Change of control review, explained.

What change of control review means: reading every signed contract a takeover can end or change

1. The deal team reads the big contractsBefore signing, under confidentiality. Licenses behind lead programs should surface here.
2. The deal closes and clocks startNorthwind closes on Kelvara on September 14, 2026. Day 0 for most of what follows.
3. Every inherited contract gets an outcomeConsent, notice, no action, a conflict, or a lawyer's read, with the clause and the clock. This guide is about this step.
4. Letters are drafted, signed and sentFrom approved wording, by the person with authority, to the address in the notices clause.
5. Replies come back and are recordedConsent, a condition, a refusal or silence. Each one has a next step before the window closes.

Four things a first-time reader of takeover contracts gets wrong

The four terms to learn before opening an inherited contract

The one-page reference in section 24 defines all 14 terms.

Anti-assignment, change of control and deemed assignment: six clauses that read alike and act differently

In shortAn anti-assignment clause protects against the contract moving. A change of control clause protects against the owner changing. Many contracts carry the first, fewer carry the second, and the review turns on telling them apart and on reading the definitions that sit pages away.

Public data gives a rare view of how common each term is. The Contract Understanding Atticus Dataset is 510 commercial contracts filed with the US Securities and Exchange Commission, labeled clause by clause under lawyers' supervision.3

We counted its published labels. Nearly three in four contracts restrict assignment. About one in four has a change of control term, which in that dataset means a right to terminate, or a consent or notice duty, when a party changes hands. Of those 121 contracts, 119 also restrict assignment, so the two usually arrive together and the reader has to work out which one applies.4

The dataset covers all industries and public companies' filings, so it says nothing exact about a biotech's estate. Your own contracts may differ widely. It is useful for the proportions and for practice: it is free, and every clause is labeled.

The dataset also shows how few contracts spell it out. Of 374 anti-assignment passages, a word search finds "operation of law" in 65 and "merger" in 75.4 Passages with neither phrase leave the reader to work out from case law what a merger does to them.

Stock purchase, forward merger, reverse triangular merger, asset purchase: which clauses each one wakes up

In shortThe question under most assignment clauses is whether the contract moved to a different legal entity. In a stock purchase or reverse triangular merger the target stays the party. In a forward merger the target disappears. In an asset purchase each contract has to be moved. A change of control clause turns on its own definition, which often looks only at who owns the shares.

Delaware's merger statute shows why survival matters. As the Court of Chancery read Section 259, the rights of a company that stops existing pass to the survivor, and the wording "suggests that the surviving corporation would not have effected any assignment".1

German practice lands in a similar place for a share deal: "Third-party consents are generally not required unless relevant change-of-control provisions are triggered."6 We did not read primary texts for other countries, and contracts in the sample run under English, Ontario and Irish law among others. Take local advice for each.

The size of that swing can be large. In the fictional Kelvara estate, 1,184 contracts carry a plain anti-assignment clause and stay at no action because the deal was a reverse triangular merger. Had the same deal been a forward merger or an asset purchase, each of those would have needed its own answer. We found no public share of biotech takeovers by structure.

A contract can also be silent and still matter. Some licenses of patents and copyrights do not transfer without consent even if they say nothing, because of federal rules on intellectual property. The next section covers those opinions and what each one decided.

Meso Scale v. Roche and the license cases: what each opinion decided, and where it stops

In shortOne Delaware trial court opinion anchors the view that a reverse triangular merger generally is no assignment. A separate federal line says many IP licenses cannot move without consent, but those cases involved a licensee that disappeared, a bankruptcy, a direct transfer or an internal reorganization. Neither line closes the question everywhere, so IP licenses go to a lawyer.

What Meso Scale decided about reverse triangular mergers

Roche acquired BioVeris in 2007 by reverse triangular merger. An earlier agreement said it could not "be assigned, in whole or in part, by operation of law or otherwise" without consent, and Meso Scale, a party to it, said the merger needed its consent.1

In 2011 the court refused to dismiss the claim and noted that no Delaware case had squarely decided the point.12 In February 2013 it granted summary judgment to Roche on that count, "on the basis that the reverse triangular merger was not an assignment by operation of law or otherwise".1 A summary judgment is a ruling without a full trial.

The reasoning has three parts. The target survives, so it "would not have effected any assignment". Delaware already treats stock purchases as no assignment, and "the law should reflect parallel results". And the parties chose their words: "Meso could have negotiated for a 'change of control provision.' They did not."1

The court also declined to follow SQL Solutions v. Oracle, a 1991 federal decision that applied California law to a software license, and the one case the plaintiffs had cited going the other way.12 It said that case "provides no further explanation for its apparent holding".1

SQL Solutions is unpublished and we did not read it, so we describe it only as the two Meso Scale opinions do. These opinions cannot tell a reader what a California court would do today, so a contract under California law, or any other state's, is a question for counsel.

Licensors know all this and draft around it. NIH's model exclusive license, revised in July 2026, bars transfer "by legal process or by operation of law", states "that the identity of the parties is material to the formation of this Agreement", and on an approved assignment takes an extra royalty of 1% of the fair market value of what the licensee receives.5 A reader who has seen that one clause learns to look for those three pieces in any license.

"And its Affiliates": when the target's exclusivity binds the buyer's own programs

In shortA covenant that binds a company and its affiliates may start to bind the buyer at closing. Whether it does depends on the definition: New York's top court reads "affiliate" as present affiliates unless the words look forward. If it does reach, the live question is whether the buyer's program overlaps, and whether the contract offers a carve-out for programs the buyer already had.

In Ellington v. EMI Music, a dispute over royalties under a 1961 agreement, New York's Court of Appeals wrote: "Absent explicit language demonstrating the parties' intent to bind future affiliates of the contracting parties, the term 'affiliate' includes only those affiliates in existence at the time that the contract was executed".2

The Federal Circuit, applying New York law five years earlier, read forward-looking words the other way. A patent cross-license defined a subsidiary as one in which a party "now or hereafter has" a majority stake, and the court said the phrase "connotes that Subsidiaries may come into existence at some unspecified future time". Two companies that became subsidiaries after the agreement's term ended were held to be licensed.13

Neither case is a takeover of a drug company, and both turned on their own contracts. Together they show why two words matter. The sample Aarenthal agreement, under New York law, says "Affiliate" means an entity "whether now or hereafter existing", so on those words it reaches Northwind. The same covenant with a plain "Affiliates" might not. Ask counsel which you have.

The filed-contracts dataset gives one view of how often this arises: 231 of 510 contracts carry exclusivity or a non-compete. Of those passages, a word search finds "affiliate" in 55. Two of the 55 also use the word "hereafter", and only one of those two speaks of future entities.4 Biotech partnering contracts may run higher, and we found no public count.

The word search looks at the labeled passage only. A definition elsewhere in the contract is missed, so treat these as lower bounds. They show why the definition has to be read, and say little about how often it looks forward.

Testing one covenant against one program: mechanism, indication, territory, timing

Once a covenant reaches the buyer, the test is plain comparison, row by row, and most of the work is having an accurate list of what the buyer has promised and is developing. Here is the sample's worst case. Kelvara's co-development agreement with Aarenthal Pharma bars Kelvara and its Affiliates from any product against a named target in inflammatory bowel disease. Northwind has Veltrimab, an antibody against that target, in Phase 3.

Real contracts carry carve-outs like this. Biotech license agreements filed with the SEC show covenants that spare an acquirer's existing programs, on condition that they stay apart from the licensed work, a separation lawyers call a firewall. We name none of them here, because a sample in this guide should not point at a real agreement.

Contracts without a carve-out are harder. The sample Pellham fill-finish agreement makes Kelvara and its affiliates buy 80% of their US pre-filled-syringe filling from Pellham, says "There is no exception for any entity that becomes an Affiliate", and charges a shortfall fee. On those words the buyer's own filling line counts, and the fix is to ask for an amendment.

Other affiliate-reaching terms work the same way: a most-favored royalty term can capture the buyer's out-licenses, a right of first negotiation can capture its assets. Five percent of the filed contracts carry a most favored nation term.4

From the data-room file to an outcome: five questions every inherited contract must answer

In shortFind the executed version with its amendments. Find the clauses and their definitions. Apply the deal structure and the governing law. Check any covenant against the buyer's commitments. Give the contract one outcome with its clause and clock, and send anything uncertain to a lawyer.

Four Kelvara contracts, four different answers from one deal

How one outcome can be wrong in each direction

A false "no action" is the dangerous error, because nobody looks again. A false "consent" has a real cost too. Asking a counterparty for consent you did not need tells them they may have leverage, and they may price it. Good review tracks both, and it records the reason for every no-action result, so that a later reader can check it. In the Kelvara estate, 2,510 contracts end at no action and each has its reason on file.

Day 30, Day 60, Day 90: counting consent and notice deadlines from the closing date

In shortThe contract sets each clock, and in the fictional deal the closing date starts them. Count the days the way the contract says, send early enough for a reply, and read a termination window carefully: it usually says how long the other side has to use its right, which means you need the answer before the window ends.

Six ways a consent or notice date goes wrong

The notices alone can swamp a small team. In the fictional deal, 119 notices are still to draft on Day 23, all due by Day 30. Nothing about them is hard. The risk is that nobody has the hours in that week, while the same counsel is deciding conflicts.

Licenses, research contracts, supply, software, leases: what each contract family tends to carry in a biotech's estate

In shortRisk is uneven. In the fictional estate, a handful of licenses and supply agreements carry the hardest clauses, while most confidentiality agreements and site contracts need nothing. The review should spend counsel's hours accordingly, and record why the rest were cleared.

A sample deal, for shape only. No public source reports these shares for a real estate. A CRO is a contract research organization, a company a trial sponsor hires to take on some of its trial work.

Employment agreements and retention letters carry change of control terms too, on their own legal rules. They are usually a separate workstream run by HR and employment counsel.

Who may write to a counterparty, and when: antitrust before closing, privacy and product ownership after it

In shortWe found no statute that sets the consent clocks: contracts do. Other rules shape the letters. Antitrust law stops the buyer running the target's business before closing, and we found no source on whether consent letters cross that line. GDPR, the EU's privacy law, governs processors, companies that handle personal data on another's behalf, and any notice of new ownership comes from the contract. On our reading, a reverse triangular merger leaves approvals with the same legal holder.

Before closing: the limit on a buyer, and no source we found on consent letters

A buyer that takes control of a target's commercial decisions before antitrust clearance risks a penalty. In the US the Hart-Scott-Rodino Act imposes a filing and waiting period on deals above $133.9 million, the threshold from February 17, 2026, subject to further tests.18,19 The penalty runs up to $53,088 a day, a level the FTC has kept for 2026.17 EU law has its own standstill and fines.20

In January 2025 three oil companies paid a record $5.6 million, for 94 days of conduct, in which the FTC says the parties coordinated on the target's customer contracts, relationships and deliveries, and on prices.16 The complaint, as Simpson Thacher summarized it, described the target's staff passing contract details to the buyer, and the buyer's staff dealing directly with the target's customers.21

We found no source that addresses whether a consent request sent before closing is gun jumping. One antitrust lawyer's summary describes gun jumping as pre-closing conduct inconsistent with the waiting period, or with competitors continuing to compete, and it does not mention third-party consent letters.22 Antitrust counsel should decide the point before any letter goes out.

Deal agreements handle consents in their own terms. One filed agreement in a public clause library has buyer and seller use "reasonable best efforts" to obtain a named consent still outstanding after closing, and says the failure alone does not fail the buyer's closing conditions, with one stated exception.23 We found no public figure for how often consents are closing conditions or promises of effort.

A team can hold all outreach to the long tail until closing, as a simple rule anyone can follow, and the fictional example does. That is a policy choice, and your antitrust counsel should write yours.

After closing: privacy and regulatory ownership

What is left is ordinary contract law, plus two working rules. First, one person with authority signs. Second, one set of approved words goes out, because a letter that concedes too much, or says a consent was required when it was not, becomes evidence.

Why inherited contracts weigh more in 2026: larger biotech takeovers, for drugs that often began elsewhere

In shortDeal value is high and rising. Large drug companies get most of their new drugs from outside, so what they buy can rest on other parties' licenses. Each deal hands the buyer an estate of contracts on a short clock.

IQVIA counts 42 deals in the first half of 2026, 84% of its 2025 total, at an average of $3.1 billion.26 EY's report, as covered by BioXconomy, counts $196 billion for the half, up 140%, with $147 billion of it in biopharma.27 Never compare the two without saying which scope you mean.

These figures show the shape of contract risk and give no direct measure of it. When what you bought began somewhere else, somebody else's license terms sit underneath the price you paid. The study above does not say how many of those licenses had a change of control term, and we found no source that does.

Why missed consents rarely become public, and the nine points where inherited-contract review fails

In shortWe found no public count of missed consents. Contracts are private, and a dispute settled quietly leaves no record. So the evidence we can offer is a list of failure modes, drawn from how the clauses work and how the reported cases turned, for you to test against your own last deal.

We can give no public figure for how often licensors terminate, reprice or ask for fees after a takeover, or what it costs. We looked in deal studies, legal surveys and trade press. Your own last two deals will tell you more: count the contracts found after Day 30, and what each one cost to fix.

What language models get right and wrong on change of control clauses, and which rules apply

In shortThe best F1, a score that balances misses and false alarms with a maximum of 1.0, was 0.644 in one 2025 test of clause finding, so a human still has to check the result. Legal ethics and privilege are the live constraints. Quality rules for making medicines and medical devices mostly sit outside this work.

That benchmark, ContractEval, tested four proprietary and 15 open models across all clause types. Its scoring counts a clause as missed if the model does not fully cover it, which is strict. The authors also report models, open ones more often, answering "no related clause" when one exists.31

For this work that is the worst way to fail, since "no clause" means "no action". A separate study of legal research tools documents invented answers in that adjacent task.32 So measure a tool's miss rate on your own contracts before trusting a no-action result, and keep a sample of those results under a lawyer's eye.

An estimated 35 working days on one inherited estate: how legal, alliance and procurement teams clear it today

In shortOne common pattern, though no public study measures it: outside counsel or a contract team reads the long tail after closing into a spreadsheet, while alliance and procurement leads chase their own counterparties. In the estimate below, hours go to finding clauses and applying the deal's rules, ahead of the judgment calls that need a lawyer.

Six roles that carry a contract review after closing

Outside counsel, an in-house team, or software reading first: how each staffing choice plays out

We found no public split between these three, and no public benchmark for contracts in an acquired biotech or for the time a first pass takes.

Day 23 at Northwind: the inherited estate, a license at risk, nine consent requests and one conflict decided

In shortLena Ortiz, integration counsel, opens the estate on Day 23. Agents read all 2,846 contracts between September 15 and 18. She confirms the Aldermoor outcome, nine consent requests are drafted in one step, Omar Haddad signs them, and he records his decision on the Aarenthal conflict a week before its clock runs out.

1

The whole estate on one screen, sized by value and colored by what the deal triggers

Every contract that needs action is a tile. Red means consent, violet a conflict with Northwind's commitments, blue a notice, amber a lawyer's read. Each family's no-action contracts form one hatched block. The chips count it: 74 consents, 9 conflicts, 212 notices, 41 lawyer's reads, 2,510 with nothing to do, 336 needing action in all.

The line under the map gives the money: $852M a year in the estate and $172M behind the 74 consents, 15 of them still open. The screens size the Aldermoor license at $26M a year. Selecting it shows the clause and the clock beside the map.

2

The Aldermoor license beside its reading, with every reason tied to a clause

Lena opens the agreement. §14.2, §14.3 and §14.4 are highlighted on page 21. The proposed outcome is Consent needed, 97% sure, with the deadline of November 13, 2026 and four numbered reasons, each linked to its passage. She can confirm, change the outcome, or send it to a lawyer.

A careful reader will note what carries this result. Northwind bought 100% of Kelvara's voting shares, which meets §14.3(a), and §14.2 names a Change of Control. The words "by operation of law" in the same sentence would not have been enough on their own under Meso Scale, and "by merger" alone would be a question for counsel.1 And §14.2 asks for prior consent, so this request is late: it asks the foundation to consent after the fact and to leave its right unused.

3

The consent runway: 74 requests against their termination windows

Below the estate, each consent request runs from closing to its counterparty's deadline, with today's line through them all. On Day 23: 27 granted, 32 waiting, 6 with the signatory, 9 to draft. A hollow dot is a draft, a solid dot a signed and sent letter, a tick a reminder, a green check a reply.

Thornbury's foundation has waited 11 days and was reminded on October 3. Merrow Rise's foundation granted consent on October 7. The largest window with nothing out is Aldermoor's.

4

"Draft the 9 requests" routes every open consent to the signatory

Lena drafts all nine in one step. The screen shows what ran: nine outcomes she had confirmed, each above the 85% confidence line; nine letters from the approved license and commercial templates, each quoting its own clause and notice address; a check that the wording is unchanged and nothing goes unsigned; and routing to Omar Haddad, earliest window first.

The summary line now reads 15 with the signatory and 0 to draft.

5

The consent request quotes §14.2 and goes to the address in the notices clause

The letter comes from "Consent request, license v3", approved by Legal. It goes to the Office of Technology Licensing named in §20.1, says Kelvara remains the same legal entity, quotes §14.2, and asks for a countersignature by November 6, a week before the window closes. Blue underlines mark the fields filled from the agreement. Omar signs consent requests.

6

Aarenthal's covenant reaches Veltrimab, and Omar Haddad decides

§7.1 bars Kelvara and its Affiliates from any product against the collaboration target in inflammatory bowel disease, and §1.2 says Affiliate means any entity "whether now or hereafter existing". The overlap table compares mechanism, indication and territory, then tests the §7.4 carve-out. Veltrimab existed at closing and the data room shows no shared people or data, so the carve-out is open if a written notice goes by October 14 and a firewall is in place.

The agents never decide this. Omar weighs relying on the carve-out, asking for a waiver, or sending it to the deal team. He records the first. The acquirer-program notice goes to him for signature and the firewall memo to the program team.

7

The decision goes on the record with its author and what follows

The decision card replaces the options: who decided, what they decided, and what was drafted because of it. The recommendation came from Lena. The decision is Omar's, and it stays in the audit trail.

The screens counsel works from: contract list, conflict cards, lawyer's reads, letters board, steering dashboard

In shortEach person works from the part of the deal they own. Counsel works the list of 336, the alliance and procurement leads follow their counterparties, the signatory sees what needs a signature, and the integration office reads one view for the steering committee. Doubtful wording goes to a lawyer with its reason.

From clause finder to signature: which contract steps belong to agents and which to named lawyers

In shortSix agents do the reading and drafting. People confirm every outcome that needs action, decide every conflict, read every IP license and everything below the confidence line, and sign every letter. Every step lands in the audit trail.

Who confirms, decides and signs on the Kelvara contracts

From data-room export to signed consents: what the review takes in, what it produces and what it connects to

In shortFour inputs from the deal, five outputs for the people who carry it. It works beside your data room, e-signature, contract repository and portfolio system. Roles, maker-checker and the audit trail apply throughout, and the design targets below are goals to measure in your own run.

Four days, 336 contracts, six minutes: the design targets on one inherited estate

These are goals and estimates on fictional data. Measure them on your own contracts before you rely on them (section 22).

Systems a change of control rollout connects to

Access runs on roles, so the people on a deal see that deal. The company's information security program is SOC 2 Type II audited and managed to ISO 27001. A contract review is legal work, so the controls that matter most are confidentiality, who may read which contracts, and an audit trail you can hand to counsel. Keep your counsel in charge of privilege questions (section 12).

The visible cost of change of control review is reading hours; the hidden one is a license or supplier in someone else's hands

In shortThe first cost is plain: lawyers' hours to read every inherited contract, many of which may need nothing. The large cost is rare and hard to see: a counterparty with leverage on a license or a supply line behind the program you bought. No public source measures it, so you will have to build your own number.

The time counsel spends confirming does not fall, on purpose. That is the part you want to keep.

Where the cost of an inherited-contract review hides, and how to find yours in a week

What a late consent costs beyond the invoices: trust, leverage and counsel's attention

How automating the first pass pays back in counsel hours, kept clocks and fewer surprises after Day 30

In shortMost of the measurable return is reading hours moved from every contract to the ones that need a lawyer. Avoiding one serious delay can outweigh that, and it cannot be promised in advance. The calculator keeps the two apart and uses your numbers.

Reading time comes first: four ways a change of control review pays back

Your first-pass cost on one inherited estate, and what a kept schedule is worth

Formulas: first pass = deals × contracts × minutes ÷ 60 × hourly cost. Confirming and sampling = deals × contracts × (share needing action × minutes to confirm + share with no action × share sampled × minutes per sample) ÷ 60 × hourly cost. Saved = the difference. Delay = days × direct daily cost. Lost sales of about $500,000 to $800,000 a day stay out of the total on purpose, since they would swamp it and depend on the drug. The software's price is not in these figures. Ask for it and subtract it.

Eight questions that show whether your deal team can automate contract review: estate size, playbook, commitments register, counsel

In shortScore your team from 0 to 2 on eight questions. The ninth is for reflection. The meter adds your score, and each question says why it matters and what a strong answer sounds like.

Testing a contract review vendor: misses on your own files, deal structure, covenants, letter wording

In shortAsk for measured misses on your own contracts, how the tool treats deal structure and IP licenses, whether it checks covenants against your register, and who controls the wording that leaves your building. Ask us too.

Test a tool on a deal you already closed: eight weeks, with your final outcome list as the answer key

In shortTake a deal you closed, with its contracts and final outcome list. Run the tool over the contracts with your structure and rules, and compare. The test that matters is how many of your real consents and conflicts it would have missed.

1. Choose a closed deal with at least 500 contracts and a final list of outcomes, and clear the use of those contracts with counsel.
2. Write the rules you would use today: structure, governing law, IP licenses, confidence line, signatories.
3. Replay the contracts, blind to the final list, with your register as it stood at closing.
4. Compare every result with your list, and have a lawyer read each disagreement. Your list may be the one that was wrong.
5. Decide with three numbers: misses on no-action results, false alarms, and counsel minutes per confirmed contract.

Numbers that show a change of control pilot worked

Twelve answers on change of control clauses, from reverse triangular mergers to privilege

In shortEach answer stands on its own. All of them are general information. Take your own counsel's advice on your own contracts.