Theranos — collapse of a fraudulent blood-testing company

1. Episode summary

Theranos was founded in 2003 by Elizabeth Holmes, a Stanford dropout, on the premise that a proprietary device — later branded the "Edison" — could run a comprehensive menu of laboratory blood tests from a few drops of capillary blood drawn by finger-stick. Over the next decade the company raised more than $700 million at a peak valuation of roughly $9 billion, assembled a board drawn heavily from former senior US officials (including two former Secretaries of State and two former Secretaries of Defense), and in 2013 launched a retail partnership with Walgreens that opened "wellness centers" offering Theranos blood tests to the public in Arizona and California. Beginning with a Wall Street Journal investigation published in October 2015, and followed by a 2016 Centers for Medicare & Medicaid Services (CMS) inspection report on the company's Newark, California laboratory, evidence emerged that the Edison device had never performed the advertised menu at clinical quality, that most Walgreens-channelled samples were in fact being run on modified third-party analyzers, and that quality-control and proficiency-testing norms had been violated. CMS revoked the laboratory's CLIA certificate and barred Holmes from laboratory ownership for two years; Theranos voided two years of Edison results; Walgreens terminated the partnership and sued for recovery. In March 2018 the SEC charged Holmes, the company, and former president Ramesh "Sunny" Balwani with "massive fraud"; in June 2018 a federal grand jury indicted Holmes and Balwani on wire-fraud charges; the company was dissolved in September 2018. The strategic question the episode turned on was whether the company's public and investor representations about the capability and deployment of its core technology corresponded to what the technology could actually do.

2. Sources

Primary:

1. Securities and Exchange Commission, Complaint: SEC v. Elizabeth Holmes and Theranos, Inc. (Case No. 5:18-cv-01602), U.S. District Court for the Northern District of California, filed 14 March 2018. Paragraphs on Edison-device capability representations, Department of Defense deployment claims, and 2014 revenue representations.
2. Securities and Exchange Commission, Press Release 2018-41, "Theranos, CEO Holmes, and Former President Balwani Charged With Massive Fraud," 14 March 2018.
3. U.S. Department of Justice, Northern District of California, United States v. Elizabeth A. Holmes and Ramesh "Sunny" Balwani, Superseding Indictment (Case No. 5:18-cr-00258-EJD), June 2018; trial record and 3 January 2022 verdict (four counts of wire fraud and conspiracy against investors); 18 November 2022 sentencing memorandum and 135-month custodial sentence.
4. Centers for Medicare & Medicaid Services, Survey and inspection report on Theranos Inc.'s Newark, CA clinical laboratory (CLIA), findings dated November 2015 and released publicly in redacted form in early 2016; and CMS sanction letter of July 2016 revoking the laboratory's CLIA certificate and imposing a two-year ownership ban on Holmes and Balwani.

Secondary (with justification):

1. John Carreyrou, Bad Blood: Secrets and Lies in a Silicon Valley Startup, Alfred A. Knopf, 2018. Investigative synthesis drawing on interviews with roughly 150 people, including approximately 60 former employees; expansion of the author's 2015–2016 Wall Street Journal reporting. Secondary because it synthesises primary testimony and documents into a narrative.
2. John Carreyrou, "Hot Startup Theranos Has Struggled With Its Blood-Test Technology," The Wall Street Journal, 16 October 2015, and follow-up articles through 2016. Contemporaneous investigative journalism; secondary under the methodology's taxonomy because it synthesises whistleblower and internal-document evidence rather than being the underlying record itself.
3. Lynda M. Applegate, William A. Sahlman and colleagues, "Theranos: The Unicorn That Wasn't," Harvard Business School Case Collection, 2019 (and updates). Peer-reviewed teaching case synthesising the public record for management-education use.
4. Erin Brodwin, "Jim Mattis testifies that Elizabeth Holmes misled him about Theranos' technology," Axios, 23 September 2021. Contemporaneous trial-day reporting on General Mattis's testimony; provides named-actor, dated evidence on board-level deception (board membership from 2013, personal investment of $85,000, stated loss of confidence in Holmes). Secondary because it reports on primary court proceedings already in §2 Primary source 3.

Tertiary (flagged):

1. Wikipedia, "Theranos" and "Elizabeth Holmes" articles (accessed 2026-04-23). Used for timeline frame only; not load-bearing on any factual claim in §1 or §3.

3. OTA narrative

Observe. The observation apparatus facing the company — internal laboratory QC data, proficiency-testing outputs, and comparisons between Edison results and reference-method results on split samples — was in place and, by the account of the 2016 CMS inspection report and the testimony of former laboratory directors recorded in the SEC complaint and the 2022 trial record, was generating signals that Edison performance did not meet clinical standards. Quality-control failures, out-of-range QC checks (reportedly on the order of 29 per cent on some Edison assays), warfarin-dose test anomalies flagged by CMS as "immediate jeopardy to patient health and safety," and dissenting reports from scientific staff and the company's own CLIA laboratory directors were produced inside the firm. The signal existed and was legible; internal whistleblowers (including former laboratory director Adam Rosendorff and junior scientist Tyler Shultz) were able to extract it and carry it to regulators and to journalists. Observe was not a root cause: the observation function produced the signal that a reasonably-resourced peer in the clinical-diagnostics archetype would have produced. Observe was a transmission step — it carried the signal through to the decision surface, where it was overridden rather than acted on.

Think. The interpretive step that converts internal QC evidence into a decision to halt the commercial rollout, disclose material limitations to investors and partners, and revise public claims was not made. Leadership continued to represent to investors, to Walgreens, and through media appearances that the Edison device performed the advertised menu at clinical quality, that the technology had been deployed with the U.S. Department of Defense on battlefield and medevac platforms, and that the company would generate revenue on the order of nine figures in 2014 — representations which, per the SEC's 2018 complaint and the DOJ's superseding indictment, did not correspond to the internal record and to actual 2014 operating revenue of approximately $100,000. The interpretive framework needed — a routine application of diagnostic-validation norms to internal QC data — was standard for the archetype and was being applied by the laboratory directors who resigned or raised dissent. The reasoning failure was therefore an Easy-Wrong Think: the correct framework existed, was accessible, and was being applied by the firm's own clinical staff; it was overridden at the executive level. Think is a root-cause phase in this episode.

Act. Execution carried the wrong interpretation into the outside world: the Walgreens rollout proceeded despite internal dissent, investor materials repeated the Edison-capability and DoD-deployment claims, and public-facing communications reinforced them. Where the evidence supports concentrated causal attribution, however, is on the act of representing the technology as something it was not — the filings in the SEC complaint and the wire-fraud counts on which Holmes was convicted in January 2022 characterise these as deliberate misrepresentations rather than as executional misfires. Act is a root-cause phase in this episode at the easy end of the task-difficulty axis: truthful disclosure of material product limitations and of actual revenue to investors is a routine obligation for the archetype, and the routine obligation was not discharged. Act is classified Wrong at the easy end of the difficulty axis — the routine disclosures were omitted or were replaced with representations the internal record contradicted.

4. Modality evidence

Direction. The episode's directional layer is constituted by a sequence of specific, attributable strategic choices that set the company's public representation of itself. Holmes founded Theranos in 2003 on the explicit premise that a proprietary device could run a comprehensive blood-test menu from a finger-stick sample — a founding direction claim that the SEC complaint and DOJ superseding indictment establish as the basis for all subsequent investor and partner representations (SEC complaint 2018; DOJ superseding indictment 2018). The decision to launch a public retail partnership with Walgreens — the agreement signed in 2012, the public announcement in September 2013, and the opening of wellness centres in Walgreens locations in Arizona and California — was a specific, dated, attributable act by Holmes and Balwani that committed the company to representing its technology as clinically operational to a consumer audience (Carreyrou, Bad Blood 2018; Carreyrou, Wall Street Journal 2015). The DoD deployment claim — that Edison devices had been deployed by the U.S. military in Afghanistan and on medevac helicopters — was a deliberate narrative choice made in investor presentations and media appearances, characterised in the SEC complaint as a material misrepresentation repeated to multiple investor classes in 2013 and 2014, and corroborated at trial by investor testimony that the DoD deployment claim was a primary basis for investment (SEC complaint 2018; DOJ superseding indictment 2018; Brodwin, Axios 2021). General James Mattis, who joined the Theranos board in 2013 and invested $85,000 of his own funds, testified in September 2021 that he was misled about the state of the technology and described being "disappointed at the level of transparency" from Holmes — establishing that the directional misrepresentation operated against the board as well as against external investors (Brodwin, Axios 2021).

Structure. The governance architecture of Theranos is the central structural evidence in this case and is governed by the Fraud Case Structure-Culture Rule, which requires separate attribution of structural enablers and cultural drivers. The board of directors assembled between 2011 and 2013 was composed predominantly of former senior US government officials — including two former Secretaries of State (George Shultz, who joined in July 2011, and Henry Kissinger), a former Secretary of Defense (William Perry), a former Senate Majority Leader (Bill Frist), a retired four-star Marine general (James Mattis), and others — none of whom held scientific, medical, or clinical-laboratory expertise (Carreyrou, Bad Blood 2018; Applegate, Sahlman et al., HBS Case 2019). This composition meant the board held no internal mechanism to evaluate the technical claims that constituted the company's value proposition. The Theranos board did not maintain an audit committee or any scientific advisory function capable of independently testing Edison's clinical performance; a scientific advisory board was created only in mid-2016, after the Wall Street Journal reporting and CMS inspection had already surfaced the laboratory's failures (Carreyrou, Wall Street Journal 2015; Applegate, Sahlman et al., HBS Case 2019). Authority over operational and regulatory decisions was concentrated in Holmes and Balwani, with laboratory directors — the personnel with the competence to evaluate QC and proficiency-testing data — carrying formal CLIA regulatory responsibility but lacking organisational authority to compel corrective action against Holmes's and Balwani's commercial priorities, as laboratory director Adam Rosendorff testified at trial (DOJ superseding indictment 2018, trial record). This structure — a board incapable of technical oversight, authority concentrated in two executives with financial incentives to misrepresent, and a formal regulatory-responsibility layer without corresponding decision-making authority — was a structural arrangement that made the fraud practically executable over more than a decade.

Processes. The laboratory operations processes provide the most direct evidence of failure in the clinical-execution layer. The CMS inspection conducted in late 2015 documented multiple failure modes in Theranos's Newark, California CLIA-certified laboratory: out-of-range quality-control checks on Edison assays at rates that CMS characterised as immediate jeopardy to patient health and safety (warfarin-dose test anomalies are explicitly cited in the CMS sanction record), deficiencies in proficiency-testing procedures, and practices that would not have been permissible under standard CLIA compliance (CMS inspection findings and sanction letter 2015–2016). The Edison device's actual operational architecture further exposed a process gap between represented and actual capability: of approximately 200 tests offered, only around 15 ran directly on Edison hardware; approximately 60 more used a hybrid approach in which finger-stick nanotainer samples were diluted to volumes compatible with conventional instruments, a dilution that experts attested could introduce error; and approximately 130 were run on commercially available third-party analysers, including modified Siemens ADVIA instruments — none of which was disclosed to Walgreens customers or to investors as the actual testing modality (Carreyrou, Bad Blood 2018; SEC complaint 2018). The failure to route the Walgreens rollout through an independent clinical-validation procedure — a standard process for a diagnostic laboratory entering a retail distribution channel — is documented in the Wall Street Journal reporting: Walgreens executives raised questions and sought to investigate the technology in detail on multiple occasions before launch, and Theranos deflected each time, with Walgreens proceeding despite unresolved technical due diligence (Carreyrou, Wall Street Journal 2015).

Scoring note (zero-modality rationale): the Processes contribution described in this subsection is classified at the boundary with Culture in the scoring record — the §4 evidence locates the operative driver of the episode's failure causation in Culture rather than in a standalone Processes contribution. Processes is acknowledged in narrative as evidenced but does not carry independent weight in the scoring; weight is borne by Direction, Structure, Culture. Categorisation under METHODOLOGY-ota-scoring-v4.md §5: classification boundary with an adjacent modality.

Capability. The capability gap in this case is fundamental rather than marginal: the Edison device and its associated nanotainer technology never achieved the clinical performance that was the company's core value proposition, as established by the CMS inspection findings, the SEC complaint's description of actual operating conditions, and Carreyrou's reconstruction of the internal record (CMS sanction letter 2016; SEC complaint 2018; Carreyrou, Bad Blood 2018). The device that could run a comprehensive blood-test menu from a finger-stick at clinical quality did not exist in deployable form during the period of the Walgreens rollout (2013–2016). The reliance on modified third-party Siemens ADVIA analysers — which were not designed to run the miniature sample volumes Theranos collected and required engineering workarounds that experts characterised as likely to introduce error — is the operational expression of this capability gap: the company substituted an approximation for a capability it did not possess (Carreyrou, Bad Blood 2018; SEC complaint 2018). Notably, the capability to detect the gap existed internally: laboratory directors including Adam Rosendorff, who resigned in late 2014 citing Holmes's unwillingness to address QC problems, and junior scientist Tyler Shultz, who contacted regulators using an alias, were able to read the QC and proficiency-testing data and identify the performance shortfall. The capability gap was thus not epistemic — the internal evidence was legible to trained laboratory personnel — but was a gap between what the technology could do and what had been represented to the outside world.

Scoring note (zero-modality rationale): the Capability contribution described in this subsection is classified at the boundary with Processes per the methodology §3 Processes / Capability replacement test ("if the current operating staff were replaced by new hires of comparable background, would the operational pattern survive?"). The §4 evidence applies the test explicitly and concludes that the strategic weight sits on the Processes side — the operational edge survives staff turnover because it lives in documented routines and tool support. The Capability component is acknowledged in narrative but does not carry standalone weight; both modalities are evidenced and the boundary call is recorded in the audit trail. Categorisation under METHODOLOGY-ota-scoring-v4.md §5: classification boundary with an adjacent modality.

Culture. The cultural layer in this case is governed by the Fraud Case Structure-Culture Rule: it is assessed separately from the structural governance failures documented above. The operative cultural mechanisms are suppression of internal dissent and systematic enforcement of secrecy as a behavioural norm. Theranos bound employees through aggressive non-disclosure agreements that threatened multi-million-dollar penalties, and through a monitored, compartmentalised work environment in which employees were systematically isolated from one another and from senior leadership except through Holmes and Balwani (Carreyrou, Bad Blood 2018). When Tyler Shultz raised quality-control concerns internally, Holmes forwarded his email to his direct supervisors; Balwani then berated Shultz in a follow-up meeting, calling him arrogant and ignorant and stating that he would have been dismissed already if not for his surname — a surname he shared with board member George Shultz (Carreyrou, Bad Blood 2018). After Shultz left the company and cooperated with regulators, Theranos's legal counsel sought to compel him to sign documents recanting his disclosures and threatened litigation that left him facing nearly half a million dollars in legal fees (Carreyrou, Bad Blood 2018; Carreyrou, Wall Street Journal 2015). Adam Rosendorff, the CLIA laboratory director, testified that Holmes held decision authority over regulatory compliance choices that Rosendorff characterised as driven by "business priorities" rather than by regulatory requirements, and that his attempts to raise QC concerns were not converted into corrective action — ultimately leading to his resignation in late 2014 (DOJ superseding indictment 2018, trial record). The pattern across the Carreyrou reporting and the DOJ/SEC record is consistent: a cultural environment in which dissent was identified, confronted, legally threatened, and suppressed as a standard operating norm, and in which the distance between the internal record and the external representation was maintained by deliberate behavioural enforcement rather than by structural channel failure alone.

---