Bad Blood

By John Carreyrou

10 min read

Why does it matter? Because the most dangerous frauds are the ones that make skepticism feel immoral.

The natural assumption is that a $9 billion fraud built on blood tests that didn't work must have been obvious to anyone paying attention. It wasn't. Dozens of people paid very close attention: engineers who watched the machines fail, scientists who flagged the numbers, a CFO who said "we've been fooling investors" to Elizabeth Holmes's face. Every single one of them was fired, surveilled, sued into silence, or threatened through their families. Bad Blood is the account of how Silicon Valley's foundational myth — fake it till you make it — migrated from software into medicine, where the collateral damage isn't disappointed customers but patients receiving wrong diagnoses. It's also the story of every institution that should have stopped it: boards, regulators, Walgreens and Safeway, George Shultz — each of them recruited, one by one, into protecting the machine instead.

The Vision Was Genuinely World-Changing — That Was the Problem

The summer of 2003, Singapore. Elizabeth Holmes, nineteen, was at the Genome Institute of Singapore during the SARS outbreak, collecting blood samples the way medicine had collected them for centuries: syringes, nasal swabs, the routine discomforts of diagnosis. She flew home to Houston convinced there had to be a better way.

She sat down at her computer and didn't stop for five days. Her mother brought food on trays. She slept an hour or two a night. When she finished, she had a patent application for an adhesive arm patch that would draw blood painlessly through microneedles, analyze it in real time, and beam the results wirelessly to a physician, diagnosing and treating at the patient's home, no needle required.

The device was almost certainly unbuildable, and she would spend the next decade quietly walking back its ambitions. But she brought it to Channing Robertson, Stanford's star chemical engineering professor, who had been teaching there since 1970. In later court testimony, Robertson said he had never encountered a student like her before: she had synthesized pieces of science and technology in ways he'd never imagined. He told her to go pursue her dream. Robertson was not a credulous man. He was completely captured.

That capture is the heart of everything that follows. What Holmes was promising — blood tests from a finger prick, results within the hour, available at a drugstore, at half the cost of a traditional lab — addressed something real. She cited a needle phobia as her own founding motivation. A Safeway executive whose husband was fighting lung cancer, his veins collapsing under repeated draws, called the fingerprick system a godsend. The gap between what medicine promised and what it routinely delivered was wide, and Holmes was responding to it with the conviction of someone who'd felt that gap personally.

The fraud that followed was built on top of something genuine. That's the reason so many brilliant people never quite found solid ground to push back from. Genuine hope asks you to wait and see — and waiting feels like patience, not credulity.

The Device Named After America's Greatest Inventor Was a $3,000 Glue Robot

The machine that eventually reached Walgreens customers started as a three-thousand-dollar industrial glue dispenser ordered from a New Jersey catalog. Tony Nugent, the engineer brought in after years of failed attempts to miniaturize the blood-testing hardware, simply didn't have time to design something from scratch. He fitted a pipette to its mechanical arm, programmed it to replicate a bench chemist's movements — aspirate, mix, measure — and had a working prototype in four months. Holmes named it the Edison, invoking America's greatest inventor. Her engineers called it the gluebot.

What distinguishes the Theranos fraud is that Holmes knew the device wasn't working and launched it on the general public anyway. In September 2013, weeks before Walgreens stores would begin offering blood tests, Anjali Laghari returned from three weeks in India to find the company on the verge of going live. She led the team responsible for the blood-test chemistry and had spent years trying to bring the Edison's error rates to something defensible. The data said they weren't there. She had the proof: a 2010 collaboration with pharmaceutical company Celgene in which the Edison had been used to track inflammatory markers in asthma patients. The error rates were too high; Celgene had ended the relationship. Nothing material had changed since.

Laghari emailed Holmes and deputy Daniel Young the Celgene data. Neither responded. She asked Holmes directly: why not wait for the next-generation device that was supposedly almost ready? Holmes answered: "Because when I promise something to a customer, I deliver." Laghari understood what this meant for the actual customers — not Walgreens, but the patients who would show up at store wellness centers and make medical decisions based on whatever result came back. She resigned. Knowing that Holmes's partner and operating chief, Sunny Balwani, monitored employee email, she printed her dissenting message and smuggled the paper out in her bag.

When the Edisons eventually came under government inspection, the numbers she had feared were in the lab's own records: quality-control failures nearly a third of the time in one month; a hormone test failing 87 percent of its quality checks; results diverging from conventional machines by as much as 146 percent. Holmes had known the shape of these numbers before the launch. She had been told by the person whose job was knowing them.

The Most Prestigious Board in Silicon Valley Was the Fraud's Most Powerful Weapon

If a board of directors exists to catch a founder who's deceiving investors, why did Theranos's board spend eight years doing the opposite?

Avie Tevanian was Steve Jobs's oldest collaborator — the man who ran Apple's software engineering through the iPod years. He joined the Theranos board in 2006, bought $1.5 million in stock, and spent the following year watching the same pattern: Holmes presented pharmaceutical deals that never materialized, offered a different technical excuse each quarter for why the product wasn't ready, and refused to produce contracts when pressed. By late 2007, Tevanian had compiled hundreds of pages of board documents showing that every executive, every projection, every stated milestone had changed in a year. He brought the stack to chairman Don Lucas and laid out the discrepancies. Lucas's response was a single question: "What are you planning to do with that stack of papers?" He didn't open the binder. Within weeks, after Tevanian blocked a governance maneuver that would have increased Holmes's voting control, Theranos's general counsel emailed him at 11:17 p.m. on Christmas Eve accusing him of "bad faith" and threatening to sue him for breach of fiduciary duty. A lawyer friend offered the clarifying question: given everything you now know about this company, do you really want to own more of it? Tevanian signed the waiver. His parting letter to Lucas warned that board members who didn't go along "100% with the program" risked "retribution from the Company/Elizabeth."

That warning went unheeded. Holmes understood what Tevanian's exit meant: the board was now hers.

When Partner Fund Management (a San Francisco hedge fund managing $4 billion) met with Holmes in late 2013, what stopped the fund's health-care analyst from pressing harder was the letterhead: Henry Kissinger, George Shultz, James Mattis, William Perry, Sam Nunn. Former secretaries of state, defense secretaries, a four-star general. The fund later described these names as giving Theranos "a stamp of legitimacy." They committed $96 million at a $9 billion valuation without a single independent test. With David Boies, the litigator who represented Al Gore in Bush v. Gore, listed as special adviser, the firm took Holmes at her word.

The board made this mechanism work: every credentialed name Holmes added made the next investor easier to capture, and every failed challenge from within — Tevanian's papers, the attempted board takeover in 2008 that Holmes defused by persuading the dissenters one by one, the two internal dissenters fired the week after they forced a board meeting — made her harder to challenge next time. It wasn't one sociopath. It was Don Lucas choosing not to open a binder. It was institutions designed to catch exactly this, declining, one by one, to do their job.

Every Person Who Raised Their Hand Was Systematically Destroyed

Tyler Shultz was twenty-three years old when he arrived at his grandfather George's house in Palo Alto on a May evening in 2015, thinking he was going to have a conversation. George was a former U.S. Secretary of State, and Elizabeth Holmes had called him to say that Tyler, who had quit Theranos the previous year after documenting data manipulation, fake validation results, and illegal proficiency testing, was about to become a problem. George told his grandson there were two people from the company upstairs who just wanted to talk.

The two people were partners at Boies, Schiller & Flexner. They had a temporary restraining order, a court summons, and a letter informing Tyler that Theranos was prepared to sue him. The senior partner, who had apparently identified Tyler as a likely journalist source in about five minutes, immediately pressed him to name every other employee who had spoken to a Wall Street Journal reporter. Tyler refused. The lawyer kept pressing. George's wife Charlotte sat nearby glowering, looking (in Carreyrou's telling) like she was ready to hit someone.

Before the lawyers came downstairs, Charlotte had pulled Tyler aside to say she was starting to wonder whether the Theranos device was even real. Henry Kissinger, she added quietly, was saying the same thing and wanted out. Then the interrogation began and the conversation was over.

His grandfather, trying to broker a compromise, offered to call the family's estate attorney to review whatever document Theranos wanted Tyler to sign. Tyler suspected George would reach the attorney first and frame the situation to make signing seem reasonable. So Tyler excused himself. Charlotte handed him the kitchen phone book with one word: "Call him." He found the number and slipped into the backyard. When he reached the attorney and mentioned the opposing firm's name, the response came back: "Holy shit, do you know who that is?" That gap — Tyler mispronouncing "Boies," the attorney nearly coming through the phone — is a precise measure of the asymmetry involved. On one side: a twenty-three-year-old with a biology degree who was correct about everything. On the other: the most feared litigation firm in America, deployed by a company valued at nine billion dollars.

Tyler didn't sign. But the campaign against him ran for months: private investigators, threats to bankrupt his parents, his grandfather used repeatedly as a conduit for messages from Holmes. Boies Schiller also reached Alan Beam, the former laboratory director who had forwarded 175 work emails to his personal account specifically to document the concerns he'd raised with management. His own attorney, intimidated by the firm's pressure, advised him to comply. One evening, Beam sat at his computer and deleted all 175 of them, one by one.

Every person who raised their hand inside Theranos was shown exactly what the cost would be. Most of them did the math. That is the precise mechanism by which large frauds survive — not because no one knows, but because knowing and proving are separated by a legal bill that could take your parents' house.

The Truth Only Survived Because One Reporter Had Something Boies Schiller Couldn't Pierce

By mid-2015, Wall Street Journal reporter John Carreyrou had what he needed: Alan Beam's specifics about diluted samples and proficiency-testing violations, comparisons showing Theranos flagging his own blood values as abnormal while LabCorp returned them all normal, a source whose cortisol Theranos reported at levels associated with Addison's disease (a condition indicating adrenal failure) while a same-day LabCorp draw showed it squarely normal. The truth was in hand. Whether it could be published was another question.

Theranos deployed David Boies to answer that question in its favor. His firm had Erika Cheung, another former Theranos employee who had contacted Carreyrou, surveilled to a secret address her own mother didn't know, and delivered a cease-and-desist letter by a man who stepped from a parked SUV. It obtained signed statements from two doctors claiming Carreyrou had mischaracterized their words. It sent the Journal a 23-page letter threatening suit and cataloguing alleged failures of journalistic fairness. Holmes even lobbied Rupert Murdoch directly: four meetings, the last in his Midtown Manhattan office three floors above Carreyrou's newsroom desk, asking him to kill the story. Murdoch had $125 million invested in Theranos and owned the Journal's parent company. He declined each time.

That Murdoch didn't intervene is the most important editorial fact in the book. What Boies's machinery couldn't touch was the institutional structure behind Carreyrou: staff lawyers, a standards editor, the financial resources to litigate for years. When the 23-page letter arrived, the Journal's legal team responded by copying the contents of Carreyrou's laptop in preparation for the fight. The paper wasn't backing down; it was getting ready.

The story ran on October 15, 2015. Days later, Holmes appeared at the Journal's own technology conference and told a livestreamed audience that running diluted finger-stick samples on commercial analyzers was "just not possible." Alan Beam texted Carreyrou in real time: "I can't believe what she just said." She had won every previous confrontation by outspending the other side. This time the other side had resources she couldn't match and an audience she couldn't control.

True Believer or Knowing Fraudster: You're Asking the Wrong Question

Was Elizabeth Holmes a genuine visionary who gradually crossed an ethical line, or was she always a knowing fraudster? The book spends 300 pages gathering evidence for both answers and then declines to choose, and that refusal is the most honest thing in it.

Carreyrou's verdict: she probably didn't set out to defraud anyone when she dropped out of Stanford at nineteen. But by November 2006 — seven years before billion-dollar startup valuations became a status symbol — Holmes fired her CFO the instant he told her, "We've been fooling investors." Not after reflection. Not after consultation. The moment the words left his mouth, her warmth vanished and she said: "Henry, you're not a team player. I think you should leave right now." The knowing deception predates everything the book reaches for to explain it.

So which is she? The honest answer is: that question won't protect the next patient.

What produced Theranos wasn't one woman's psychology. It was a startup culture that celebrates shipping before you're ready, applied to a medical device rather than software. It was a board structure that substitutes famous names for actual scrutiny. It was a legal apparatus that can outspend any individual who raises their hand. Remove Holmes entirely and those conditions remain, ready to organize around the next person who says what investors want to hear.

The book answers the question it set out to answer: how did this specific fraud happen? But the question it leaves open is harder. If the fake-it culture, the prestige board, and the Boies Schiller letter are still in place, the name at the top of the org chart barely matters. Theranos was a singular catastrophe. The conditions that made it possible are not.

The Machine Is Still Running

The conviction landed in January 2022: eleven counts of fraud, up to twenty years' exposure. One fraud resolved. The machinery intact.

The FDA finalized a rule requiring lab-developed tests to go through standard premarket review in April 2024. The draft had been sitting since 2014 — the year Theranos was already running Edison machines in Walgreens pharmacies. The rule is being challenged in federal court.