I first heard about Theranos in late 2014 or early 2015.
I was a college freshman then and had just decided to major in bioengineering. On a bulletin board in a ground-floor hallway of the engineering building, I came across a profile of the company. At the time, Theranos was a rising star in Silicon Valley, with a valuation that would reach $9 billion. It claimed it could run a wide range of tests using just a small amount of blood from a fingertip—tests that would otherwise require blood drawn from a vein. This engineering breakthrough promised to make blood testing cheaper and more convenient than ever.
What made that promise exciting was more than the prospect of having fewer tubes of blood drawn. If testing became inexpensive and easy, people might learn about their health sooner and detect diseases earlier. Theranos’s partnership with Walgreens, the ubiquitous American pharmacy chain, made that future seem within reach. Perhaps you could simply walk into your neighborhood pharmacy for tests that once required a trip to a hospital or lab.
For a student who had just chosen his major, this was almost exactly what I imagined bioengineering should do. In a technology hub where internet companies commanded the spotlight, the rise of a medical technology company made the future feel full of possibility. I thought about how wonderful it would be to take part in something like this after graduation.
In the fall of 2015, reporting by Wall Street Journal journalist John Carreyrou began to reveal another side of Theranos. The company’s own devices fell far short of its claims. Many of its tests were actually performed on commercial equipment made by other manufacturers, and the reliability of its proprietary devices was also called into question. Over the next few years, the technological mythology surrounding the company gradually collapsed.
When Carreyrou’s book Bad Blood came out in 2018, I bought it right away.
I felt a sense of loss. Not because Holmes had lost her chance to become a great entrepreneur, but because the collapse of Theranos, once a standard-bearer for medical innovation, left me disillusioned. A breakthrough that looked so exciting could turn out to be a carefully sustained story. How many other innovations supposedly on the verge of changing the world would fail under scrutiny?
After the disappointment came anger. I had thought medical innovation could follow the same path as Silicon Valley’s other success stories: break through the old order and change the world in a few short years. With Theranos gone, all I could see ahead was years of research, painstaking validation, and slow, incremental progress. At the time, I found that discouraging. I did not yet really understand why some steps could not be skipped.
Rereading Bad Blood After Starting a Company
Years later, I found myself in Silicon Valley, though I had entered AI rather than biotech.
It is an industry moving at extraordinary speed, in an era when it is easy to believe that enormous changes are just around the corner. Working in it, my wife and I began a startup of our own. Perhaps some of that student dream was still alive: although we were building an AI application, we chose to work on regulatory compliance for medical devices.
Only once we were doing the work did we begin to understand how much specialized knowledge it required, how difficult reliable data was to obtain, and how long it took customers to make decisions and put a product into use. A seemingly simple feature could involve complex professional judgments and responsibilities we could not simply disclaim. We eventually decided to move on, but the experience changed how I thought about entrepreneurship. It also helped me begin to understand what those procedures I had once found so slow were there to keep in check.
When I reread Bad Blood after starting a company myself, I found that I had different questions about Holmes.
I began to wonder how a young person caught up in Silicon Valley’s startup boom would make sense of the success stories unfolding around her. When everyone is talking about vision, speed, and changing the world, how much resolve does it take to admit that something cannot yet be done?
In the B2B SaaS world we encountered, “sell first, build later” is common startup advice. There is a practical reason for it. Engineers can easily spend months building something they think is cool, only to discover that no one wants to buy it. Better to take a demo or prototype to potential customers, establish that there is demand and a willingness to pay, and then invest in development.
Having gone through the experience myself, I understand the appeal. When resources are limited, you want every step to minimize wasted effort. And in the face of uncertainty, you need some faith that you can turn your promises into reality.
But there is a boundary here that cannot be overlooked. Letting customers know that a product is unfinished and inviting them to explore it with you is different from leading them to believe a capability already exists. Believing you will be able to do something is no substitute for demonstrating that you can do it now.
Theranos’s problems cannot be explained simply as “promising too soon.” Evidence presented at trial showed that Holmes had given investors reports bearing pharmaceutical companies’ logos, creating the impression that those companies had validated and endorsed Theranos’s technology. This was about more than excessive optimism about the future. It was about how she described the present and represented what had already happened.
The controversy surrounding Delve, a YC startup, in 2026 brought similar questions to mind. Delve markets AI tools to accelerate corporate compliance work. An anonymous investigator accused it of supplying fabricated records of meetings, tests, and procedures, and generating audit conclusions before an independent review. The company denied the allegations, saying the materials were only templates and that final reports were issued by independent auditors.
The allegations are disputed, but the question they raise deserves attention: technology can make it faster to organize evidence, but it cannot stand in for the facts that evidence is supposed to document. Whether in medical testing or corporate compliance, when reliability and trust are precisely what customers are buying, the gap between a promise and a fact cannot be treated as a minor issue to sort out later.
So my understanding of Holmes now comes with more questions, and more caution. Silicon Valley’s culture may help explain why someone could become so certain about the future. But when reality repeatedly contradicted her promises, how she handled that evidence—and what she told other people—remains for her to answer.
Does Seeing More Help Us Understand More?
You Can See Everything brought this question back to me.
Directed by Nathan Fielder and Lance Oppenheim, the documentary began filming 34 days before Holmes went to prison. She invited the crew into her home to document her life with her partner and children. Before watching it, I hoped this close view of her life would help me understand her better. Afterward, I found that greater access to a person does not necessarily make her easier to understand.
One conversation stayed with me.
Nathan repeatedly asks Holmes whether Theranos’s machine could actually perform more than a hundred tests. She insists that it could. Nathan brings up evidence that it could perform only twelve. Why, he asks, does she keep saying otherwise?
As the conversation continues, they seem finally to reach the source of their disagreement. Nathan is asking whether the machine could already perform those tests at the time. Holmes is saying that it had the capacity to do so, provided the corresponding chemistry was developed. In her explanation, the fact that this research remained unfinished did not stop her from answering “yes.”
Nathan seems caught between disbelief and exasperation. Why can’t she understand that, on hearing “this machine can perform more than a hundred tests,” most people would take it to mean an existing capability, rather than a possibility with critical problems still to be solved?
What troubles me more is that even after this misunderstanding has been pointed out repeatedly, she continues to insist that she has explained herself clearly.
Does she truly not understand that the public would hear “theoretically possible” as “already possible”? Or is it enough for her that her explanation holds together logically, regardless of how other people understand it? Or does she see the distance between those two interpretations perfectly well—and use it to make people believe in a promise that has not yet been fulfilled?
I cannot tell which it is. Perhaps these states of mind are not always clearly separated. Someone might believe that the technology will eventually succeed while also accepting that others are mistaking a future possibility for a present fact, because that misunderstanding happens to bring her the support she needs.
Part of the film’s complexity is that it also shows us how this struggle over interpretation affects the people who come into contact with her.
Early on, Nathan asks Holmes whether she chose him to direct because he does not understand science and would therefore be easier to influence. As they spend more time together, he seems to begin to care about her. In a conversation with Amanda Seyfried, when she presses him on whether he has become protective of Holmes, he does not give a straightforward answer.
Watching this, it is easy for me to think that Nathan, too, has fallen under her influence. But then I remember that he is also the person who chose to leave that hesitation in the film.
The film thus asks us to consider not only how Holmes explains herself, but also how the director presents his relationship with her. Nathan’s uncertainty may be a trace of what that relationship has done to him. It may also be an experience the film deliberately creates for its audience. And when Amanda takes part in reenacting interactions that were not directly captured on camera, another layer of memory and performance enters what we see.
We see more and more, without becoming correspondingly more certain.
Of course, that uncertainty has limits. These questions do not erase the false statements or technical failures that have already occurred. A film can make a person more complex without making her actions impossible to judge.
In the Name of a Mission, Who Bears the Responsibility?
What truly unsettles me is that Holmes seems to remain driven by an extraordinarily strong sense of mission, yet I cannot tell how she thinks about what her pursuit of that mission has done to other people.
Her second child was only a few months old when she was due to enter prison. In her remaining days of freedom, the film still shows her discussing new medical technology projects with her partner and joining fundraising calls. When Nathan asks why she is so determined to build a second Theranos, she says she feels she was born to do this. Faced with someone like that, I find it difficult to say that everything she did was simply for money. At least in the life the film presents, she still seems to believe deeply that she has unfinished work to do.
Another contrast stays with me. When they discuss how many tests the machine could perform, Holmes keeps insisting that she has explained herself clearly, even as Nathan repeatedly points out that her statements would mislead people. She seems unconcerned about how others might misunderstand the product. Yet when it comes to how Amanda portrayed her in The Dropout, and where that portrayal departs from reality, she seems deeply concerned about how the public will see her.
Does she care about being misunderstood? Perhaps the more revealing question is: which misunderstandings matter to her?
If it matters so much to her that other people understand her fairly, shouldn’t the people who made decisions based on her promises deserve the same care? Believing she can transform healthcare does not relieve her of the responsibility to explain her technology’s limitations. The more important the mission, the more necessary it becomes to ask: until it is achieved, who bears the risk?
I cannot know how much of her persistence comes from sincerity and how much from self-protection. It has also made me question whether the pursuit of money and a sense of mission are necessarily opposing motives. Someone can sincerely want to change the world while hurting the people who believe in her. She can also become so convinced of the importance of her goal that she keeps finding reasons to justify what she is doing along the way.
I think back to myself standing in that engineering building hallway, looking at the profile on the bulletin board. What I aspired to then was precisely that story: someone who believed she could change the world and gave everything to it. Having started a company myself, I still think that kind of conviction is precious. Without it, many things might never begin.
What I want to understand now is what a person chooses to do when the evidence conflicts with that conviction.
Can she acknowledge that a capability does not yet exist, that a promise has not been kept, that a criticism is right? Can she accept that the problems other people point out do not all arise from their failure to understand her?
A sense of mission can carry a person through difficulty. But if that conviction becomes so strong that she no longer allows reality to test it, or seriously reckons with the consequences of her words and actions for others, how much damage can it do?
I cannot say whether that is more dangerous than defrauding people purely for money. But the thought that someone could be hurting others while remaining certain that he/she is building a better future for them leaves me with a deeper unease.
