Theranos: How Trust Without Proof Carried a Billion-Dollar Company

Why a Silicon Valley lab company matters to owners

Theranos promised to determine hundreds of blood values from a few drops of fingertip blood, quickly and cheaply. The company was founded in 2003 by then 19-year-old Elizabeth Holmes and valued at around nine billion US dollars in 2014. In 2018 it was dissolved. In between lie eleven years in which investors, partners, and board members trusted a promise that nobody outside the company had verified.

This article asks: How could trust without proof carry a billion-dollar company, and how could outsiders have recognized the missing proof? To answer it, we first recount the documented sequence, then explain signaling theory as a thinking model, weigh two readings, and derive a test for your own business. The article relies on court decisions, agency statements, and press reports.

What happened: the documented sequence

1. 2003 to 2013: building in secret

   Holmes drops out of Stanford University and founds the company. For years it works without publications in scientific journals. The board is filled with former ministers, senators, and military leaders, hardly any of them with experience in laboratory medicine.

2. 2013 to 2014: partnership and peak valuation

   The drugstore chain Walgreens offers Theranos tests in its stores, first in Arizona. In a funding round, the company is valued at around nine billion dollars.

3. October 2015: the newspaper report

   The Wall Street Journal reports that the company runs most tests not on its own devices but on modified devices from other manufacturers, and that employees doubt their accuracy. Several tips came from former employees.

4. 2016: authorities step in

   The US agency for health services finds serious deficiencies in the lab and imposes sanctions. Theranos voids two years of results. Walgreens ends the partnership.

5. 2018: securities regulator and indictment

   In March, the US Securities and Exchange Commission (SEC) accuses the company, Holmes, and former president Ramesh Balwani of a massive fraud through which more than 700 million dollars were raised. Holmes settles with the SEC without admitting guilt. The indictment follows in June, the company's dissolution in September.

6. 2022 to 2026: verdicts

   In January 2022, a jury finds Holmes guilty on four counts of defrauding investors and acquits her of the charges relating to patients. In November 2022 she is sentenced to eleven years and three months in prison, Balwani in December to twelve years and eleven months. The appeals court upholds the verdicts and restitution of around 452 million dollars in February 2025. In March 2026, Holmes' sentence is reduced by one year under a new sentencing rule.

The distinction drawn by the court itself is important. Holmes was convicted for deceiving investors, not for harm to patients; on those counts the jury reached no guilty verdict or acquitted. Anything said about the intentions and knowledge of individuals in this article rests solely on the court decisions.

The thinking model: signals instead of proof

In 1973, the economist Michael Spence described signaling theory. When one side knows more than the other, it sends signals meant to indicate its quality. A signal is credible when it would be costly or impossible to send for someone without the claimed quality. A certificate from a demanding exam is a strong signal; a confident appearance is a weak one.

Theranos sent many signals, but almost only weak ones. A board of former ministers proves political connections, not medical quality. A partnership with a drugstore chain proves that the partner was convinced, not why. A high valuation proves that others invested. The strong signal, an independent verification of measurement accuracy, was missing. Each weak signal became the next party's reason to skip its own check.

An everyday analogy: when buying a used car, people like to be reassured by the friendly salesman, the well-kept dealership, and the hint that a well-known neighbor also bought there. None of this proves the condition of the engine. Only a garage chosen by the buyer does.

A signal that anyone can send proves nothing, no matter how many believe it.

How employees wavered between resistance and complicity

Young companies have flat hierarchies and short paths. One would expect problems to surface faster there. In a case study spanning 15 years, Crivelli, Hess, Shepherd, and Wincent (2026) examine why this did not happen at Theranos. Their finding: founders shape how employees respond to irregularities, eliciting both resistance and complicity. Employees do not judge events once but waver over time between condoning and condemning.

The authors stress that fraud in young companies arises differently than in established ones. There, routines, benchmarks, and audits anchor judgment. In a young company, much is open; what counts as right is negotiated in the moment, and the founder has great influence on that interpretation. In a review of startup fraud, Gleason, Kannan, and Rauch (2022) reach a similar conclusion: the attitude that success can be faked until it arrives creates opportunities that an early independent audit would close.

That the case was eventually exposed was largely due to former employees who turned to journalists and authorities. This matches research showing that tips from within a company are the most common way fraud is discovered. Whistleblower protection describes how a business receives such tips and protects those who give them.

Two readings: a lone perpetrator or a failure of the controllers?

Both readings have evidence. Individual responsibility has been established in court and confirmed on appeal. At the same time, the case shows how little the usual controls took hold. The board was prominent but lacked relevant expertise. This fits a finding by Nili (2022): independent controllers on boards are often independent in name only and have neither the powers nor the distance their role demands.

I consider the second reading the more instructive one, without weakening the first. Deception is a decision made by people. That it held for eleven years was a consequence of many clever participants taking a signal for proof. Enron 2001 shows the same pattern of pressure, opportunity, and rationalization, and the FTX case a similar failure of control.

The signal-or-proof test for your business

A small business rarely invests in lab companies. But it checks promises every day: from suppliers citing a certification, from software vendors with well-known customer names, from applicants with impressive references. And it sends signals to its own customers. From the case I derive the signal-or-proof test. For every promise, it asks whether it is a signal or proof.

1. Name the promise

   Write down in one sentence exactly what is promised: delivery reliability, measurement accuracy, savings. Vague promises cannot be checked.

2. Separate signal from proof

   Note what speaks for the promise and ask of each point: could someone without this performance also show it? Well-known customer names, prices, and appearances usually fail the test.

3. Demand independent proof

   Ask for evidence the provider does not control: a reference you call yourself, a third-party test, a trial under your conditions. Due diligence on business partners shows how to vet partners beyond that.

4. Question secrecy

   If a provider grants no insight, ask why. Protecting know-how is legitimate; dispensing with verifiability altogether is not.

The same applies to your own oversight. An advisory board of respected people is valuable if it may ask questions and sees the figures. Advisory boards for SMEs describes how to set one up this way.

Transferability and limits

Theranos shows a rule that applies beyond Silicon Valley: the more participants trust a promise, the less each individual checks. This is convenient and dangerous, because the check everyone assumes the others made is ultimately made by nobody. Trust as an economic factor discusses why trust is nevertheless an economic factor and how it can be proven.

The limits of this case study are clear. It relies on court decisions and press reports, not on its own access to documents. The study by Crivelli and colleagues rests on a single, exceptionally well-documented case. How often the pattern occurs in smaller companies cannot be derived from it.

Conclusion: trust needs proof that others can check

To the guiding question: Theranos could raise billions because a chain of weak signals covered up the missing proof. Prominent controllers, a large partner, and a high valuation reinforced one another, and secrecy prevented anyone from verifying the technology independently. The missing proof was recognizable at exactly this point: there was no independent verification, and nobody insisted on one.

The lesson for a business is sober. Trust does not arise from the number of those who trust, but from proof that someone can check.

• For your three most important suppliers or service providers, note which promises are proven and which are only signaled.
• For an upcoming decision, demand at least one piece of evidence the provider does not control.
• Check your own sales materials for whether they contain proof or only signals.
• Clarify whether your advisory board or tax advisor may inspect documents directly.