet been made. List rules reward a clean inventory at a point in time. Transaction rules reward the ability to respond quickly once a decision lands. Organizations that treated the Commerce proposal as another list to check against were preparing for the wrong kind of rule.

The two approaches also differed in who they bound. Section 889 bound federal contractors in their dealings with the government. The Commerce proposal reached any person subject to U.S. jurisdiction, so a state agency, a community bank and a physician group were all within its terms.

Mitigation, prohibition or unwinding

The possible outcomes ran from permission to an order to stop. Between them sat mitigation: conditions that would make an otherwise prohibited transaction acceptable. The proposal asked how mitigation should be monitored and how the government would learn if conditions became obsolete. Those questions implied that mitigation could be a lasting obligation, with reporting, for the buyer as well as the seller.

At the far end, the Secretary could require parties to cease using the technology immediately, even if it was already installed. That is the clause a chief information officer should have read twice. It meant the remedy could arrive in production, not at the procurement desk.

Who could start a review, and what records mattered

Reviews could start on the Secretary’s own motion or on referral from another federal agency. That widened the set of eyes on any supplier. An agency that learned of a risky product through its own work could hand the matter to Commerce, and the buyer would not see that step.

The recordkeeping language was short and consequential. Parties were expected to keep transaction records in the ordinary course of business. Once notified of a review, they had to retain everything related to the transaction. For a large organization, that resembles a litigation hold. Someone has to know which contracts, invoices, configuration records, support tickets and emails relate to a given supplier, and be able to freeze them quickly.

In our view, that differs from a litigation hold keyed to people and date ranges. A supplier-keyed hold needs a different index: every system the supplier touches and every team that deals with it. Building that index is part of the estate work described next.

The estate: exposure lived in what was already running

Why the installed base mattered more than the pipeline

Procurement processes are built around new purchases. The continuing transactions that keep old purchases alive can escape them. Maintenance can renew automatically. Updates can flow through vendor portals or directly to devices. Managed service fees can appear as recurring invoices that nobody reopens. Under the proposal, those are the transactions most likely to be live at any moment.

A regulated organization therefore needed a view of its estate organized by continuing relationship, not by asset. For each running system, the useful facts were practical: who supplies updates, who has remote access for support, who hosts it and where, and which subcontractors touch it. An asset inventory that lists hardware models and software versions answers few of those.

Categories that carried the most weight

The proposal did not rank technologies. Its own background section, however, described the supply chain as supporting critical infrastructure and the nation’s ability to store, process and transmit data. For banks, health systems and public agencies, four categories followed from that framing.

  • Network and communications equipment, including routers, switches, wireless gear and the management software that controls them.
  • Endpoint and security software that runs with high privilege and updates itself, so the vendor can change its behavior at any time.
  • Hosted and managed services that hold sensitive data, where the supplier’s staff and systems are part of the control environment.
  • Connected devices in clinical, building and field settings, which are often maintained by the manufacturer for years and are hard to replace.

These are our categories, offered as a way to prioritize, not a list the rule contained. They share one property: the supplier keeps acting on the buyer’s environment long after the sale.

Building the map from records that already exist

The map of continuing relationships does not need a new survey to start. Most of it can be assembled from five existing sources.

  1. Accounts payable. Recurring payments show every supplier that is paid monthly or annually, which is a close proxy for continuing service.
  2. The contract repository. Renewal dates, auto-renewal clauses and support terms show which relationships refresh and when.
  3. Remote access records. Vendor accounts in privileged access tools, VPN logs and support portals show who can reach production systems.
  4. Update channels. Endpoint and device management tools, and outbound network traffic to vendor update services, show who can change software in the environment.
  5. Hosting and data flows. Cloud and hosting agreements, plus data flow diagrams from privacy and security reviews, show where sensitive data leaves the organization.

Joining those five views by supplier name is tedious but finite. The output is a list of suppliers ranked by how many critical systems they can change and how much sensitive data they hold. That ranking decides where provenance and contract work start.

Ranking within the map

Not every supplier deserves the same attention. A workable ranking uses three factors. The first is privilege: whether the supplier can change software or configuration on critical systems. The second is data: whether it stores or can reach sensitive customer, patient or resident information. The third is substitutability: whether a replacement could be in place within a quarter.

Suppliers that score high on all three belong at the top. That short list is likely to include endpoint security, network management, core processing or electronic health record hosting, and a few managed service providers. Starting with perhaps twenty suppliers, rather than hundreds, makes the rest of the work possible within a year.

What an examiner or auditor would ask for

In our reading, existing expectations already covered much of the ground. Bank third-party risk management, the HIPAA risk analysis that covered entities and business associates perform, and federal agencies’ own supply chain duties all ask related questions.

The evidence that would satisfy any of those reviewers is the same evidence the Commerce proposal made urgent. It is a list of continuing supplier relationships for critical systems, each with an owner, a data classification and a record of who can push changes. We would expect the parts to be scattered: contracts in legal, assets in IT, vendors in procurement, risk ratings in compliance.

Provenance: knowing who stands behind each supplier

Ownership, control and direction

The proposal’s second test turned on whether technology was designed, developed, manufactured or supplied by persons owned by, controlled by, or subject to the jurisdiction or direction of a foreign adversary. Each word widens the net. Ownership is a cap table question. Control can come through board seats, contracts or financing. Jurisdiction depends on where the company and its staff operate. Direction is the hardest to see from outside.

A due diligence questionnaire built around security controls, financial health and insurance would not capture this. Nor would one that asks only where data is stored. The proposal made it relevant where code was written, who owned the parent company and which engineering teams could push updates.

The depth problem

Provenance also runs down the chain. A domestic integrator may resell a foreign-made appliance. A software vendor may embed a component from another company. A managed service provider may use subcontracted support staff in several countries. The proposal’s definition of transaction reached any of these if the covered technology was involved.

A regulated buyer could not trace every component. It could, however, require its critical suppliers to disclose material ownership, the countries where development and support take place, and the third parties whose software or hardware is embedded in what they deliver. It could also require notice when any of that changed. Those are contract terms and questionnaire fields, not research projects.

What the absence of a list did to diligence

Without a named list of adversaries, buyers had to choose their own working assumptions. The proposal described a foreign adversary as a government or non-government person the Secretary determined had engaged in a long-term pattern or serious instances of conduct significantly adverse to U.S. national security. A buyer could reasonably infer which countries were likely candidates. It could not know.

The practical response was to collect provenance broadly and decide narrowly. Gather ownership and location data for all critical suppliers, then apply whatever list the government eventually published. That approach costs more at the start. It avoids a second round of questionnaires once the list appears, and it produces data useful for other purposes, such as concentration risk and sanctions screening.

Questions that produce usable answers

Supplier questionnaires can fail when their questions invite marketing replies. Provenance questions work better when they ask for facts that can be checked and kept current.

  • The legal name and country of the ultimate parent, and any holder of a significant ownership stake.
  • The countries where the product is designed, where code is written and where it is built.
  • The countries from which support staff can access customer systems or data.
  • The third parties whose software or hardware is embedded in the product, at least for components that run with high privilege.
  • Any pending government review of the supplier or its products, in any jurisdiction, to the extent the supplier may disclose it.

Answers can be partly verified. Public company filings, corporate registries and sanctions lists confirm some ownership facts. Development and support locations are harder to verify, so the contract should make the answers representations, with a duty to update. That turns a questionnaire into an enforceable statement.

Provenance changes after the contract is signed

Ownership is not fixed. Technology suppliers are bought and sold, take outside investment and move development work between countries. A supplier that was clearly domestic at signing can change hands during a five-year contract. Under the proposal, the next update after that change was a new transaction.

Two controls address this. The first is a contractual duty to notify the buyer of a change of control or a material change in development or support locations. The second is periodic monitoring of news and filings for the top-ranked suppliers, so the buyer does not depend on the supplier’s notice alone. A change should trigger a fresh provenance review and, where needed, a decision on whether to continue at renewal.

Who owns the answer

Provenance data falls between functions. Procurement collects vendor information but rarely tracks ownership changes. Security assesses technical risk but not corporate structure. Legal and compliance handle sanctions screening but often only for payments. The proposal gave organizations a reason to assign one owner for supplier provenance on critical systems, with a defined refresh cycle and a trigger for review when a supplier is acquired.

Contracts: who pays when a deal must be unwound

The gap in standard terms

A technology contract written around outages, breaches and price changes may not contemplate a government order to stop using a product mid-term. Termination for convenience clauses, where they exist, can carry fees. Termination for cause generally depends on a breach by the supplier. A Commerce prohibition is neither.

The result could be an organization ordered to stop using a product while still owing payments for it. It might have no right to a refund, no transition support and no data return terms written for the situation. The proposal did not address private contracts at all. It left allocation to the parties.

Clauses that do the work

A buyer could close most of the gap with a small set of terms in new contracts and renewals.

  • Regulatory termination. A right to terminate without penalty if a government order prohibits or materially restricts use of the product or service.
  • Disclosure and notice. A duty on the supplier to disclose material foreign ownership, control and development locations, and to give prompt notice of any change or of any government review.
  • Transition assistance. A defined period of continued support, data export in usable formats and cooperation with a replacement supplier, at agreed rates.
  • Refund of prepaid fees. Return of unused subscription or maintenance fees if the relationship must end because of a prohibition.
  • Mitigation cooperation. A duty to accept reasonable mitigation conditions and to provide the evidence the buyer would need to show compliance.

Renewals as the moment of bargaining power

Reopening every contract is unrealistic. Renewals, however, come around on their own. Because the proposal treated continuing services as new transactions, the renewal was also when exposure was refreshed. Tying the new clauses to the renewal calendar for critical suppliers spreads the work over two or three years and puts the effort where it buys the most.

Resellers, integrators and flow-down

A regulated organization may not contract with the manufacturer at all. It may buy through a reseller, a systems integrator or a managed service provider. The supplier of record may hold no information about the manufacturer’s ownership and may have no power to agree to transition terms on its behalf.

Flow-down clauses close part of that gap. The buyer can require its direct supplier to obtain disclosure from upstream manufacturers and to pass through any notice of government review. Where the direct supplier cannot do so, that fact is itself useful. It shows which relationships carry provenance the buyer cannot see, and those deserve a higher replacement priority.

Mitigation as a continuing obligation

The proposal allowed the Secretary to permit an otherwise prohibited transaction subject to mitigation. It asked commenters how compliance with mitigation could be monitored over time. A buyer should assume that some of those conditions would fall on it, for example limits on where data is stored or on who may access a system.

Contracts can prepare for that. A clause can oblige the supplier to cooperate with any mitigation measure and to supply the logs and attestations the buyer needs. It can also make the supplier bear the cost where the measure addresses its own ownership or operations. Without such a clause, the buyer may carry the compliance burden alone.

Public buyers and standard terms

Public agencies may buy through cooperative contracts, statewide agreements or vendor terms set by another body. Their room to add clauses can be limited. They can still use purchase orders, statements of work and data protection addenda to add disclosure and notice terms. They can also ask the body that holds the master agreement to add a regulatory termination clause at its next renewal, which helps every agency that buys from it.

The trade-offs

Suppliers can be expected to resist open-ended termination rights, and smaller buyers have less bargaining power. Some suppliers may price the risk. A credit union buying a core banking service from a large provider may have little room to obtain bespoke terms. It can still ask for disclosure and notice, which cost the supplier little, and it can compare suppliers partly on their willingness to agree.

Replacement: the operational test

A deadline set by someone else

The proposal’s final step was the hardest for operations. If the Secretary prohibited a transaction, the parties might have to stop immediately. Even where a transition period was allowed, the timetable would belong to the government, not the buyer. Replacement capacity, not contract terms, would decide how painful that was.

Replacement difficulty varies widely by category. Endpoint software can often be swapped across a fleet in weeks if management tooling is in place. Network equipment may need hardware purchases, configuration work and outage windows. Clinical devices may need revalidation and staff training. A hosted service that holds years of records needs a data migration and a parallel run.

What readiness looks like in operating terms

An organization able to absorb a sudden order would have a few traits. It would know which systems it could replace quickly and which it could not. It would have at least one alternative supplier identified for critical categories. It would keep configuration and data in exportable forms. And it would have a decision path that could approve emergency spending without waiting for the next budget cycle.

A simple test captures most of this. For each critical supplier, the owner should be able to state how long a full replacement would take, what it would cost within a range, and which services would be degraded during the switch. If no one can answer, the organization does not yet know its exposure.

A tabletop exercise that tests the plan

The cheapest way to find replacement gaps is a short exercise. Pick one critical supplier and assume a prohibition with a 30-day transition. Walk through who learns of it, who decides, what is bought, what is turned off, and what customers, patients or residents experience. Problems such an exercise is likely to surface include no alternative supplier pre-qualified, no exportable configuration and no authority to spend quickly.

Data portability deserves its own check. A hosted service that cannot export records in a usable format turns a supplier change into a data recovery project. Testing an export once a year, and keeping the result, is a small cost against that risk.

Approval paths differ by sector

Fast replacement depends on someone being able to say yes. In a bank, that may mean a technology risk committee and a spending authority within management, with the board informed afterward. In a health system, clinical engineering and patient safety committees must sign off on device changes. In a public agency, procurement law may require competitive bids, and emergency procurement powers vary by jurisdiction.

Each of those paths can be prepared. A bank can pre-approve a spending limit for regulatory replacements. A health system can pre-qualify alternative devices for its highest-risk categories. An agency can confirm in advance whether its emergency procurement authority covers a federal prohibition, and document that reading.

The cost no budget carried

Replacement under a deadline costs more than planned replacement. It compresses procurement, pulls staff from other work and often requires overlap licensing. Public agencies may feel this most, because an appropriation may carry no contingency for a forced swap. Health systems face clinical risk on top of cost. Banks face examination scrutiny if the change disrupts customer services.

The proposal did not price any of this for buyers. Its classification section described the action as significant under Executive Order 12866, but the burden on downstream users was left largely to the comment process. A regulated organization had to estimate its own exposure.

Where the argument is weakest

The rule might never reach most buyers

The strongest objection is scale. A case-by-case process run by one office could review only a handful of transactions. Most banks, hospitals and agencies would never receive a notice. On that view, the work described here was insurance against a remote event.

The objection is fair about frequency and wrong about consequence. A single determination against a widely used product would reach every customer at once. Those customers would not be parties to a negotiation with Commerce. They would simply have a deadline. The relevant probability was whether any product in the estate would be named, not whether the organization itself would be reviewed.

Existing programs already covered it

A second objection is that mature third-party risk programs already handled supplier risk. The proposal added little.

That holds for the risk rating and fails for the data. A third-party program focused on security and financial health would not capture ownership, control, jurisdiction and development locations, or track continuing transactions under old contracts. The proposal’s value was in exposing those gaps.

Supplier risk is the government’s job

A third objection is that national security screening belongs to the government. Buyers lack intelligence about suppliers and should not be asked to replicate it.

The objection is right about intelligence and wrong about readiness. No buyer can judge whether a supplier is directed by a foreign government. Every buyer can know which suppliers it depends on, what its contracts allow and how fast it can switch. Those are the facts that determine the cost of a government decision, whichever way it goes.

The diligence crowds out security work

A fourth objection is opportunity cost. Security and procurement teams are stretched, and provenance work competes with patching, monitoring and incident response.

The cost is real, which is why the sequence below starts with records that already exist and applies new effort only to the top-ranked suppliers. The same work also feeds incident response. Knowing who can push updates and who has remote access is among the first questions we would ask in any supply chain incident, whatever its origin.

Differences by kind of organization

A federal agency faced the Commerce proposal alongside federal acquisition rules such as the section 889 provisions. Its contractors had to make representations about covered telecommunications. Its own estate still needed the provenance work, because the Commerce rule could reach products the FAR did not.

A state or local government had no FAR obligations for its own purchases. It did have critical infrastructure functions, sensitive resident data and little budget slack. For a county, the replacement part of this argument carried the most weight.

A regional bank or credit union operated inside supervisory expectations for third-party risk. Its main exposure was likely to come through core processors, network gear and security software, which may be bought through intermediaries. Provenance through those intermediaries was the hard part.

A health system was likely to have some of the longest-lived and least replaceable technology: clinical devices, imaging systems and building controls. Its replacement problem could be the most severe, and its contracts with device makers among the least flexible.

Where the argument does not hold

For organizations whose critical suppliers were plainly domestic, with transparent ownership and limited subcontracting, the provenance and contract work was modest. The argument applies with most force where the estate is large, long-lived and bought through layers of resellers.

What to put in place first

Sequencing matters because the four parts depend on each other. Contracts cannot be fixed until the critical suppliers are known. Replacement plans cannot be priced until provenance shows which suppliers carry the most risk.

  1. Map continuing relationships. For each critical system, list the suppliers that update, support, host or access it, with the contract and renewal date.
  2. Collect provenance. Ask those suppliers for ownership, control, development and support locations, and embedded third-party components. Assign one owner for the data.
  3. Fix contracts at renewal. Add regulatory termination, disclosure and notice, transition assistance and refund terms to the renewal template for critical suppliers.
  4. Rate replacement difficulty. Record time, cost range and service impact for replacing each critical supplier, and name an alternative where one exists.
  5. Set a decision path. Agree in advance who can approve emergency replacement spending and how the board or oversight body will be informed.

Three measures show whether the program is working. The first is the share of critical suppliers with current provenance data on file. The second is the share of critical contracts with a regulatory termination right. The third is the share of critical suppliers with a recorded replacement estimate. Each should rise every quarter until it reaches the whole critical list.

Each step produces evidence that serves other purposes as well: third-party risk reviews, HIPAA risk analysis, federal supply chain duties and business continuity planning. That overlap is what makes the work defensible in a budget meeting before any determination arrives.

With hindsight

Since 2019, the proposal’s procedures have been replaced twice, and its core reach has survived. On December 23, 2019, Commerce extended the comment period to January 10, 2020. On January 19, 2021, it issued an interim final rule, effective March 22, 2021.

The interim rule answered a common request. Many commenters had asked Commerce to name the countries it considered foreign adversaries. It did: China, including Hong Kong; Cuba; Iran; North Korea; Russia; and Venezuelan politician Nicolás Maduro. It said the list could be revised immediately on publication, without notice or comment.

The interim rule also described six types of covered transactions. One was technology used by a party in a sector designated as critical infrastructure under Presidential Policy Directive 21. Another covered data hosting or computing services handling sensitive personal data on more than one million U.S. persons. Commerce declined calls for categorical exemptions, including for medical devices. It said wholesale exemptions would show adversaries which transactions escaped oversight.

Two smaller choices in the interim rule mattered to regulated buyers. Acquisitions by a U.S. person as a party to a transaction authorized under a U.S. government industrial security program were not ICTS transactions. And Commerce said transactions involving only personal hardware devices, such as handsets, did not warrant particular scrutiny. Neither carve-out helped a bank, hospital or county with its core infrastructure.

A widely felt use came in 2024. On June 24, 2024, Commerce published a final determination on Kaspersky Lab. From July 20, 2024, Kaspersky could not enter new agreements with U.S. persons for the covered products. From September 29, 2024, it could not provide antivirus signature and codebase updates to U.S. customers, among other activities. The determination also covered Kaspersky software integrated into third-party, white-labeled products. That was the 2019 proposal’s logic in action: the update, not the original sale, was the transaction stopped.

On December 6, 2024, Commerce issued a final rule finalizing the interim rule with changes, effective February 4, 2025. It removed the one-million-person threshold for sensitive personal data and added Macau to the definition of China. The regulations had by then moved to 15 CFR part 791. The final rule also recorded that the national emergency behind the order had been extended every year since 2019.

Commerce then shifted toward rules for whole classes of transactions. A connected vehicles final rule was published on January 16, 2025 and took effect on March 17, 2025. On January 3, 2025, an advance notice opened a similar inquiry into unmanned aircraft systems. A Federal Register search on November 14, 2026 found no later final ICTS rule.

For banks, health systems and agencies, the 2024 change to the data threshold mattered most. Under the interim rule, data hosting and computing services were in scope when they handled sensitive personal data on more than one million people. Removing that test brought smaller hosting relationships within reach, including the kind a regional bank, community hospital or local government might hold.

What held up was the claim that exposure lived in running relationships and updates. What did not hold up was the proposal’s refusal to name adversaries or define classes. By the January 2021 interim rule Commerce had done both, and by 2025 it was writing sector rules rather than relying only on individual reviews.

Spotted an error? Tell us and we will check it and correct the article where needed.

  • supply chain
  • Commerce
  • ICTS
  • third-party risk
  • contracts