Oceans are under threat. Can cell-cultured fish help?
San Diego, California, was once known as the “Tuna Capital of the World.” Throughout much of the 20th century, thousands of workers caught cheap albacore tuna off the coast and packed it in canneries that lined the city’s waterfront. But by the 1980s, operating costs and foreign competition were too high, ships stopped their journeys out to sea, and the canneries closed down.
Now, some 40 years later, San Diego startup BlueNalu is looking to put the city back on the culinary map — but with a very different approach to producing seafood that doesn’t involve fishing.
Five miles inland, in a sprawling office park, BlueNalu is growing fish cells in large stainless steel tanks, known as bioreactors. Instead of producing the cheap, $5-per-pound albacore tuna San Diego was once known for, BlueNalu is brewing up Pacific bluefin tuna toro — the prized fatty belly portion of the near-threatened fish, which fetches over $100 a pound on the retail market.
The meat is made by taking a small amount of fish cells (sourced from a San Diego fishery), placing them in the bioreactor, and feeding them a mix of nutrients, such as amino acids, salts, and sugars, for several weeks until they can be harvested for consumption. It’s colloquially called “lab-grown meat,” though BlueNalu prefers the term cell-cultured, while most in the nascent industry call it “cultivated” meat. Whatever the name, it does taste good.
I should add that I’ve never eaten the belly fat of a bluefin tuna — the fish is on Seafood Watch’s red list because of its near-threatened status (its population has dwindled to just 3 percent of its historical, pre-fishing levels). But BlueNalu’s version, which was served up as sushi and nigiri, didn’t taste far off from how food critics have described the wild-caught version: a butter-soft flavor bomb. It left a thin lining of oil on my mouth in a way mere plant-based meat never has — something that maybe only real fish could achieve, even when made by replicating cell after cell.
/cdn.vox-cdn.com/uploads/chorus_asset/file/24288730/1_Traditional_saku_of_bluefin_tuna_toro_sliced.jpg)
BlueNalu
It was my second such tasting of fish made from the cell up this year; in May, I tried cell-cultured salmon served in a poke bowl from the San Francisco-based startup Wildtype. It was similarly soft, buttery, and fishy — again, traits you wouldn’t find in a purely plant-based product. Wildtype makes coho salmon saku, the most expensive part of the fish (usually reserved for sushi), which can run you $40 per pound and up. Some populations of coho salmon — the species Wildtype is cultivating — are threatened or endangered.
Neither is available for sale yet, as both startups are still in talks with regulators for approval. And neither BlueNalu nor Wildtype would disclose to me just how much it cost them to make what amounted to an appetizer’s worth of seafood — but it sure wasn’t cheap. While companies are bringing down the cost of cultivating meat from cells, it’s still a pricey endeavor, in the range of hundreds to thousands of dollars per pound in recent years. (In 2013, the first-ever cell-cultured hamburger weighed in at five ounces and cost $325,000 to produce, or around $1 million per pound.)
Other companies making cell-cultured chicken, pork, and beef have a long way to go to get within striking distance of the low price of commodity meat, which has benefited not only from economies of scale, but lax environmental, labor, and animal welfare regulation, along with decades of subsidies and government R&D. Chicken can run as low as $1.50 per pound in the US, with pork and beef in the range of $4 to $10 a pound. Some experts say the cell-cultured meat companies will never catch up.
But seafood — especially sushi-grade salmon and tuna — has a much more exclusive price point, one that is likely to only rise as overfishing taxes wild fish populations. That changes the economics for cultivated seafood and startups like BlueNalu and Wildtype.
/cdn.vox-cdn.com/uploads/chorus_asset/file/24289088/GettyImages_630979134.jpg)
Toshifumi Kitamura/AFP via Getty Images
“If we can start with a more premium product and more premium cut of that fish, it eases the economic journey considerably,” said Justin Kolbeck, co-founder of Wildtype. “I think, honestly, the folks working on chicken are gonna have a really hard time, right?”
It’s an approach common in novel technology, from cellphones to Teslas: Start with a high-end product for the early (affluent) adopters, then scale up to reach a price point low enough to feed the masses and displace the incumbents.
And it shouldn’t require extra production costs to make a rare cut of fish versus one that is still relatively abundant in nature. “The complexity between growing a very high-value product and growing a lower-value product is negligible or nonexistent,” said Kate Krueger, founder and CEO of Helikon Consulting, a biotechnology consultancy. Shiok Meats, a startup in Singapore, is making cell-cultured crab and lobster, two species that also tend to be expensive, as well as shrimp, which tends to cost less.
In the simple evolutionary calculus of numbers, the billions of pigs, chickens, and cattle we breed and farm for food have been wildly successful. But their enormous population not only leads to enormous suffering, through the way they’re torturously factory-farmed, but it also endangers us: Growing demand for meat is a major contributor to climate change, and the waste from billions of animals chokes the world’s rivers and streams. The hope in growing red meat and poultry from cells is that it could lead to a shrinking of domesticated animal populations and their heavy carbon and pollution footprint.
The opposite has been true for fish. Growing demand for seafood has come at the expense of wild populations — one-third of fish stocks are overexploited, a threefold increase since 1974, according to the United Nations. And the International Union for Conservation of Nature recently added more marine species to its Red List, citing illegal and unsustainable fishing as leading culprits.
Fish farming was supposed to relieve the pressure of commercial fishing on oceans, but that hasn’t come to pass. As fish farming has grown six-fold since 1990, the rate of wild-caught fish has plateaued, not declined. We farm fish so that we can eat more of them, not so we can preserve wild populations.
If the cell-cultured seafood startups can figure out how to grow fish from cells and compete with the fishing industry on cost, they could not only prevent the suffering of the individual fish we catch or farm, but they could also play a role in preserving threatened wild species.
The case for cultivating seafood
Americans don’t eat a lot of seafood (at least, relative to our insatiable appetite for red meat and poultry). But globally, it’s consumed at a higher rate than poultry, and just a little lower than red meat. And depending on how you look at it, the seafood industry is either a better alternative to the factory farming of mammals, or one of the more destructive forces on the planet.
Fish is on the whole healthier than pork and beef, but it can also present a unique health hazard, with its mercury, parasites, and microplastics. Fish has the lowest carbon footprint of all meats, yet it’s uniquely harmful to oceanic ecosystems and marine life — commercial fishing is a major threat to coral reefs, and discarded fishing gear accounts for 10 percent of plastic found in the ocean.
Wild-caught fish only suffer briefly during slaughter, as opposed to the weeks or months a chicken or pig suffers on a factory farm on land. But now more than half of the estimated 1-3 trillion fish we eat each year — who we believe can feel pain — are raised in fish farms, where they suffer from many of the same problems as farmed chickens: disease, overcrowding, and painful slaughter.
/cdn.vox-cdn.com/uploads/chorus_asset/file/24289099/GettyImages_1245441427.jpg)
Wojtek Radwanski/AFP via Getty Images
The seafood supply chain is also dizzyingly complex, lacking the traceability of most other meat products. For example, a recent study found that out of 105 samples of fish used in sushi, ceviche, and poke dishes at a smattering of Orange County, California, restaurants, almost two-thirds were mislabeled. That won’t be true of cultivated seafood.
“We can be just exquisitely precise about where something was made,” said Aryé Elfenbein, co-founder of Wildtype. “Exactly who made it, when it was made, what the transportation looked like, and just levels of information and detail that by far exceed what we see in conventional seafood.”
The lack of empathy for fish and the opaque, globalized supply chain have made commercial fishing even harder to reform than the factory farms that churn out red meat and poultry. A group of UK and Canadian researchers estimated in 2009 that one out of every five wild fish sold is caught through illegal, unregulated, or unreported activity. That could make cultivated seafood an important tool in the fight to clean up oceans and reduce the suffering of fish, if it can indeed displace wild-caught and farmed fish.
But as is often the case for any kind of new technology, whether or not cultivated seafood can actually solve welfare and conservation challenges will come down to dollars and cents.
Making cell-cultured fish affordable
Startups like BlueNalu and Wildtype have their sights set on first cracking into the high-end sushi market, with hopes of eventually cultivating more affordable species to make a dent in the 179 million-ton global fish trade. Wildtype’s Kolbeck said that if they ever do take over a sizable portion of the seafood market, it won’t be until he and his co-founder have a lot more gray hair.
“This has the ability to actually transform how we eat animal proteins for the better, unquestionably, but the journey there is really hard,” Kolbeck said. “It’s going to take the better part of our professional lives to see this done.”
/cdn.vox-cdn.com/uploads/chorus_asset/file/24288723/Wildtype_Salmon_Nigiri_Topped_With_Finger_Lime___Photo_Credit__Wildtype_a.jpg)
Wildtype
That realism is welcome, but technical progress is being made, if slowly. In October, BlueNalu announced two advancements in its goal of making bluefin tuna toro. The first was growing cells in suspension. Cells usually require some kind of surface on which to grow, so single-cell suspension saves precious space inside the bioreactor, meaning more cell-cultured tuna per batch.
The second is what it calls its “lipid-loading” technology, in which the company, through how they process the cells and what they feed them, has figured out how to make the cells grow a customized amount of fat — which, as the name suggests, is the critical component of fatty bluefin tuna toro. Lauran Madden, BlueNalu’s chief technology officer, said most tuna fat is 30 to 40 percent saturated, 25 percent omega-3 fatty acids, and the rest is other unsaturated fats. But they can tweak these amounts, enabling them to generate products based on what kind of fat profile a consumer would want.
“We can actually create cells, if we want, that have exactly the same profile [as wild-caught fish], or we can create cells that, for example, might have 90 percent omega-3 [fatty acids],” Madden said. And instead of growing fat cells in one bioreactor and muscle cells in another, they can be grown at once, further saving on costs.
Despite the scientific advancements and frenzied investor interest, cultivated seafood companies are still very much in uncharted seas. Researchers at the University of California Santa Barbara are skeptical they can succeed — meaning displace conventional seafood — due to the technical and economic challenges of producing it affordably at scale, the subsidization of the fish industry, and the challenge of securing consumer acceptance of novel food products.
Researchers also say that fish farming likely hasn’t had any kind of conservation effect and instead has increased overall demand for seafood. That could happen with cell-cultured seafood too, but it may not be a fair comparison.
Some species, like tuna, are practically impossible to farm, meaning that unless we grow them from fish cells, we have to take them out of the ocean. Fish farming has also been advertised as a more sustainable way to produce seafood, but much of it still depends on commercial fishing, as many farmed fish are fed wild-caught fish. Cell-cultured seafood avoids the plundering of oceans altogether.
Of course, if culturing fish cells never becomes economically feasible, lab-grown seafood will become, at best, a technological delicacy. But that’s the gamble dozens of investors have eagerly taken, considering the oceanic stakes and the potential windfall to be earned.
For at least half of homo sapiens’ 300,000-year run, we’ve ventured into the ocean for food. Today, fish remains a critical source of sustenance and income for many, but with a population that has surpassed 8 billion, we’ll need to figure out how to feed the world without further decimating oceans and rivers. The scale of the problem requires imaginative thinking, even as wild as making fish cells swim in bioreactors instead of out in the open water.
The long shadow of James Burnham’s The Managerial Revolution.
Since the launch of the so-called “Twitter Files” — Elon Musk’s self-styled exposé of the alleged excesses of the “woke” managers of Twitter before he bought it — there’s been a lively debate over what exactly Musk is trying to accomplish.
His statements clearly indicate that he sees himself as being engaged in some kind of culture war — “The woke mind virus is either defeated or nothing else matters,” as he tweeted on Monday morning. But what does the highly public censure of Twitter’s previous management team have to do with fighting the “woke mind virus”?
To understand how Musk sees the connection, it’s helpful to look at a recent tweet from Antonio García Martínez, a writer who’s very plugged into the world of right-leaning Silicon Valley founders. García Martínez describes a project that looks something like reverse class warfare: the revenge of the capitalist class against uppity woke managers at their companies.
“What Elon is doing is a revolt by entrepreneurial capital against the professional-managerial class regime that otherwise everywhere dominates (including and especially large tech companies),” García Martínez writes.
On the face of it, this seems absurd: Why would billionaires who own entire companies need to “revolt” against anything, let alone their own employees? To explain, García Martínez cites a book by conservative political theorist James Burnham: The Managerial Revolution: What Is Happening in the World.
Published in 1941, Burnham’s book predicted that capitalism had reached a terminal stage; the capitalist class’s power would soon decline, giving way to the rise of the “managerial class” — people who direct industry and the complex operations of the state. His examples of this new state were Nazi Germany and Soviet Russia, which Burnham believed to be rooted in a more efficient economic model than liberal capitalism. For this reason, he predicted a Nazi victory in World War II was all but inevitable.
Burnham’s predictions were wildly wrong, in ways that should cast significant doubt on the viability of his entire theory of “the managerial revolution.” But his conceptualization of an unaccountable managerial class has nonetheless been extremely influential in the right-leaning tech world and in the broader conservative intellectual firmament.
Marc Andreessen, a leading tech venture capitalist, called Burnham’s work “the best explanation for the current structure of our society and politics.” Julius Krein, a leading conservative policy intellectual, wrote that Burnham was “enjoying something of a revival” because “David Brooks, Ross Douthat, and Matthew Continetti, among others, have recently pointed to his work as essential to understanding the current political moment.”
So, while Burnham’s work got some big things wrong, it’s still worth taking seriously. In many ways, he’s the progenitor of the right’s current cultural obsessions with so-called “woke managers” — and the godfather of the approach to politics that Musk has spent $44 billion advancing.
Who is James Burnham, and what is the managerial revolution?
Burnham, born in 1905, began political life as a Marxist of the Trotskyite persuasion — but would decisively break with this tradition and become a stalwart of the American right. After Burnham’s death in 1987, then-President Ronald Reagan said that he was “one of those principally responsible for the great intellectual odyssey of our century — the journey away from totalitarian statism and toward the uplifting doctrines of freedom.”
The Managerial Revolution, one of Burnham’s earliest and most influential works, shows signs of both his youthful Marxism and his later turn toward anti-communist conservatism. It begins with the very Marxist idea that history is, at root, a story of conflicting social groups struggling for control over a society’s wealth and means of production.
“It is a historical law, with no apparent exceptions so far known, that all social and economic groups of any size strive to improve their relative position with respect to power and privilege in society,” he writes.
Burnham also concurred with Marxists that the capitalist class would inevitably lose the contemporary iteration of this struggle — but disagreed about who would win it. In his view, the working class was too weak and disorganized to overthrow the capitalists in the way that Marxist theory predicted. Instead, he argues, a new group was rising: the managerial class.
Managers, in Burnham’s definition, are the people responsible for “the tasks of the technical direction and coordination of the process of production.” This does not mean technical experts, like chemists or architects, but the people that direct these technical experts: “operating executives, superintendents, administrative engineers, supervisory technicians; or, in government…administrators, commissioners, bureau heads, and so on.”
The managerial class’s growing strength stems from two elements of the modern economy: its technical complexity and its scope. Because the tasks needed to manage the construction of something like an automobile require very specific technical knowledge, the capitalist class — the factory’s owners, in this example — can’t do everything on their own. And because these tasks need to be done at scale given the sheer size of a car company’s consumer base, its owners need to employ others to manage the people doing the technical work.
/cdn.vox-cdn.com/uploads/chorus_asset/file/24289332/542352569.jpg)
Ullstein Bild via Getty Images
As a result, the capitalists have unintentionally made themselves irrelevant: It is the managers who actually control the means of production. While managers may in theory still be employed by the capitalist class, and thus subject to their orders, this is an unsustainable state of affairs: Eventually, the people who actually control the means of production will seize power from those who have it in name only.
How would this happen? Mainly, through nationalization of major industry.
“An economic structure based upon state ownership of production provides the framework for the social domination of the managers,” he writes. “It must also be noticed that this apparently is the only economic structure through which the social domination of the managers can be consolidated.”
Given the complexity and scale of modern economic tasks, Burnham argues, it is simply more efficient for the state to take over from individual capitalists. Nazi Germany and the Soviet Union, with their state-directed economies, provided a blueprint that the rest of the world would soon follow. The United States hadn’t made the leap, but the rise of the administrative state after the New Deal proved it was heading there.
Hence, “the managerial revolution”: the inevitable decline of capitalist democracy and the rise of a new social regime defined by managerial control of the economy using “the unlimited state” as a vehicle.
How James Burnham explains Elon Musk and the right’s obsessions
At first blush, Burnham’s theory is not an especially promising candidate for explaining the world today.
Virtually all of The Managerial Revolution’s major predictions — the coming collapse of capitalism, an Axis victory in World War II, the superior efficiency of state-run enterprises — were all proven wrong. The power of the capitalist class has become more entrenched since the neoliberal revolution of the 1970s and ’80s and attendant skyrocketing inequality. The rise of tech capitalism, with firms founded by individual innovators and technical experts, seems to disprove his theory that capitalists cannot themselves perform technical and management tasks at scale.
Yet Burnham’s early thought has in fact experienced a renaissance of late, including in unexpected quarters: the right-leaning titans of Silicon Valley and allied political thinkers. Why?
The answer, in brief, is the culture war. The right’s new Burnhamites have revived his theory of managers as a distinct social class — the one, in their view, most responsible for imposing the malign ideology of “wokeness” on the American public.
“Woke managers want to impose a new political and social order,” Malcom Kyeyune argues in City Journal, a publication of the right-wing Manhattan Institute. “Wokeness has accomplished what New Dealism never set out to do in the 1940s: it serves as a comprehensive, flexible, and ruthless ideology that can justify almost any act of institutional subversion and overreach.”
Modern Burnhamites have come to see their most fundamental enemy not so much as the Democratic Party but as a series of institutions that they believe have been fully captured by this ideology: academia, Hollywood, the media, and (for some) Big Tech. These institutions make up what Curtis Yarvin, a Silicon Valley entrepreneur and anti-democratic political theorist, calls “the Cathedral”: America’s true power elite, one that uses its cultural hegemony to impose far-left values on everyone else. (Yarvin, not coincidentally, is deeply influenced by Burnham.)
The brute reality of economic inequality creates some difficulty for this theory. In today’s America, the billionaire class has far more power to shape society than random university professors or even most celebrities. Their sheer wealth allows right-wing billionaires like Musk and Peter Thiel to do things, like purchase Twitter or shut down Gawker or bankroll a Senate campaign, that are unthinkable for 99.999 percent of Americans. When these people exist, it seems implausible to describe the Harvard faculty office or Twitter Trust and Safety team as the true loci of social power.
Burnham’s theory helps the modern right square this circle. If power is increasingly moving away from capitalists and toward the managers they employ, then it’s totally coherent for even the wealthiest people in the country to see themselves as victims of a “woke mind virus” infecting middle and upper management. This is how you get the odd spectacle of people like Musk deploring alleged censorship perpetrated by their own companies: They see their staff not as subordinates whose conduct is an internal company matter, but as rivals in the struggle for power who must be defeated.
/cdn.vox-cdn.com/uploads/chorus_asset/file/24289347/1242720271.jpg)
Michael Gonzalez/Getty Images
“In Burnham’s formulation, this new managerial class would supplant the former business-owning bourgeois and even capital itself as the elite ruling class,” García Martínez writes. “Most woke ‘labor’ scandals in tech are an entitled middle-management class at odds with founders.”
The Burnham of 1941 likely would not have agreed with this analysis. In his view, the managerial revolution could never take place as long as capitalists still owned the corporations where they worked — that the right cannot coherently adopt his theory of a managerial revolution while jettisoning the underlying material analysis.
“Within capitalist society the power of the managers is… interfered with, limited by the capitalists and by capitalist economic relations,” he writes. “The manager is never secure. He can always be fired by someone or some group of persons possessing capitalist ownership rights.”
The real lesson of Musk’s ownership of Twitter is that this remains true today.
Can tech bros save the Pentagon? The messy tale of Rebellion Defense.
Rebellion Defense set out to disrupt the way the Pentagon handles new technologies. Silicon Valley’s elite and Washington’s national security leaders lined up behind the startup. Three and a half years later, Rebellion is falling short.
Founded in 2019, Rebellion wants to create AI-powered software for the military, intelligence agencies, and law enforcement. The startup has claimed its tools could sort through heaps of sensitive data to help officials make decisions, and that it will ultimately build software capable of making battlefield decisions. Its backers are as big as they come, with high-profile investors like former Google CEO Eric Schmidt. Even Obama’s personal IT guy joined. Rebellion has also won several million dollars of military contracts from the Pentagon, the start of what some observers think will be a gusher of government dollars. In 2022, the company raised $150 million in funding and was valued at a staggering $1.15 billion.
Now some industry experts say Rebellion has failed to meet its own mission, and some former employees allege the company has been stymied by problematic internal politics. Many startups face disarray, but the dangers are bigger here, far beyond the potential waste of taxpayer dollars on products that don’t actually exist. That’s because Rebellion is quickly developing military technologies, according to former staffers, without ethical guardrails on which governments products would be sold to or how they would be used. (Rebellion responded to this by sharing an “Ethical Principles” page from its website.) At worst, Rebellion’s ambition to automate decision-making could lead to algorithms with lethal power. Think Skynet in the Terminator films.
Recode spoke with seven former Rebellion employees who, speaking on the condition of anonymity, alleged that the company is mired in dysfunction, due to a toxic workplace. Two of them said that Rebellion’s products are still not market-ready. And the startup has been sloppy: Sources claimed that classified material, which is typically handled on secure platforms, has been shared in Rebellion’s unsecured Slack channels and Google documents, which poses risks to US national security. Spokesperson S.Y. Lee disputed this and said Rebellion “strictly complies with applicable government regulations.”
Rebellion is just one startup in a new generation of billion-dollar companies focused on selling to the military. If and when the technologies they’re building — including next-level facial recognition and autonomous decision-making — reach the battlefield, they could usher in an age of algorithmic warfare with lives at stake. In the coming years, AI may be capable of making decisions on whom to target abroad or eventually in the US.
“The stakes are different when you’re talking about creating a market for some plugin for Google Chrome or something like that, versus creating a market for weapons systems and surveillance technologies,” said Jathan Sadowski, a researcher at the Emerging Technologies Research Lab at Monash University. Rebellion’s software might endanger the soldiers using its products and, Sadowski emphasizes, “the lives of the people who are the targets of these technologies.”
The Silicon Valley mindset has led to breakthroughs in apps and smartphones. But is the move-fast-and-break-things culture what we want shaping the future of war?
Revolving-door power
Rebellion’s name evokes the good guys in Star Wars. In 2019, Rebellion opened up a slick office in downtown DC, around the corner from the Apple Store. In the common room was a mural of two hands gripping a lightsaber set against the Washington Monument. Employees one-upped each other with Star Wars kitsch at their desks. The deodorizing spray in the bathroom was labeled “Emperor Poo-Patine.”
Pop culture references aside, Rebellion aimed to capitalize on the revolving door between the government’s seat of power and Silicon Valley. The startup proclaimed to rebel against the entrenched government bureaucracy and, by extension, the military-industrial complex, in order to advance new technologies. Rebellion also wanted to bring Silicon Valley talent — and that fun, laid-back office culture — to Washington, hoping that big contracts would follow. The company needed connected people to make that happen.
Chris Lynch has been the company’s chief hype man. The onetime Microsoft employee co-founded the company with lawyer Nicole Camarillo, a former chief strategist for US Army Cyber Command. They met, and began secretly dating, while Lynch served as the first director of the Pentagon’s Defense Digital Service. Lynch was the force behind JEDI, a $10 billion Pentagon project to bring it onto a universal cloud. (The Defense Department ended up canceling JEDI amid lawsuits and controversy.)
In 2019, Lynch and Camarillo began pitching investors on AI products for the military, cybersecurity for ultra-classified data, and using AI to make satellite imagery easier to read. Rebellion’s pilot product, Iris, was a comprehensive battlespace awareness software that automates decision-making with AI, not unlike Skynet.
To create software as a service and sell it to the military, Rebellion relied on its network. Two Rebellion engineers landed seats on the Biden transition team in fall 2020. Then Rebellion adviser David Recordon, an accomplished Facebook alum who had also been Obama’s closest tech staffer, went to the White House as director for technology and special assistant to the president.
In addition to Eric Schmidt, who had led powerful national security advisory bodies in Washington, the company invited media mogul James Murdoch and Nick Sinai, deputy chief technology officer of the US during the Obama administration, to join its board. Rebellion also built out a fleet of former officials from both parties hyping the company in Washington. Pentagon budgeting virtuoso Bob Daigle became Rebellion’s chief operations officer. Jane Lee, who was Mitch McConnell’s adviser on appropriations, linked Rebellion to Congress.
The two years since Rebellion launched have also seen a steady flow of investment. By 2021, it had secured over $150 million at a $1.15 billion valuation. At the time, Axios reported that the funding “reflects a new mindset among venture capitalists, who long avoided defense startups.”
In September, after his stint at the White House, Recordon returned to Rebellion as chief technology officer. By that time, the company had grown to about 300 employees. But the ranks were full of unrest.
What Silicon Valley can do for the Pentagon
Lynch wanted to create a tech company suited for military contracting, in response to what he saw as the piqued tensions between Silicon Valley and the Pentagon. Rebellion’s campaign to tie down military contracts went to the core of tech companies’ complex relationship with national security. But for all the hype around Rebellion’s cutting-edge technologies, former employees claim it’s unclear how big those contracts are or how well its products work.
Lynch specifically thought that Rebellion could offer a solution to the controversy surrounding Project Maven, a Defense Department program launched in 2017 to integrate AI into the military. One of Project Maven’s primary goals was to use the private sector’s state-of-the-art software to label and sort the thousands of hours of footage captured from US drones flying abroad. Through another contractor, the military hired Google in late 2017, but once workers in Mountain View learned about the project, thousands of them revolted, outraged that the work had been kept secret. Googlers also said the mixing of surveillance technology with artificial intelligence set a dangerous precedent. This spurred a broader ongoing debate about the ethics of Silicon Valley partnering with the military to create technologies that could kill. Google did not renew the contract.
/cdn.vox-cdn.com/uploads/chorus_asset/file/24288357/601059584.jpg)
Andrew Harrer/Bloomberg via Getty Images
Lynch, on the other hand, said the partnership between tech and the military was the future of both industries. Google’s problem, according to Lynch, was that the company didn’t go far enough in standing by the military’s patriotic mission when faced with internal criticism. Rebellion even described itself as “an unconstrained ‘Project Maven’” in an early pitch deck. And in early 2021, Rebellion did win a Maven subcontract to analyze captured enemy materials, like the cellphone data of enemy combatants, “to improve decision making for users within the military community.” Though the contract was only for $700,000, it seemed like exactly what Lynch had set out to do in the first place.
Things did not go as planned. Rebellion was so excited about the project that it pledged to use five to six times the resources that it would make from the contract, making a bet that its products would gain further momentum across government. But former employees allege the development team was frustrated with the terms of the subcontract and the amount of data Rebellion had access to. The startup ultimately decided to stop working on this aspect of Project Maven, according to two former employees. When asked about the incident, a Rebellion spokesperson told Recode it does not comment on specific contracts.
Rebellion has continued to work on a project related to Maven: a $650,000 obligation to colorize satellite imagery so that AI can more easily label it. But when the Pentagon’s Joint Artificial Intelligence Command hired Rebellion for a $600,000 contract to do something similar, the experiment failed, and the Pentagon discontinued the contract, according to a former senior Pentagon official familiar with the contract. Rebellion disputes this version of events but declined to comment specifically.
In 2022, Rebellion has secured only about $5.5 million in public US military contracts. That seems small given the startup’s $1 billion-plus valuation. Contracting experts say that, over time, a company would need to reach about $100 million of government work to realize such a valuation. By comparison, the rising software company Two Six Technologies has about $105 million in obligations for the year, while BigBear, an AI company with similar funding to Rebellion’s, has roughly $16 million. Rebellion told Recode that it has “paying customers across all of our products” and is “experiencing rapid growth,” but did not specify who those customers were or which products were generating that revenue.
Meanwhile, some have suspicions about how Rebellion advertises its products as AI solutions. Many of the products rely on data processing, which just isn’t nearly as sexy as machine learning. “They don’t really have any AI products,” one former Rebellion engineer alleged to Recode. “If I put it shortly, I would say, yeah, they’re doing AI, but none of it’s valuable,” said another former employee. A Rebellion spokesperson described its technology as “proven.”
The stakes of rush-developing military AI may seem to clash with what former employees describe as incompetence at the company. But even the seemingly underdeveloped products pose ethical concerns and could lead to unproven technologies in the hands of government officials with major potential for misuse.
One investor claims that Rebellion has been too focused on what are sometimes called “Made for Pentagon” products, chasing military contracts rather than creating innovative solutions. “‘Made for Pentagon’ companies don’t work, never were going to work, and are self-evidently bad ideas,” said Joseph Malchow, founding partner of the investment group Hanover.
Yet Rebellion’s inner circle has achieved the distinction of military tech influencers, glowingly profiled in the Atlantic, Fortune, and trade magazines. Lynch, usually in a hoodie, has taken the stage at South by Southwest and the premier Aspen Security Forum.
“Rebellion is like the Fyre Festival led by Jar Jar Binks,” said another former employee.
Rebellion’s Nova cybersecurity product, according to three former employees, does work. The Naval Postgraduate School, Special Operations Command, and the Air Force have purchased licenses for it. But two former employees emphasized that none of Rebellion’s products are being used operationally in the field by the US military. (Rebellion disputes this.)
/cdn.vox-cdn.com/uploads/chorus_asset/file/24288367/1234996377.jpg)
Andrew Harrer/Bloomberg via Getty Images
Rebellion has also done work to modernize surveillance systems for US Special Operations Command, as part of a previously unpublished contract obtained by Recode. At $633,600, Rebellion’s portion of the sub-award is small compared to those of other participating contractors that work on internet surveillance, cellphone tracking, and satellite imaging. According to Jack Poulson of the watchdog Tech Inquiry, “Rebellion was one of many players, and it was far from the most interesting player on this contract.”
Along the way, Rebellion capitalized on the exaggerated threat of a US war with China and Russia to drum up appeal. A confidential product overview from 2021 obtained by Recode says that AI-enabled software is “the defining factor for deterring global conflict” and “the final frontier for a software revolution.” That rhetoric hasn’t convinced everyone.
“The technology that they’re building is not revolutionary or innovative,” said Sadowski upon reviewing the document. “What Rebellion has seemingly done is just put it in the language of military-speak.”
Many of its other products remain undeveloped, according to former employees, who told Recode that mismanagement and disorganization are to blame.
“It reminds me of an early-stage WeWork or a Theranos. They just keep raising money,” said one former employee.
Rebellion has not lived up to its stated values
Employees from Amazon, Microsoft, and Netflix took a leap to join a military-focused company because they bought into Lynch’s appeal to “top-notch rogue agents” to do meaningful, patriotic work for American service members that was grounded in values (and competitive salaries).
Like Silicon Valley, military contractors have struggled with diversity. That’s why Rebellion made inclusion so central to its pitch to new employees. The company went out of its way to portray itself as progressive, in stark contrast to the culture at multibillion-dollar military tech startups founded by conservatives, like Peter Thiel’s Palantir and Palmer Luckey’s Anduril. On social media, Lynch declared that the company needed “extraordinary, diverse, and brilliant people to stand up and help define the future.” Blog posts conveyed the importance of Women’s History Month and Black History Month to Rebellion employees.
Despite his stated commitment to progressive values, Lynch often skipped diversity programming that Rebellion conducted on Juneteenth and Black History Month, per former employees. (According to Lee, “Rebellion has several forums to commemorate cultural and historical occasions with involvement and support from Chris Lynch, Rebellion’s leadership, and employees.”) Former employees pointed out that, apart from Camarillo, the board members appeared to be all white men. Meanwhile, Rebellion’s managers ignored women’s perspectives in meetings, according to several employees. Comments that verged on racism went unaddressed by management, two employees alleged to Recode. In recent layoffs, Rebellion dismissed a manager advertised as the leader of its DEI Council.
“Rebellion has a strong track record of investing company resources and time into supporting DEI initiatives since our founding,” and it “does not condone inappropriate behavior in the workplace,” said Lee, the Rebellion representative.
The lack of underrepresented employees in leadership roles at a military tech company can lead to products with inherent bias and inherent racism.
“You want that diversity and inclusiveness to make sure that different perspectives and experiences are embedded into the systems that you’re developing,” said Merve Hickok, an AI ethicist at the University of Michigan. Otherwise, she added, “You’re going to be not only replicating, repeating, but also deepening those inequalities and injustices.” (Rebellion said in a statement that 80 percent of product leaders come from underrepresented groups.)
One image that stuck with one former employee: a large poster board used to demonstrate the capability of Rebellion’s facial recognition software. On it were the faces of celebrities, Santa Claus, and Osama bin Laden, but the majority of the faces were people of color, especially Middle Easterners, displayed on the movable whiteboard in the office. It made employees from underrepresented groups feel particularly uncomfortable.
“I remember one time I looked at the poster, and I looked at the people around, and I looked more like the people on the poster,” the former employee explained. “Psychologically, I was feeling gross about what I was doing.”
/cdn.vox-cdn.com/uploads/chorus_asset/file/24288571/GettyImages_946069720.jpg)
Michael S. Williamson/The Washington Post via Getty Images
Rebellion had posted a section on its website about the company’s values: transparency, empathy, diversity. But transparency did not include a statement on how the company’s products would be used, whether by non-democratic governments or by agencies with reputations for misuse, like Immigration and Customs Enforcement, which irked employees. Another statement on Rebellion’s website says, “We work with the military and intelligence agencies in the United States, the United Kingdom, and allied nations,” but offers no further details.
“One of the things that came up was, ‘Can we make pledges not to sell this stuff to law enforcement, or the Saudis, stuff like that,’ and they wouldn’t take any firm stance,” that same former employee said.
Other surprises threw employees. Co-founders Lynch and Camarillo had been engaged in a romantic relationship since before Rebellion launched. Former staff said they were demoralized to learn in an all-hands meeting that Camarillo, who served as the head of talent for two years, was dating the CEO.
Employees knew they couldn’t say anything about that, due to the cliquey nature of the workplace, but they did try to raise substantive issues to management, like how to improve products or basic procedures. Feedback to management was not welcome, they say.
“If you came in saying anything that would disrupt the echo chamber, despite the radical transparency and empathy, you’d get fired,” said a former employee. (Rebellion said that, in an anonymous survey, 70 percent of Rebellion employees said they “feel free to speak my mind without fear of negative consequences.”)
Data from LinkedIn shows that the median employee tenure at the company is about a year. This fall, many key personnel blasted off. Oliver Lewis changed his title on LinkedIn from UK chair to co-founder, a telling indication as Rebellion let go the majority of staff in its UK office. The engineer who had served on Biden’s transition left military contracting altogether. Chief operations officer Bob Daigle, the Pentagon budget wizard, no longer appears on Rebellion’s website.
Rebellion had set out to solve the Pentagon’s software problems and go even further by rebelling against the military-industrial base. The billion-dollar valuation obscured systemic shortcomings within the company. The influential board and staff insulated Rebellion from criticism. Rebellion has yet to deliver on the hype of its lofty goals of disruption.
As one former employee wrote on GlassDoor: “Dumpster fire.”