In one sentence: John Carreyrou’s Bad Blood is the definitive account of how Elizabeth Holmes built Theranos into a nine billion dollar company on a blood-testing device that never worked, how a famous board and powerful partners failed to notice, and how a reporter and a few young whistleblowers brought it down.
At a Glance
Author: John Carreyrou
First published: 2018 (Alfred A. Knopf)
Category: Business / True Crime
Length: about 339 pages, about 100,000 words (Knopf hardcover)
ISBN-13: 978-0-525-43199-2 (Alfred A. Knopf hardcover)
Summary reading time: about 12 minutes
Book reading time: about 6.5 hours
Notable adaptations: the 2019 HBO documentary The Inventor, and the 2022 Hulu series The Dropout, which drew on the same story
Content note: the book came out in 2018, before Holmes and Ramesh “Sunny” Balwani were tried, so it ends with criminal charges described only as a possibility. The outcomes are now known and this summary states them. The book also describes the suicide of the Theranos chemist Ian Gibbons in some detail. Almost all of its dialogue is reconstructed from participants’ memories, some sources appear under pseudonyms the text does not identify, and Holmes declined to be interviewed, so her side and Balwani’s are largely absent.
John Carreyrou was a two-time Pulitzer winner at the Wall Street Journal when he broke the Theranos story in October 2015. Bad Blood reads like a corporate thriller but is built like an investigation: hundreds of interviews with more than 150 people, including more than sixty former employees, plus depositions, regulatory records and internal emails. Quotations from documents are verbatim. Dialogue is reconstructed from memory. The first two thirds rotate through insiders, and from chapter nineteen Carreyrou writes in the first person.
Read it if you want to understand how a fraud this large stayed hidden for a decade, how Silicon Valley’s tolerance for hype collided with medicine, or what it costs to blow the whistle.
The Big Idea
Elizabeth Holmes dropped out of Stanford in 2003 at nineteen and founded Theranos on a single promise: hundreds of blood tests from one finger prick. Over fifteen years she raised roughly 900 million dollars, reached a valuation of nine billion, and signed deals with Walgreens and Safeway. The technology never worked. According to the former employees Carreyrou interviewed, the company ran patient blood through commercial analyzers modified beyond their approved limits, hid that from partners and regulators, and silenced anyone who objected.
Carreyrou’s larger argument is about why nobody stopped it. Everyone who saw the gap between claim and laboratory was sidelined, fired or sued, and nobody with power asked for the one thing that would have settled it: an independent comparison against a real lab. That repeating pattern is the mechanism of the fraud.
Key Ideas
1. The device was miniaturization, not invention
Theranos 1.0 was a microfluidic cartridge and reader that never gave the same answer twice, partly because Holmes insisted on about ten microliters of blood, forcing heavy dilution. The Edison, built in late 2007 by Tony Nugent out of a roughly 3,000 dollar glue-dispensing robot and nicknamed the gluebot, automated one kind of test and nothing else. The miniLab, begun in 2010 and meant to do everything, was never finished. Carreyrou’s point is that none of the instruments inside were new, and competitors like the Piccolo Xpress already did some of what Theranos promised.
2. The demos were staged from the start
The book opens in November 2006 with a Novartis demonstration in Switzerland where, as a former executive recalls it, a reader failed and a result was beamed over by a colleague back home. Carreyrou reports that investor demos routinely ended by displaying a stored result. Later a journalist’s demo ran through an app that masked malfunctions, and Joe Biden toured what former employees describe as a staged “automated lab.”
3. Walgreens was told what it wanted to hear
Walgreens was told Theranos had a commercially ready lab and 192 available tests, when the room downstairs was a research lab and roughly half those tests could not run on the Edison. The Johns Hopkins validation Walgreens relied on was a two-page letter summarizing one meeting, which said on its face that Hopkins had verified nothing. Walgreens executives gave blood at the kickoff party and never got results, and its consultant Kevin Hunter, who asked for a fifty-patient comparison against Stanford Hospital, was barred from the calls. Carreyrou says Walgreens stayed because it feared losing out to CVS.
4. The real lab ran on other companies’ machines
By 2013, according to former lab staff, Theranos was running patient samples on more than a dozen commercial analyzers, with Siemens machines adapted to take diluted finger-stick blood, pushing samples below their cleared range. Staff nicknamed the commercial room Jurassic Park and the Edison room Normandy, and inspectors never saw the latter. Tyler Shultz, grandson of board member George Shultz, found that the syphilis test detected only sixty-five and then eighty percent of known positives while the validation report claimed ninety-five, and that failed quality-control runs were discarded as outliers. Erika Cheung watched a patient result go out after two failed checks.
5. Holmes had the last word, and the votes
Carreyrou rejects the convenient story that Balwani, who joined in 2009 after a forty million dollar dot-com windfall and had been Holmes’s partner since she was eighteen, was the real villain. He describes a regime of badge logs, surveillance and abrupt firings under Balwani, but Holmes had misled pharmaceutical partners before he arrived and had the final say throughout. The board, which came to include Henry Kissinger, George Shultz, James Mattis, Sam Nunn and William Perry, never functioned as a check. In March 2008 four directors voted to remove Holmes as chief executive and she talked them out of it in two hours. In December 2013 she took shares carrying a hundred votes each, giving her 99.7 percent of the voting power. Shultz, deposed later, said the board never took any votes. Carreyrou leaves any diagnosis to psychologists, but says her moral compass was badly askew and doubts she set out to defraud anyone: she stopped listening to sound advice and began cutting corners.
6. Lawyers did the work the technology could not
David Boies took on Theranos in its patent case against Richard Fuisz, accepting 300,000 shares worth about four and a half million dollars in lieu of fees, and Carreyrou reports that private investigators were put on the Fuisz family. Boies later joined the board, and his firm pressured employees and sources. Tyler Shultz was ambushed at his grandfather’s house with a restraining order and a threat letter, and his parents spent more than 400,000 dollars on lawyers. Ian Gibbons, the British biochemist credited with some fifty patents, was demoted after questioning Holmes’s honesty, was subpoenaed for the Fuisz case, and died by suicide in May 2013. Holmes did not return his widow’s call, and a company lawyer asked for his laptop. Theranos also told regulators its devices were in effect dumb fax machines and that one lab certificate covered everything, and when Army and FDA officials disagreed, Holmes complained to General Mattis about the Army officer.
7. Patients paid, and a reporter with a tip ended it
Carreyrou’s investigation began in February 2015 with a tip routed through a pathology blogger, and Alan Beam, the former lab director, spoke first on condition of anonymity. Through Yelp, Carreyrou found Dr. Nicole Sundene in Phoenix and her patient Maureen Glunz, who after a wrong result ran up about 3,000 dollars in emergency scans. His own comparison against LabCorp came back badly wrong. The Journal published on October 15, 2015, and the next day the FDA forced the nanotainer off the market. Erika Cheung’s email to a federal inspector had triggered a CMS inspection, which found the Edisons ran only twelve of about 250 tests, that one hormone test failed quality control eighty-seven percent of the time, and that results differed from conventional machines by up to 146 percent. Theranos voided or corrected close to a million results, Walgreens sued, Partner Fund settled for 43 million dollars, Rupert Murdoch sold his stake back for one dollar, and in March 2018 the SEC called it an elaborate years-long fraud.
Context and Analysis
Where the paper trail exists, Bad Blood is unusually well sourced: court filings and depositions, the CMS inspection reports, dated internal emails, the Fuisz patent record, the Johns Hopkins letter, regulatory correspondence and the settlement figures. Its structural findings have survived two criminal trials it could not include. Holmes and Balwani were indicted in June 2018. In January 2022 Holmes was convicted on four counts of defrauding investors and acquitted on the counts relating to patients, and in November 2022 she was sentenced to eleven years and three months. Balwani was convicted on all twelve counts in July 2022 and sentenced to just under thirteen years. Theranos was dissolved in 2018 and George Shultz died in 2021. The book predates all of it, and its central claim no longer rests on Carreyrou’s word.
The fair criticisms matter. The author’s note concedes that all dialogue comes from participants’ memories. That covers the prologue confrontation, Holmes’s childhood vow to be a billionaire, every boardroom exchange, and private scenes with interior thought and physical detail that cite no source, including Ian Gibbons’s last night. Several chapters rest on one person’s account, often someone who was fired. Holmes refused to be interviewed and Balwani never speaks beyond emails, so their positions appear only in lawyers’ letters, a press release and two television appearances, and both later testified at trial. Some people appear under pseudonyms the note does not identify, so a reader cannot weigh any particular account. Carreyrou is a participant from the midpoint, and while he admits that one of his own emails exposed Tyler Shultz as a source, the narration flatters him and he reaches for words like tyrant and queen in his own voice. There is also loose material: a rumor about bribery in Thailand, a joke implying drug use, and a passage that characterizes Indian consultants and executives in ethnic terms that are the author’s framing, not a source’s.
On this site it pairs naturally with Steve Jobs by Walter Isaacson, the life Holmes consciously imitated, down to the black turtlenecks and the reality-distortion field, and with Elon Musk by Ashlee Vance, another founder who promised far beyond what existed, which makes a sharp test of where the line between vision and fraud falls.
What It Teaches
Five lessons from the story.
1. Secrecy is not the same as a trade secret. Theranos used confidentiality to hide a failure, not an invention, and the tell was its refusal to allow a simple independent test. 2. Governance is a verb. A board that never votes, lets a founder hold 99.7 percent of the voting power, and treats a request for data as disloyalty is an audience, not a board. 3. Whistleblowers pay first and are thanked last. Tyler Shultz and Erika Cheung were in their twenties, had no power, and faced the best lawyers in the country, and the collapse began with one of them emailing a regulator. 4. Software can ship broken. Medicine cannot. Hyping a product while concealing its progress is tolerated in technology, but Theranos was first and foremost a health-care company. 5. A famous board is worth exactly what it checks. Kissinger, Shultz, Mattis, Nunn and Perry lent credibility earned elsewhere and verified nothing, and their names made it harder to ask whether the device worked.
Memorable Lines
“Sunny was a force of nature, and not in a good way.” (John Carreyrou)
“It was as if working at Theranos was gradually stripping them all of their humanity.” (John Carreyrou)
“It was one thing to do that with software or a smartphone app, but doing it with a medical product that people relied on to make important health decisions was unconscionable.” (John Carreyrou)
“Such overpromising became a defining feature of Silicon Valley.” (John Carreyrou)
“Holmes knew exactly what she was doing and she was firmly in control.” (John Carreyrou)
“Her ambition was voracious and it brooked no interference.” (John Carreyrou)
Should You Read the Full Book?
Verdict: Essential
This summary carries the shape of the fraud, how it was hidden, and the outcomes the book could not know. The full book adds the experience of watching it happen one person at a time, inside a company where badge logs and lawyers replaced data. That texture is why it reads like a thriller, and it is where the reconstruction from memory does its heaviest lifting. Read it knowing that Holmes and Balwani do not speak in it, that some names are false, that the author becomes a character halfway through, and that two juries have since confirmed its core. The Bad Blood book page has the full details and where to get a copy.