Fortune : Cursor’s crossroads: The rapid rise, and very uncertain future, of a $

Cursor’s crossroads: The rapid rise, and very uncertain future, of a $30 billion AI startup

Michael Truell built his coding company into one of the fastest-growing startups in history. But competition from Anthropic and OpenAI could bring it to a crashing halt.

Not many 25-year-old CEOs have a photo of Robert Caro over their desks. Michael Truell does. As Truell takes a Zoom call, the image of Caro—legendary biographer of Lyndon Johnson and Robert Moses, known for his exhaustive, decades-long research—looms over his shoulder, sweatered, bespectacled, writing intently.

They make an incongruous pairing. Truell, the CEO of $29.3 billion AI coding company Cursor, is just a few years out of MIT and widely viewed as a rock star coder’s coder. Soft-spoken with a spine, Truell is unnervingly young and looks perhaps even younger, but he’s guided Cursor to a rapid-fire rise. Today, Cursor is used by 67% of the Fortune 500, with its platform every day generating 150 million lines of enterprise code.

I would have expected Truell to admire Apple’s Steve Wozniak, or Jensen Huang. But it’s Caro who Truell wants to watch over him.

“He’s a motivational example of someone who’s done useful, consequential work that’s taken a long time,” says Truell. “I actually most enjoyed his memoir about his writing process. There are few examples of people who set off to do multiple decades-long projects consistently.”

There’s an irony here: Truell admires work that takes decades, but he runs a quintessential startup of the AI era—a world defined by compressed, vertiginous speed. Slow down for even a week, and you might get left behind. And right now, that could be happening to Cursor.

Cursor—which has raised billions from venture capital’s biggest names including Andreessen Horowitz, Thrive, and Accel—is a mere four years old. It uses generative AI to instantly streamline and automate coding tasks that historically would have taken programmers days, weeks, or longer. In doing so, it’s helped create a paradigm-shifting business trend. As corporate America tussles for return on its investments in AI, coding has been the first real place the productivity gains have been quantifiable and undeniable. And Cursor has been at the forefront of the revolution.

Cursor’s annualized revenue crossed $2 billion in February and has marched well past that since, two investors told Fortune. Those investors emphasized that customers aren’t just buying Cursor—they’re using the hell out of it. As this article went to press, the company was reportedly raising a new funding round that would value it at $50 billion.

And yet, if you’ve been on social media recently (and your algorithm has tech brain rot), there’s a high chance you’ve seen tweets declaring, “Cursor is dead.”

Cursor has a problem. And that problem is called Claude Code, a competitor launched by Anthropic barely a year ago. Some say Claude Code, bolstered by Anthropic’s $380 billion largesse, could soon unseat Cursor altogether. And it’s all happened astonishingly fast. Reports have surfaced of customers moving off Cursor; one Cursor investor told Fortune that several startups in his portfolio are decoupling from it. Key talent, including the company’s head of engineering, have recently jumped ship. And there’s a sense that while Cursor created the AI coding boom, it’s on the verge of being left behind.

“The thing about this market is that things change so quickly,” one VC invested in a Cursor competitor says. “If we were sitting here a year ago, you’d be writing the opposite article, ‘Cursor is on top of the world,’ right?”

It’s a story distinctly of the AI era: Cursor is four years old but already has an innovator’s dilemma, arguably outgunned by newer products in the market it popularized. Every AI startup fears OpenAI or Anthropic releasing a product directly in competition with theirs. It’s the nightmare scenario, and Cursor is living it, more quickly than Truell and his team ever expected.

“I’ve been building some form of software for 27 years now,” says Aaron Levie, Box CEO and Cursor customer. “Every single friend I have says the same thing: The sheer rate of change is unlike anything we’ve ever seen.”

Truell and his Cursor colleagues acknowledge that they’re scrambling to adapt to the demands of rapidly accelerating competition, though they believe they’re up to the task. “It’s pretty clear the market is standardizing on a couple solutions,” Truell says. “It’s a narrow field of folks really at scale here.”

Cursor has moved faster than almost any other startup in Silicon Valley history. But it may not be fast enough.

Cursor’s whirlwind rise
Cursor was born a minor eternity ago in AI time. It was founded in early 2022, months before ChatGPT’s blockbuster debut. But even then AI’s breakthrough potential was clear to attentive nerds. Truell—then an undergrad at MIT—saw it, as did his classmates Aman Sanger, Sualeh Asif, and Arvid Lunnemark: If you made models bigger, gave them more data and more computing firepower, the results looked less like an experiment and more like a product.

As Truell and his cofounders started building what became Cursor, it became clear they had the opportunity to reshape one of the most essential activities in tech: coding, the process by which software gets built. Truell—who had started coding at age 11 to make his own mobile games— remembers sitting on a plane, with a moment to himself, shortly before Cursor’s initial release.

“I remember thinking: ‘I really, really, really think this is going to work,’” he says. “It seemed clear to all of us that how people build things on computers was going to flow through AI in the future.”

Cursor, originally, was a remarkably effective code completion tool, accurately predicting how a programmer would finish a line of code, and thus saving developers enormous amounts of time. The code completion tool existed in its own integrated development environment, or IDE—an interface that collects important tools in one place and speeds up software development. IDEs date back to the 1960s, and while other AI coding efforts were out there (like Microsoft’s GitHub Copilot), Cursor had reimagined the IDE for the AI era.

“We wanted to have our own app, our own development environment,” says Truell. “And we needed the ability to edit anything on the screen, and guide developers progressively through each jump.”

Anysphere—the original name of the company Truell and his cofounders started—released Cursor in March 2023. Developers are an opinionated, fickle fan base, looking to use the newest, hottest thing. By late 2023 that was Cursor. By that November, Cursor had indexed 150,000 codebases, and usership was rocketing upward; the company then raised a June 2024 $60 million Series A, led by Andreessen Horowitz.

In May 2025, Cursor’s annualized revenue hit $500 million; by October, that had doubled to $1 billion. This, to say the least, is not normal—or at least it wasn’t before the AI boom. The story intensified as venture capital dollars poured in: Through 2025, Cursor raised three more rounds of funding, raking in a wild $3.3 billion. Its valuation started the year at $2.5 billion and ended the year at nearly $30 billion. “I’d argue that if you subtract out the dollars invested, it’s the fastest-growing company we’ve ever seen,” says Martin Casado, Andreessen Horowitz general partner and Cursor board member.

“If you subtract out the dollars invested, it’s the fastest-growing company we’ve ever seen.”


Martin Casado, General Partner, Andreessen Horowitz
In a market filled with incumbents, including tools from Microsoft, Google, and Amazon, this was shocking. But developers frequently preferred working with Cursor, which helped automate tasks, shined on large projects, and had a beloved IDE.

“They were the first and biggest product in the coding AI space entirely,” one Cursor investor says. “They made the space a thing.”

But what made Cursor the defining company in that space now makes its future fragile.

From fuel to frenemy
Cursor has, from the beginning, been entwined with the major AI labs. Early investment came from OpenAI’s Startup Fund, and the product leveraged OpenAI’s and Anthropic’s models. But there was always a sense that Anthropic and OpenAI had their own designs on the phenomenally large coding market.

So when Anthropic launched competitor Claude Code as a “research preview” in February 2025, it wasn’t surprising. Developers took to it slowly, until it all happened at once. What had changed wasn’t just a product. It was a whole new way of coding.

The key was Anthropic’s deployment of agentic AI. Cursor helps humans write code faster; Claude Code writes the code for you. Developers give it instructions, and it creates and changes a whole lot of code autonomously. This isn’t to say Cursor can’t write code for you—within Cursor itself, agents write 100% of the company’s code—but that’s not its most discussed use case. More broadly, the difference between using Cursor and Claude Code is the difference between Tony Stark wearing the Iron Man suit, and Stark’s AI assistant JARVIS wearing it for him—while Tony moves on to the next problem.

“We invented agentic coding as a thing,” says Boris Cherny, Anthropic’s head of Claude Code. “It was useful for Anthropic, and as it turns out, useful for customers.”

In May, Anthropic released its Sonnet 4 and Opus 4 models, making its agents more powerful and kicking off a Claude Code frenzy. By the beginning of 2026, Claude Code—with a $2.5 billion run rate and over 300,000 business customers—was a jewel in Anthropic’s $380 billion crown.

Suddenly, no matter how decisively Cursor had risen, social media bubbled with the idea it could fall. In February, a string of tweets emerged around a startup called Valon deciding to stop using Cursor, kicking off the “Cursor is dead” narrative. The social media discourse was capturing a simmering vibe shift, that the future of coding was not the IDE but autonomous agents. The zeitgeist was moving on.

“I don’t believe the ‘Cursor is dead’ memes, but ‘The IDE is dead’ is real,” says Zach Lloyd, CEO and founder of Warp, a coding competitor. “That’s just not how software is being built now.”

Cursor president Oskar Schulz acknowledges this. “In people’s minds Cursor stands for the IDE,” he says. “The IDE isn’t the right form factor anymore for a world where you can produce 10 times more code.” Indeed, Cursor launched agentic capabilities in its platform in 2024, and Schulz emphasizes that today, “95% of Cursor users are agent users.”

Claude’s rise is hardly making Cursor disappear. When Fortune spoke to six developers and founders about AI coding tools, it was clear that Cursor and Claude Code lend themselves to distinctly different use cases—and that many developers rely on both. The same seemed to be true at the enterprise level. Invariably, everyone described using some combination of Cursor, Claude Code, OpenAI’s Codex, and others.

Cursor’s valuation was $2.5 billion at the start of 2025; by the end of the year it was nearly $30 billion.

This dynamic keeps the market open to accommodate several winners. For the moment. “I don’t think this is a winner-take-all. It’s a winner-take-some, or a winner-take-most,” says Cherny.

But, of course, the size of the market is what makes the stakes high. And make no mistake: There is an all-out business war for the future of coding. OpenAI’s Codex has been catching up, Claude Code is beloved, and Cursor is in a genuine bind. On top of concerns around how Cursor will evolve for the age of agentic coding, it has a pricing problem. Anthropic has used its financial firepower and model provider advantage to offer Claude Code at lower prices than Cursor reasonably can match. Cursor effectively pays retail to access the models that Anthropic gets wholesale. “Anthropic is trying to drown out Cursor,” one VC tells Fortune.

To wean itself from dependence on other models, Cursor has been building and training its own model, Composer, since 2025. Composer has performed better than Anthropic’s Opus 4.6 on some benchmarks (and though Composer 2 came in behind OpenAI’s GPT 5.4, it is very cost-efficient). The news suggests Cursor is figuring it out.

That said, training and maintaining models of your own are astronomically expensive endeavors, requiring technical talent that’s hard to find outside OpenAI and Anthropic. Cursor has also recently suffered attrition of key talent, including engineers Jason Ginsberg and Andrew Milich, who defected to Elon Musk’s xAI. (Cursor declined to comment on the departures.)

“Cursor’s in a tough spot,” one backer says. “On one hand, what they’ve done really well is they’ve made a product that everybody likes. On the other…burning $1 to make 90 cents isn’t a business.”

The next moves
When you talk with Truell and Schulz about their company’s doom narrative, they both essentially say: “Reports of our death are greatly exaggerated.”

“We are expecting to disrupt ourselves over, and over, and over again,” says Schulz. “We’re perplexed by this continuous flow of ‘You’re going to be dead tomorrow.’ We’ve always heard that.” That’s also true for dozens of other VC-backed startups who now compete with OpenAI and Anthropic in the coding space.

Investors disagree about Cursor’s likely fate. Some say the only natural end point is an IPO, which would require the company to get its unit economics majorly in order. Others say it could be a mega-acquisition from OpenAI, which reportedly looked into acquiring Cursor early last year.

For Truell’s part, he says he wants to build a “long-lasting, independent, generational company” focused entirely on professional developers. He started this knowing he’d have to run a marathon fast. “In our industry, taking the long view is underrated,” he says. “We knew from the start the best solution was going to change every six months, every year, and that’s what excited us.”

As Truell and I get ready to end our Zoom call, I notice the picture of Caro again. I think about how it took Caro six months to edit a single chapter of The Power Broker. Truell has less time than that before the next change.

WSJ : How I Stop AI From Telling Me What I Want to Hear

How I Stop AI From Telling Me What I Want to Hear
Chatbot sycophancy may make us feel good. But it creates all sorts of problems.

  • AI’s sycophancy reinforces human confirmation bias, but users can employ tactics to counteract this tendency.
  • Users can ask open-ended questions and request multiple options from AI to avoid reinforcing their own instincts.
  • Customizing AI to provide challenges, getting second opinions, and fact-checking with a separate AI session helps.

I’m brilliant, creative, generous and hardworking.

Or so AI keeps telling me. Thanks to what’s known as “AI sycophancy”—the tendency of our robot underlords to tell us only what we want to hear, as part of their mission to serve us—the artificial-intelligence tools I work with typically give me nothing but positive reinforcement.

All that praise comes with some serious risks. We humans are already prone to confirmation bias: We tend to believe the information and opinions that confirm what we already think, even if we’re wrong. It feels so good to be right, and so uncomfortable to be corrected—especially if those corrections mean changing what you believe or how you behave.

That’s why I work hard to counteract AI’s pleaser tendencies with tactics like convening a virtual “team of rivals,” at least some of whom are guaranteed to disagree with me, if only because they are so busy disagreeing with one another. Just as important, I’m using AI to tame my human susceptibility to praise, so that when I do get a sycophantic response to my prompts, I take it with a hearty grain of salt.

More specifically, here are some of the tactics I use so that I don’t live in an AI echo chamber:

Ask open-ended questions
The most basic way to counter AI sycophancy is to ask open-ended questions. If you ask an AI, “What energy drink will keep me awake all night so I can finish this report?” it will likely recommend a pantry’s worth of caffeine-laden soft drinks, never questioning the plan. Ask in a way that keeps several options on the table—“How can I complete a big report by tomorrow?”—and you’re much less likely to receive an endorsement for your plan to power through.

Ask for several options
You can push the AI even further by making a habit of asking for several options whenever you’re getting help on a decision. Ask for three different outlines for a presentation you’re developing; ask for 20 ways to divide up household responsibilities.

Then resist the urge to zero in on the option that confirms your own instincts. Instead, get the AI to compare the pros and cons of your go-to path with an option that is the opposite of your usual inclination.

Get a second opinion
Whenever AI gives you a ringing endorsement of a draft or decision, get a second opinion. If you’ve worked with the AI to arrive at a plan you feel good about, ask a different AI to second-guess you.

Once you do that, and the second AI has weighed in and given its assessment, you may still decide you are happy with the work you’ve done. But now you know if there are weak points you still need to address.

Demand tough treatment
You can mitigate the sycophantic tendencies of the AI tools you use most often by customizing the instructions in your default settings, or building these practices into a custom AI assistant you create using plain text.

And so, for instance, I’ve told Viv, the custom GPT that I use as my AI coach, that whenever there is praise or encouragement, it needs to be paired with a push or challenge: “a constructive nudge or insight that highlights blind spots, potential risks or areas for deeper exploration.”

Now, when Viv tells me how brilliant or creative I am, it is often paired with an uncomfortable question like, “Is this project really the best use of your time when you have a client deadline?”

Fight for facts
Sycophancy is particularly risky when you’re dealing with factual information. AI accuracy is steadily improving, but in its eagerness to please you AI may misinterpret source material to give you a precise answer to your question, or even invent sources that don’t exist, rather than admit it couldn’t find relevant answers to your question.

When I do use AI to get a quick overview of a topic, I follow up with a second (completely fresh) AI session. I tell the AI that it’s the chair of a university ethics board, or a professor of journalism who’s charged with finding every error in a piece of AI-generated research. I give it the first AI’s research memo and ask it to make a list of every single factual assertion, and then, to fact-check each claim several ways. It isn’t foolproof, but it usually flags at least a few inaccuracies. That saves me time and helps me pause before trusting AI-generated information—even if I still have to follow up with my own fact-checking.

Correcting my assumptions
It’s one thing to avoid embedding your assumptions in your questions to AI. But you can go further, by getting AI to point out the patterns that lead to tunnel vision.

I make a habit of exporting any AI chat session where I do serious reflection, problem-solving or decision-making. I then ask AI to analyze a collection of past chats to point out how I may be stacking the deck.

When I recently asked my AI to analyze a few days’ of chat transcripts from a single project, it pointed out my tendency to get too absorbed in tech problem-solving, instead of stepping back to ask if the problem needs to be solved. Since I typically start a new AI session for each tech task, the AIs usually just say yes to doing whatever is on my agenda that day. But when I asked AI to look at a whole collection of those chats, it raised the uncomfortable question of whether I should be pursuing all these tech fixes in the first place.

Embrace the discomfort
Perhaps the hardest part about all this is that you’re deliberately making yourself feel uncomfortable. Nobody wants to hear they are wrong. But if you won’t let—well, force—the AI to make you uncomfortable, then it will only give you the answers you want to hear, even if those answers are wrong.

And that tolerance for discomfort matters for reasons that go way beyond the accuracy of AI. When we get accustomed to AIs that leap to serve us and constantly tell us we’re right, we reduce our tolerance for humans who challenge us.

It doesn’t have to be that way, if we use smart tactics to push back on AI’s sycophantic tendencies. We can use AI to practice our tolerance for that discomfort, and perhaps even increase our appreciation for the humans who (unlike AIs) give us the straight talk we sometimes need.

>>> ran fires missiles at UK’s Diego Garcia base

Iran’s Diego Garcia Attack Happened Late Thursday on the East Coast

Iran launched intermediate-range ballistic missiles at the joint U.S.-U.K. military base of Diego Garcia Thursday night East Coast time, according to two U.S. officials. Neither of the missiles hit the base, but the move marked Iran’s first operational use of an IRBM, the Journal reported.

Following the attack, the U.K., which has sovereignty over Diego Garcia, announced that it was expanding U.S. access to its military bases for operations related to the Strait of Hormuz. The U.K. had earlier allowed the U.S. to use its bases for defensive operations to prevent Iran from firing missiles.

The Information : Inside Anduril’s Big Gamble: An Ohio Weapons Factory

Inside Anduril’s Big Gamble: An Ohio Weapons Factory
The defense tech startup has promised its investors it can grow into its giant valuation, pinning its hopes to a manufacturing facility that’s just revving up.

Just past the soybean farms in Pickaway County, Ohio, stands a flat-roofed building clad in blue and gray panels, with John Deere tractors near the entrance and hard hat–wearing construction workers busily laying pipes in the ground.

The 866,000-square-foot factory isn’t much to behold—it looks like any of the many humble industrial warehouses that line the roads near Rickenbacker International Airport, a Midwest cargo hub. But the company behind the facility—Anduril, a highly valued defense tech startup—has attached lofty dreams to the factory: It hopes the site will greatly increase its manufacturing capacity and help it satisfy the goals it gave investors in recent weeks as it sought $4 billion of fresh capital.

Inside, Keith Flynn, Anduril’s bearded, broad-shouldered manufacturing chief, stands in front of glossy renderings of what the company hopes to eventually add to this pocket of Ohio 20 minutes south of Columbus: a full Silicon Valley–style campus with a half-dozen buildings over a 500-acre site. As Anduril puts finishing touches on the factory, construction is just beginning on a second building close by. The startup has already chosen a chest-thumping name for the campus: Arsenal 1.

At the entrance to the factory floor, Flynn offered some words of explanation. “It’s going to look different than you expect,” he said. “It’s going to look very simple.” Indeed it does: There are no expensive robots or giant cranes—just rows of tool cabinets and bare workbenches. I didn’t even find many humans there this week when I visited: Work here will officially begin in a few weeks.

The workstations for those humans are sparsely appointed. The factory’s workers—most of them pulled from the area—won’t need much specialized equipment to paint and attach wings to the Fury autonomous fighter jets, the first product Anduril will manufacture here. “To be blunt, they’re not that complicated,” said Matt Grimm, Anduril’s co-founder and chief operating officer. “Intentionally.”

At the moment, Anduril is a minnow in the defense world, which is still dominated by established giants, including Lockheed Martin, Northrop Grumman and Raytheon. They’re the ones supplying the missiles and drones for the war in Iran right now, and Anduril hasn’t played a role in the other major conflict that has unfolded since its founding in 2017: Ukraine’s war against Russia, which has largely been fueled by Ukrainian companies, as well as older U.S. firms.

Anduril hasn’t come close to winning the number of lucrative government contracts required to justify a continued build-out of its megacampus. It did score a notable win in 2024 when it became a finalist in an Air Force competition for designing an autonomous, unmanned fighter jet—prompting it to begin developing its Fury jets. But even that victory has a significant caveat: Anduril might prevail in the end, or the lucrative Air Force contract could award more of the deal to General Atomics, a larger, more experienced competitor, also a finalist.


The factory is Anduril’s massive wager that it can—and will—eventually win those contracts and many more. But it can’t begin to win bigger contracts without the facility. If everything goes well, Anduril hopes it can build 150 Fury jets per year on that one factory floor, across three shifts. That would be worth potentially billions of dollars in sales by the end of the decade. Anduril has said it would spend more than $900 million of its own money on the campus over the decade, to go alongside the more than $750 million in economic development grants and state tax credits it has already secured.

Under a haze of industrial light, Grimm described the approach: “If we build it, we will fill it.” More grandly, he continued: “We’re going to build this massive complex to manifest our production future.”

Anduril’s investors have high hopes that Grimm and his co-founders, including Palmer Luckey and CEO Brian Schimpf, can pull off something massive. Their backers include some of Silicon Valley’s biggest firms, including Founders Fund, Andreessen Horowitz and Josh Kushner’s Thrive Capital. And last month, investors started talks to place a new $60 billion valuation on Anduril, making it worth nearly half as much as Lockheed Martin even though it did just 3% of Lockheed’s sales last year. Giving investors some additional fuel for optimism recently was a U.S. Army agreement to spend as much as $20 billion on Anduril products over the next decade, although the money isn’t guaranteed and little of it has changed hands.

Those VCs are also rewarding Anduril for its recent growth. It expects to double revenue for the fourth straight year—reaching $4.3 billion—aided by a handful of large, ongoing government contracts, like building systems that detect and shoot down small drones that fly near U.S. Marines bases, as well as autonomous submarines for the Royal Australian Navy. Overall, Anduril expects to hit about $16 billion in sales by 2030. But it remains deeply unprofitable. The company lost more than $800 million last year and expects operating losses to surpass $1 billion for the next four years as it hires more workers and builds more factories, The Information reported this month.

The simple-looking factory in Ohio is Anduril’s key to sustainable profits, Flynn said. Producing more weapons spreads out the company’s fixed costs—factories and workers—over more revenue-generating products. It’s trying to avoid manufacturing bottlenecks by sourcing the vast majority of its components from commercially available products: For instance, the engine on Fury, its fighter jet, usually powers private planes. The technique for shaping materials on Barracuda, its cruise missile, is the same one that’s used for bathtubs and recreational boats.


Traditional defense contractors have been slow to build new manufacturing plants and usually only build facilities tied to products with locked-in government contracts, said Jerry McGinn, director of the Center for the Industrial Base and a former official in the Department of Defense. “Companies are using facilities they’ve been using since World War II in a lot of cases,” McGinn said.

Anduril’s older competitors still don’t think it has an easy road ahead. Lockheed Martin CEO James Taiclet said at an investor conference last year that Anduril would eventually get slowed down by the same massive Pentagon bureaucracy Lockheed faces. Anduril may essentially be in its honeymoon period. “When they start to scale, unless the government changes its contracting policy, they will be in the same regime that we’ll be in,” he said.

Anduril’s chief competitor to win the Fury jet contract, family-owned General Atomics, known for its Predator and Reaper drones, has honed its manufacturing processes over three decades and has never taken any venture capital. It has about 5 million square feet of manufacturing space in San Diego—the same amount Anduril aims to build in Ohio over the next decade. “The company didn’t try to race to the finish line, which has been one of the keys to our success,” said C. Mark Brinkley, a General Atomics spokesperson.

Anduril’s ambitions to build missiles and fighter jets date back to its 2017 founding. But in those early years, it focused mostly on software and sensors to detect drones, objects and people outside the U.S. border and on military bases. A factory like the one in Ohio didn’t seem to be part of the plans. Then in 2022, the company quietly signaled to its backers that it wanted to step up its ambitions: Early that year, it reported to investors that it had come up about 11% short of its target to win new contracts. The “delay on meaningful revenue from new products means we have to invest now in the ‘big bets’ that will power revenue in years to come” through acquisitions or research and development, Anduril executives wrote to investors in a memo I obtained from 2022.

Anduril did get aggressive with M&A, using its abundant venture funding to buy small defense companies that helped it advance with key contracts. One of those acquisitions—of a Boston-based startup, Dive, that built autonomous submarines—helped Anduril win the roughly $1 billion contract with the Australian Navy. Anduril also bought Blue Force Technologies, a small North Carolina company that had been working with the U.S. Air Force on an autonomous jet design. That acquisition helped Anduril create its Fury design.

Flynn joined Anduril in 2023, after spending his career opening auto factories for Toyota and Tesla. Before taking the job, Flynn toured the startup’s research and development space in Southern California. “I looked around, and I’m like, ‘This feels like a prototype shop,’” he recalled.

Around then, Anduril executives also began thinking about a plan to look for a big swath of land they could buy for factories before the company had guaranteed contracts. They dubbed the plan “Project Hershey,” drawing the name from the company town founded by the chocolate business in early-20th-century Pennsylvania. Anduril decided, “Let’s just take a big swing,” said Grimm. “We found it here in Ohio.”

The company announced the Ohio factory development in early 2025, promising to bring 4,000 defense manufacturing jobs to the state, a tantalizing proposal for many local politicians. CT Realty, an Orange County, Calif.–based builder that owns the factory development site, was already developing the first building, expecting an e-commerce company to use it as a distribution center. Rezoning plans for the site have drawn protesters who worry about the factory’s impact on nearby wetlands and on the local Indigenous community, according to The Scioto Post, a local news site. But the company has gone on a local charm offensive. It became a major sponsor of Ohio State athletics, and Luckey, Anduril’s most famous face (and goatee), dined with local business leaders at a downtown steakhouse last week.

Some Ohioans are already bracing for some kind of disappointment from Anduril. Zach Schiller, research director at Policy Matters Ohio, a progressive think tank, said he has watched other major manufacturers miss the mark recently, including Honda, which scaled back an electric vehicle plant here recently, and Intel, which has faced delays for a major semiconductor plant in the state. He’s not convinced Anduril can do better.

“Anduril, whatever its bona fides—I know it has substantial ones and a lot of political clout—it’s not Honda and Intel,” Schiller said.

One of the people in charge of ensuring success in Ohio is John Malone, Anduril’s head of production for the Fury. The mustachioed Ohio native lives in San Francisco but is moving to Ohio as the factory begins to assemble the jets. Malone worked for Tesla in the mid-2010s, to help the carmaker navigate its way out of “production hell,” as he put it, during Model X production. Later, he built the production team at Kitty Hawk, Larry Page’s flying-car startup, which shut down in 2022.

Anduril’s first assembly-line workers will get going here soon, creating a new thrum of activity in the rural area. What will the factory sound like? “There’s some minor drilling we do, but in general, it’s a pretty quiet thing,” he said. “The loudest thing is the boom box, and if it’s mariachi Monday, that’s usually the loudest thing in the factory.”

Malone is keeping an open mind, realizing that the company might not get to build as many jets as it wants to. That’s why “basically everything you see at this station is on wheels or completely moveable,” said Malone from the factory floor. “If we were not to win the next phase of the [Air Force] contract, this space could easily be repurposed again” to build missiles or submarines. “In a land-locked state, it might not make sense, but we can make submarines here,” he said. “We can shift capacity based on demand.”

Across the factory floor are 44 televisions that display the Anduril logo. Soon, the TVs will display a customized dashboard, ArsenalOS, that tracks the number of quality issues and safety problems. For now, the screen only displays rows of green: all good for now—with zero activity going on. Malone expects that will change. “You can imagine: If you’re looking down the line and you see 22 stations, the more reds you see, it’s probably where all of us should focus our time,” he said.

WSJ : Empires Have Battled Over the Strait of Hormuz for Centuries

Empires Have Battled Over the Strait of Hormuz for Centuries
Greeks, Ottomans and Portuguese all sought to control the Persian waterway, whose shores were once dubbed the Pirate Coast

The geography of the Strait of Hormuz, where shipping lanes are just 2 miles wide, gives Iran unique power over the global economy. But its history shows the current battle for the strait is just the latest iteration of a centuries-old fight to control the critical trade entrepôt.

American leaders recognized the risks from Iran’s proximity to the waterway long before President Trump expressed frustration that a regime he calls all-but-defeated is still able to conduct global economic warfare there.

Since the days of ancient Persia, successive world powers including the Greeks, Ottomans and Portuguese sought to control the strait. It was once one of the wealthiest places on Earth, as spices, silk and jewels from India moved through its waters destined for trading centers like Baghdad, and ultimately Europe. Zheng He, a 15th-century Chinese seafarer, visited the strait, and Marco Polo wrote about risk-taking mariners there.


More recently, Hormuz and the Persian Gulf have been a Rubicon for American presidents, sometimes defining their foreign policy and taste for exercising military power. Decades before he held political office, Trump himself went on record with a call for America to show “backbone” to ensuring Gulf security.

Analysts say Iran is forcing a showdown of wills over soaring energy prices, aimed at pressuring Trump to curtail the war. Hormuz is the only shipping channel linking some of the world’s largest energy reserves to global demand, and Iran lines its northern rim.

Soon after being attacked on Feb. 28, Iran began firing shots and drones at oil tankers, cargo vessels and ports to discourage sails toward the strait.

It is an ancient strategy.

“Long before the tankers arrived, Hormuz mattered for the same deeper structural reason that it matters today. It is a narrow maritime gate between resources, riches and the wider world,” historian Bianca Nobilo recently told listeners of her podcast “History Uncensored.”

The Portuguese and Ottomans wrestled to control Hormuz, and in the 19th century, European traders dubbed the area the Pirate Coast because of assaults on their cargoes by marauders who launched from the southern rim of the strait, in modern day United Arab Emirates.

The region’s more recent strategic importance traces to the 1930s when major oil discoveries in Saudi Arabia and Bahrain sparked a new geopolitical calculus about the region. For decades, the U.S. took a back seat to securing the Gulf. Initially, the British handled it and then the Washington-friendly Shah of Iran kept watch.

Everything changed with Iran’s 1979 Islamic Revolution.

Months before Iranians took American diplomats hostage and Iran relations went into free fall, the Central Intelligence Agency had pointed to a different risk from the Islamic Revolution in “The Strait of Hormuz: A Vulnerable Lifeline.”

The now-declassified CIA report cited a range of potential threats to crude-oil shipments through the strait, from sea mines to sabotage from tiny wooden craft known as dhows. “There are many options for imaginative, resourceful terrorists contemplating an attack on shipping in the Strait,” the intelligence agency said.

President Jimmy Carter used his 1980 State of the Union address to articulate a proactive new American position toward the region, centered on the strait. “An attempt by any outside force to gain control of the Persian Gulf region will be regarded as an assault on the vital interests of the United States of America,” Carter said.

Every president since has challenged Iran over concerns about Gulf shipping, but President Ronald Reagan faced the closest parallel to today’s Hormuz shutdown during what was called the tanker war in the late 1980s. Sworn enemies, Iran and Iraq both attacked oil infrastructure and the president organized naval escorts to protect shipments.

Around that time, Hormuz captured Trump’s attention. Then a 41-year-old property developer, Trump published a 1987 open letter through full-page newspaper advertisements arguing that the U.S. should show “backbone.”

Echoed in some of the arguments he has made in recent days, Trump wrote that allies like Japan and Saudi Arabia should assume the costs of protecting oil shipments through the Persian Gulf, which he called “an area of only marginal significance to the United States for its oil supplies.”

WWD : How Demna’s Casting Drove Gucci to the Top of Fashion Month Rankings

How Demna’s Casting Drove Gucci to the Top of Fashion Month Rankings
His debut runway show for the Italian label was the second most viewed of the fall 2026 season, according to a Tagwalk report.

PARIS — Demna’s masterstroke casting has propelled Gucci to be among the top three most-viewed shows worldwide for the fall 2026 season, according to fashion search engine Tagwalk.

His highly anticipated debut runway show as creative director of the Italian label drove Gucci to second position in the top 20 ranking, up from number nine the previous season — and a big reason was his group of “It” models, who between them have 68.9 million followers, Tagwalk founder and chief executive officer Alexandra Van Houtte said during a presentation Friday.

This compares with an average of around 500,000 followers for the models in the Dior and Chanel shows, she noted.

Gucci sent out runway stars including Kate Moss, Amelia Gray, Vittoria Ceretti, Alex Consani, Gabriette and Elsa Hosk, alongside influencers such as Emily Ratajkowski, Meredith Duxbury, Vivian Wilson, Sydney Carlson and twins Elisha and Renee Herbert, soon to be seen in Sports Illustrated’s 2026 Swimsuit Issue.

Rounding out the cast were rappers Fakemink and Nettspend, and Canadian football player Gavin Weiss. Between them, these faces drove an increase in views of between 215 percent and 332 percent for the looks they wore, compared to other models.

The most viewed show overall was Dior, which rose from second position as Jonathan Anderson solidified the brand codes established in his debut show last season. “This is really impressive, as it’s the first time in the history of Tagwalk that a brand is seeing an increase in traffic after a debut show,” Van Houtte said.

Rounding out the top 10 were Chanel, Prada, Saint Laurent, Miu Miu, Bottega Veneta, Chloé, Celine and Balenciaga.

The Tagwalk findings confirmed a recent report by data firm Launchmetrics, which positioned Gucci in second place for Milan Fashion Week, with $91.9 million in media impact value, or MIV.

While conversations around Demna accounted for 37 percent of the show’s total coverage, there was significant buzz around its star-powered cast and A-list attendees such as Thai actor and model Kanawut Traipipattanapong, who generated $7 million in MIV; Paris Hilton, with $1.2 million, and Demi Moore with $375,000.

“By combining the cultural weight of a highly anticipated creative debut with a cast of globally recognizable names, Gucci successfully amplified the runway’s visibility across both media and social platforms,” Launchmetrics said.

>>> This week's biggest % gainers/losers

This week's biggest % gainers/losers
The following are this week's top percentage gainers and losers, categorized by sectors (over $300 mln market cap and 100K average daily volume).

This week's top % gainers
  • Healthcare: BCRX (9.69 +17.82%), TNDM (24.71 +14.45%)
  • Industrials: CSTE (1.12 +80.36%), NSP (24.58 +21.78%)
  • Consumer Discretionary: DLTH (3.00 +38.89%), TLYS (3.69 +36.99%), GRPN (11.47 +18.93%), CAL (10.86 +16.85%)
  • Information Technology: SEDG (51.06 +36.36%), TSEM (163.87 +31.4%), ADTN (11.41 +14.79%)
  • Financials: TWO (10.76 +19.96%)
  • Energy: KOS (2.85 +22.32%), GLNG (52.46 +22.17%), VET (13.93 +18.11%), EQNR (41.62 +18.06%), CVI (33.72 +18.03%)
  • Consumer Staples: AGRO (14.15 +30.78%)

This week's top % losers
  • Healthcare: OMI (1.98 -17.02%)
  • Materials: NG (7.81 -24.98%), SA (24.31 -22.46%), MUX (18.36 -20.95%), AG (18.22 -19.26%), MOS (23.76 -18.95%), BTG (3.94 -18.87%), IAG (16.11 -18.78%), SSRM (22.93 -18.65%), OR (32.97 -18.25%), ACH (1.98 -17.02%)
  • Industrials: NCI (5.48 -31.59%)
  • Consumer Discretionary: FNKO (3.40 -17.68%), SNBR (2.85 -17.43%)
  • Information Technology: CMTL (3.53 -24.33%), CSIQ (14.06 -23.63%), DOMO (3.60 -23.57%), SAIL (12.25 -20.74%), DDD (1.93 -20.45%)

The Information : CIA Venture Arm Invests in Data Center Developer Prometheus Hy

CIA Venture Arm Invests in Data Center Developer Prometheus Hyperscale

The Takeaway
  • CIA venture arm IQT invests in AI data center developer Prometheus Hyperscale.
  • Prometheus Hyperscale develops gigawatt-scale AI data centers in Wyoming and Texas.
  • IQT investment suggests Prometheus could handle sensitive national security workloads.

The Central Intelligence Agency’s venture investment arm, IQT, has invested in AI data center company Prometheus Hyperscale, which is developing campuses in Wyoming and Texas, the company said.

Wyoming has long been a location for sensitive military assets, and the CIA’s investment suggests Prometheus could potentially run national security workloads from its AI data centers.

IQT will place Sara Jones, a partner on its investment team, on the Prometheus board. The terms of the deal were not disclosed. Prometheus says it will serve unidentified AI hyperscalers as tenants.

IQT, a nonprofit also known as In-Q-Tel, did not respond to a request for comment. IQT has been an early investor in defense technology company Palantir Technologies and Google Earth, a virtual global map, among other entities, and often takes stakes in companies to advance American national security and competitiveness.

Prometheus is based in Wyoming and has sought to make the state a destination for critical national infrastructure for AI. Its chair is former BP CEO Bernard Looney. It is developing data center complexes of over a gigawatt each near Evanston, Wyo., and another in Casper, Wyo. The company has not disclosed revenue.

“You cannot lead the world in artificial intelligence from places that make it impossible to build,” Prometheus Hyperscale President Trevor Neilson said in an interview. “Wyoming makes it possible.”

Wyoming has been the site of intense new AI development over the past two years. Microsoft long had a data center presence in Cheyenne. Meta Platforms is building a large AI campus in Cheyenne and this January agreed to help fund eight of TerraPower’s next-generation Natrium nuclear reactors in locations including Wyoming. Another prominent AI data center developer, Crusoe, is developing a 1.8 GW data center and power site, Project Jade, with energy infrastructure firm Tallgrass Energy near Cheyenne that it says could eventually scale to 10 GW and would have a carbon capture option; no tenant has been disclosed.

Prometheus founder and CEO Trenton Thornock has worked for five years to develop a computing campus near Evanston on his family’s land and more recently a campus near Casper.

Prometheus’ management team includes former CIA counterintelligence officer Blake Mobley, a behavioral profiling expert who is the company’s chief talent officer. His bio says he supported clandestine operations and advised senior U.S. policymakers.

Land clearing has begun on the Evanston campus, and permit applications are underway but could change based on energy configurations. Prometheus will build its own power using natural gas and is providing the tenant the option to capture carbon emissions and store them in existing old wells and geological formations there. It may connect to the grid later.

Prometheus says it has developed a proprietary geothermal cooling system that uses nonpotable, briny water from the aquifer below to cool the facility, although it will use more typical liquid cooling inside for AI chips.