What Counts as an Asset in the 23andMe Sale?
In the 23andMe bankruptcy, the hard question was not whether “data” has economic value. It plainly did. The harder question was which interests in that data belonged to the bankruptcy estate, which remained connected to customers, and how a buyer could receive the business without receiving unlimited authority over people’s biological information. Separating those layers makes the sale easier to understand: the estate could transfer its business-held interests, but that did not automatically extinguish customer control or convert genetic information into an ordinary commodity.
The word “data” hides several different assets
The bankruptcy court described 23andMe’s customer-related holdings in both physical and electronic terms. Customers sent saliva samples for testing. The company maintained those samples in biobanks, produced genetic results, and kept account information such as names and email addresses. These items were connected by the company’s service, but they were not interchangeable.
A saliva sample is a physical biological material. A genetic result is an analysis derived from that material. An account record is an identifiable relationship between a person and a service. Each can be stored, transferred, deleted, or exposed through a different operational process. Treating all three as rows in a database would hide the decisions that matter: who may handle a sample, who may interpret a result, and who may link an account to a person.
The business also involved research participation. More than 80 percent of customers had elected to participate in 23andMe’s medical-research programs. The court noted that data provided to researchers was deidentified, meaning it was not associated with individual customers, and that customers could opt out of medical research at any time. That creates a meaningful distinction between research data as supplied under a particular program and the broader set of customer information held by the company. Deidentification describes how one category of information is presented to researchers; it does not answer who controls a customer’s account, what happens to a saliva sample, or what a successor may do with identifiable records.
Finally, the company’s value was not located only in the information itself. The asset package included the operating business: the employees, service processes, customer relationships, and organizational capability that turned samples and results into a continuing service. The court later emphasized that TTAM expected to operate in essentially the same way, with the same limited group of people accessing customer data. The business therefore had both informational assets and the machinery of custody. A buyer was not merely purchasing a file; it was acquiring the position of a custodian that could continue collecting, analyzing, storing, and responding to customer requests.
That distinction is important for two reasons. First, it explains why the data had value in the bankruptcy sale: it was embedded in a functioning business rather than detached from its service. Second, it explains why the sale raised privacy concerns: the value of the business depended on continuing relationships with people who had not agreed to become anonymous inventory.
How the asset bundle connects to the transaction
The following model separates the customer relationship, the company-held assets, the new corporate structure, and the controls customers retained. It is not a claim that each relationship has the same legal character. Its purpose is to show where interests meet and where they diverge.
The diagram’s most important line is not the one connecting Newco to TTAM. It is the separation between the estate’s interests and the customer’s controls. In the selected transaction structure, Newco would receive the debtors’ assets, including customer data, in exchange for a promissory note. The debtors would then sell all of Newco’s equity to TTAM Research Institute. After the two steps, TTAM would own the subsidiary that housed the data and other assets.
That structure changed legal ownership and corporate control even if the information remained in the same systems or the same employees continued to handle it. Physical location is therefore an unreliable guide to whether an asset interest has moved. A record can stay on a server while the entity entitled to administer the business changes. Conversely, a transfer of an estate’s interest does not necessarily mean that the information has been newly disclosed to a different population of people.
The distinction also explains why the court analyzed the transaction as a transfer of the debtors’ interests rather than as a sale of every right associated with every customer. The bankruptcy estate could sell what the debtors owned. It could not simply manufacture new customer permissions by listing the data among estate assets.
The same distinction explains why the court treated the operating business as part of the value rather than as a wrapper around the “real” asset. A company that holds genetic information also has to maintain the systems, personnel, policies, and customer-facing processes that make the information usable. Those arrangements are where deletion requests, research choices, and access restrictions become real. If a buyer received only a database without the operating capacity or the obligations that accompanied it, it would be acquiring something different from the going concern that was offered. In the 23andMe case, the proposed continuity of employees and service functions supported the argument that TTAM would remain a custodian of the same relationship, while the promised privacy enhancements supplied reasons not to treat continuity as a complete answer by itself.
The two-step structure also shows why consideration and control should not be confused. Newco’s promissory note and the later transfer of Newco equity were the financial mechanics by which the estate obtained value and TTAM obtained ownership. Neither step, standing alone, answered what customers could delete, what research participation meant, or whether a new use would require consent. Financial consideration can explain why an asset transfer is economically justified; it does not define the scope of the customer’s rights.
The estate’s interest is not the customer’s entire relationship
The court’s discussion of Texas law makes the boundary especially clear. Texas law, as summarized in the opinion, gives an individual a property right in the biological sample provided to a direct-to-consumer genetic-testing company and in the results of testing or analysis conducted on that individual’s DNA. The right includes control over matters such as collection, use, retention, maintenance, disclosure, and destruction. Texas also treats the results as confidential.
The court did not accept the idea that the bankruptcy sale would transfer those customer interests away. The debtors said they were selling only their own property interests, and the court found no record that they intended to sell more. If a Texas customer had a property interest in a sample or result before TTAM took possession, the court reasoned, that interest remained the same afterward. The purchaser acquired the company’s position as custodian and operator; it did not acquire the customer’s personal legal identity or a power to rewrite the statute.
That conclusion is narrower than saying that customers own all genetic data everywhere. The court was interpreting a particular Texas provision in a particular transaction. Other jurisdictions may define property, confidentiality, consent, or control differently. The point that travels beyond Texas is methodological: before asking whether a bankruptcy estate can transfer an asset, identify the interest that the estate actually possesses. A company’s control over a record and a customer’s legal control over a sample are not automatically the same thing.
Customer controls also have an operational form. Before the sale, 23andMe customers could delete their accounts, direct the company to dispose of saliva samples, and opt out of medical research. The court said approximately 1.9 million customers had deleted their accounts after the Chapter 11 cases began. Those actions show that “control” was not merely an abstract promise in a policy document. It was tied to requests that affected what the company could retain and how it could continue a relationship.
The court treated account deletion as the clearest example of a customer affirmatively exercising control later in the relationship. It did not say deletion was the only possible form of control, nor did it say that silence automatically granted the company permission for every future activity. It rejected a different proposition: that doing nothing necessarily required the company to obtain fresh express permission before every change in corporate structure. That distinction becomes central when consent rules are examined later.
Deidentified research information does not settle the ownership question
Research participation adds another boundary. The court recorded that information provided to researchers was deidentified and that customers could opt out. Those facts helped describe the company’s existing practices, but they did not answer the larger sale question.
A research dataset supplied without individual identifiers is not the same operational object as an account linked to a person’s name and email address. Nor is either one identical to a saliva sample stored in a biobank. The fact that one category is deidentified does not prove that every customer-related asset is deidentified, and it does not establish that a change in business ownership has no effect on the customer’s choices. Conversely, the existence of a customer right to opt out does not mean that the estate owns no transferable business interest at all.
The useful approach is to follow each control to the asset or process it governs. A research opt-out concerns participation in research. An account-deletion request concerns the customer relationship and associated records. A direction to dispose of a sample concerns physical biological material. They can be carried forward by a successor, but they are not interchangeable permissions.
This is also why the court’s commitment to preserve deletion and research opt-out rights mattered to the sale analysis. Those rights were part of the practical continuity promised by TTAM. The buyer was not presented as receiving a clean slate. It was expected to operate the same business while honoring the petition-date privacy policy and adding protections, including limits on certain future transactions and a commitment not to share personal information with insurance companies.
Why the boundary matters before the next question
Three mistaken conclusions are easy to draw if the asset bundle is not separated.
The first is that customers’ rights block any transfer because the information is “theirs.” The case does not support that categorical answer. The estate had business-held interests that could be sold, and the court found that the proposed transaction did not transfer away the customer property interest recognized under the Texas law before it.
The second is that the estate could sell the information without restriction because the information was “its asset.” The court rejected that frame as well. Customer rights, privacy policies, state statutes, and the purpose and structure of the transaction remained relevant. An estate cannot create broader authority merely by giving a sensitive record an asset label.
The third is that no privacy issue exists if the bytes do not move or no new employee opens the file. The court treated the transfer into Newco as a transfer in legal form even though the data was expected to remain with the same operating team. At the same time, it distinguished that transfer from a new disclosure to people who previously lacked access. The two propositions can coexist: a custodial interest can change without a new audience receiving the information.
The practical test is therefore more precise than “who owns the database?” Ask four questions instead. What does the company hold? What can the estate transfer? What does the customer retain? And what will the new custodian be allowed—and expected—to do? Those questions connect the physical sample, the electronic result, the account relationship, the research choice, and the operating business without collapsing them into one asset.
Once that boundary is clear, the next issue is transaction design. The 23andMe case did not use a single uncomplicated conveyance. It used an Equity Toggle that moved the assets into Newco before TTAM acquired the subsidiary, while preserving a Plan Toggle that would have changed ownership without moving the assets out of the debtors. Understanding why those alternatives mattered is the next step in judging whether the bankruptcy process changed custody, privacy, or both.