請同意我們的隱私權規範，才能啟用聽新聞的功能。

（中央社舊金山20日綜合外電報導）科技媒體The Information今天報導，輝達（Nvidia）2名員工稱公司計劃在年底之前，開始小批量出貨一款專為中國客戶設計的AI晶片，以搶攻快速成長的AI推論市場。

路透社報導，目前已有數家中國客戶訂購這款晶片，這是輝達語言處理器（LPU）的一個版本，採用Groq授權的技術，可與輝達繪圖處理器（GPU）搭配使用，以加快AI聊天機器人的回應速度。

The Information報導指出，這款晶片符合美國出口管制規定，但目前不清楚北京當局是否批准相關銷售。

輝達未立即回應路透社的置評請求，路透社也無法獨立查證這則報導內容。

儘管輝達主導AI系統訓練市場，但在推論（inference）市場卻面臨更激烈的競爭。百度（Baidu）等多家中國大型企業，目前已自行生產推論晶片。

路透社今年3月曾報導，輝達正在籌備符合中國市場規範的AI晶片，積極爭取在快速成長的AI推論市場中保有競爭力。（編譯：張茗喧）1150821

Until both OpenAI and Anthropic get close enough to their planned IPOs to release their financials, we have to look to other sources for signs of how well their businesses are doing. One of those sources, Ramp, the corporate credit card and expense management company, has just released some surprising new data: OpenAI has started gaining on Anthropic with U.S. businesses.

OpenAI, which was once the runaway leader with both businesses and consumers, lost the lead among Ramp’s paying business users back in May. That’s when Anthropic hit 41% market share to OpenAI’s 39%. The ChatGPT maker has never regained that lead. As of July, Anthropic has nearly 44% to OpenAI’s nearly 40%.

The data covers more than 70,000 American businesses that spend billions via Ramp’s bill pay and corporate card products. Ramp’s customers are spread across industries but, as a popular Silicon Valley corporate credit card, they do skew toward the tech industry.

A closer look at the most recent data, according to Ramp economist Ara Kharazian, shows that OpenAI is currently growing faster among this segment in Q3 to date than Anthropic. Mind you, there’s still a month left in the quarter and that’s like 30 AI years, so the trend could easily shift again before it’s over. Ramp also declined to provide actual dollars spent, sharing only percentages.

To borrow ChatGPT’s own hedging style for a moment: This isn’t a measure of the total market. It excludes large enterprises that use spend-management tools from providers like American Express, rather than Ramp. But it’s enough data to show market indications. And what it shows is that Anthropic hasn’t won permanently. Businesses are willing to flop back and forth as each lab releases new models, volatility that should give both companies’ investors pause about how “sticky” enterprise AI spending really is.

“GPT-5.6 Sol is really good, increasingly the choice for developers,” Kharazian posted on X about OpenAI’s new growth. “Fable 5, meanwhile, disappointed both in adoption and real-world application given price + data retention requirements imposed by regulators,” he continued.

That may be an over simplification. Fable — Anthropic’s higher-end model tier — is expensive but it’s also built for a more targeted set of use cases than a general chatbot. Still, Anthropic did cause some outrage when it warned Fable users that it must retain their data for 30 days.

Ramp’s data also suggests that both companies should be growing business revenue, even as they duke it out for market share, because the market overall is expanding. The percentage of companies that pay for AI among these Ramp customers has been steadily climbing. It topped 50% in March. It reached nearly 56% by July.

On Wednesday night, Runlayer and Rippling dropped their respective lawsuits against each other. No settlement was made. No money changed hands. Not even lawyers’ fees, according to court documents seen by TechCrunch.

Rippling celebrated by instantly releasing its MCP gateway, the product at the heart of the dueling lawsuits and the one that competes with Runlayer’s offering.

This public fight is a cautionary tale to founders: In the age of AI, when building new software has become almost trivial, you never know who your next competitor will be. It might even be a prospective customer.

To recap the short-lived legal brouhaha: Runlayer is an early-stage startup that launched out of stealth in November 2025 and has raised a total of $42 million from VCs like Khosla Ventures’ Keith Rabois and Felicis. It’s led by third-time founder Andrew Berman (previous companies include baby-monitor maker Nanit and an AI video conferencing tool, Vowel, that sold to Zapier in 2024).

After Rippling tested Runlayer’s MCP gateway for more than a year, with the two engineering teams working closely together, Rippling never signed on to become a customer, according to Runlayer’s lawsuit. Instead, Berman received a text from a Rippling employee that said his employer was building its own MCP gateway and planned to release it as a product. This employee described Rippling’s product as a clone of Runlayer’s.

Runlayer sued, claiming that Rippling violated contractual agreements covering the tests of its products.

An MCP gateway securely handles an enterprise’s AI agent requests for data from other software systems. So, for instance, when a hiring professional asks for details on the top five candidates for a job, including their emails, that data must be retrieved from the company’s recruitment system. The gateway handles the retrieval process, rather than granting agents direct access to the company’s software systems. It can then also layer on other features like employee role-based access control (managers getting different access than interns), observability (logs and usage trails), and so on.

Then Rippling countersued, alleging that Runlayer was violating some of its patents. The move was seen by Runlayer as a way to induce it to drop its suit while ratcheting up legal expenses.

Runlayer dropped its suit after spending the last three weeks in discovery. Rippling dropped its own suit and didn’t collect a settlement either.

So, while the lawsuits didn’t lead to anything but a lot of public flaming, there is a deeper takeaway for founders. The AI landscape is changing so rapidly that the long-running technical shoot-outs that enterprises love to impose on startups need to be rethought. Between the time an AI startup enters into one and however many months later, an enterprise’s needs and desires may drastically change.

In the meantime, in the span of weeks, Rippling, whose bread and butter has historically been payroll and benefits management, has now entered the AI gateway market with a tool that can route to different models while dashboarding token spend by employee. The product is competing with the likes of Stripe, Ramp, and Databricks.

Now Rippling is also in the AI security business with this MCP gateway that ties AI access to employee roles. It competes with the likes of Runlayer, Docker, and Amazon Bedrock.

As for Runlayer, its pitch is a broader bundle of agent security services tied to the gateway, ranging from agent creation to spotting shadow AI agents running in an enterprise unbeknownst to IT.

In Brief

If you’ve ever wanted to share all of your digital conversations with OpenAI, we have good news for you: The AI lab has just launched an Apple Messages plug-in for ChatGPT, allowing interested users to connect their Messages inbox with the chatbot.

The benefits of doing this, OpenAI argues, are numerous. Users can use the plug-in to sort, analyze, or edit their messages directly from ChatGPT. The plug-in also works with Codex and ChatGPT Work, so users can use it professionally, not just personally.

A brief commercial advertising the new plug-in shows a user asking the chatbot to suggest follow-up messages to their contacts based on the messages received the previous day.

You can also ask ChatGPT to delete messages for you, draft and send messages on your behalf, or search for information buried deep in your message history.

As with most things related to AI, this new feature raises some privacy questions. OpenAI told Bloomberg that the plug-in runs locally on a user’s machine and that it “doesn’t create an index of all someone’s messages.” Still, the specifics of what that means aren’t immediately clear. TechCrunch has reached out to OpenAI for more information.

When it comes to message sending, OpenAI encourages users to keep an eye on what ChatGPT is doing and discourages turning on persistent approval, warning that doing so “removes your final chance to review a message before ChatGPT sends it as you,” the company writes.

I traveled to Las Vegas in May to watch the Enhanced Games — a one-of-a-kind sports competition that lets athletes compete while using the kind of performance-enhancing drugs typically banned in professional sports. The games, derided as the “steroid Olympics,” were put on by a telehealth company backed by the likes of Peter Thiel and staffed by veterans of the crypto, AI, and biotech industries.

The event ended up being more than a little anticlimactic. Hailed by its creators as an event that would fundamentally transform the world of organized sports, the games resulted in few exciting feats. Only one world record fell, and it came in swimming, a sport where records are broken often.

Now it can be said that, in addition to the games being a competitive flop, they were a commercial failure, too. Earlier this week, the Enhanced Group, the company that put on the games, posted its second-quarter earnings report, revealing that the company had suffered a net loss of nearly $62 million. Much of that loss came from hosting the games.

Enhanced Group, founded only several years ago in 2023, enjoyed an IPO earlier this year at a $1.2 billion valuation, and sells personalized health treatments via a digital telehealth platform. The treatments it sells are all FDA-approved, including peptides, testosterone injections, GLP-1s for weight loss, and other similar products.

The company’s Q2 earnings report says it brought in $17.7 million in the last quarter, but most of that money came from sponsorships tied to the games, not the telehealth business on which the company was built. Scant information is available about how that core business is doing. The report casts doubt on claims previously made by Enhanced executives that the games will be an annual event (the company will either have to start making a whole lot more money or be comfortable losing tens of millions of dollars a year).

Enhanced may already be signaling a pivot. The company’s report also highlights the recent launch of a new online series, Enhanced Breakers, which it says “operates at a fraction of the cost of a full Games event” while still keeping “athletes competing, audiences engaged, sponsors interested, and performance medicine in front of the world year-round.”

Enhanced’s own woes aside, the industry around it is gaining ground. The peptide business is booming, helped along by a recent decision from the Trump administration’s Food and Drug Administration to reclassify a number of substances that have long resided in a legally gray area. The government’s support doesn’t open the floodgates for the sale of those substances quite yet — an additional review process still needs to take place — but it highlights the government’s interest in deregulating the industry.

The FDA’s parent agency, the U.S. Department of Health and Human Services, is notably overseen by Robert F. Kennedy Jr., who has long been known for his unconventional views on health. Kennedy’s ideas have been castigated by health professionals both inside and outside of the government who have routinely characterized his thinking as dangerous — criticism that hasn’t slowed the industry’s momentum.

Silicon Valley remains one of the hotbeds for peptide startups, where companies like Superpower and Noho Labs are capitalizing on the tech industry’s penchant for biohacking and trendy health supplements. And the sector’s growth is outpacing the rules meant to govern it, with state governments struggling to keep up with regulatory schemes.

In a cheeky marketing campaign, Liquid Death teamed up with former Philadelphia Eagles star Jason Kelce to share a solution to mitigate the environmental impact of AI data centers, which require massive quantities of water to prevent servers from overheating.

“AI data centers waste millions of gallons of water,” Kelce quips in the video campaign. “That’s why Liquid Death and Garage Beer have teamed up. We want your pee to cool these data centers.”

Then, as a crowd of people walk through a field sipping their branded beverages, they sing in unison: “Let’s pee on computers together to save humanity!”

It’s a funny commercial. What’s even funnier is that Kelce has unwittingly stumbled upon a real tactic for cooling down data centers.

“The Liquid Death commercial is funny and tongue-in-cheek,” Michael Obradovitch, vice president of Data Center Global Accounts at Ecolab, told TechCrunch. “But in reality, there is a fair amount of alternative water sources already being used to a similar extent to cool these data centers.”

These alternative water sources, when used in data centers, at least partially offset the demand for potable drinking water. One such alternative water source is recycled water, which is made by treating wastewater and sewage water so that they’re safe to use again. Wastewater and sewage water contain many things, including — you guessed it! — human urine.

“You wouldn’t just use pee, but you can clean it and make it into useful water, and that’s what we advocate,” Bruno Pigott, executive director of the WateReuse Association and former acting assistant administrator in water for the U.S. Environmental Protection Agency (EPA), told TechCrunch.

To be clear: You should not actually contribute gallons of your pee to help cool data centers, as Kelce facetiously suggests. But just for the sake of the thought experiment: What would happen if you did try to cool a data center with a steady stream of pee?

“Pee contains all sorts of stuff. It contains salts, it contains urea, bacteria, organic matter of all sorts that can leave mineral deposits. If you just put that into a cooling tower or something else, it would require constant cleaning,” said Pigott. “One of the methods of cooling is called evaporative cooling, where hot air is passed through water to remove heat through evaporation. Can you imagine if you just poured urine through hot air?”

We do have the technology to turn our urine into potable drinking water — that’s what astronauts do in space, since they can only bring so much water with them on their spacecraft. But that isn’t efficient at a large scale, and even if it were, it’s not like scientists can just access millions of gallons of pee at will (well, not unless Kelce really commits to the bit). Instead, our toilet water ends up in wastewater and sewage.

That’s where water treatment facilities come in, providing recycled water to spare us from the smell of evaporated urine. These facilities use membrane bioreactors, reverse osmosis, ultraviolet light, and other processes to treat water until it’s clean enough for industrial use. In some cases, this water can even be treated to the point that it’s drinkable.

“We use recycled water for cooling for all kinds of industries, and we have for decades,” Dr. Greta Zornes, practice leader for water reuse at the engineering firm CDM Smith, told TechCrunch. “So this is only one application, but definitely, there’s been a boom in recycled water for data center cooling.”

Though Zornes has worked on water reuse technology for more than two decades, her day-to-day work has shifted with the rising demand for data centers.

“Every day right now, I’m working on recycled water for data centers,” she said.

When data centers use more recycled water, they don’t pose as much of a burden to the local potable water supply. But industries can only pivot to recycled water use when there is proper infrastructure in place to treat millions of gallons of water every day.

“You have to be somewhat near a waste water treatment facility that’s sizable enough that you have enough water to use,” Zornes said. “So when data centers go out into rural areas, a lot of times the wastewater treatment plants just aren’t big enough — they’re not treating enough water for them to be able to take it and treat it and use it.”

Aerial view of data centers in LOudoun County, Virginia. Image Credits:Gerville / Getty Images

Loudoun County, Virginia, located outside of Washington, D.C., is home to more than 250 data centers, with plans to construct at least another two dozen. As of 2025, Loudoun data centers collectively used about 200 million gallons of recycled water each day, but the water footprint of these data centers is so extreme that this only accounts for 43% of daily data center water usage in the area. The other 260 million gallons, or 57% of daily data center water usage, come from potable water supplies, according to Loudoun Water.

“There’s a lot of infrastructure that has to be built out and usually isn’t existing today, and that takes time,” Zornes said. “That’s one of the problems — it’s just the time that it takes to get that done.”

Obradovitch thinks that the AI industry could even drive resources toward building out this kind of infrastructure to scale water treatment. Meta, for example, will invest at least $270 million in wastewater infrastructure projects near its data centers (the company also loses about $4 billion each quarter on its Reality Labs division).

“That’s where data centers can actually come in and be anchors of water infrastructure,” Obradovitch said. “There’s a number of cases and examples where data centers, as part of their engagement with communities, have committed funding and capital to some of these municipalities to help in addressing some of those exact challenges.”

On the policy side, Pigott is advocating for legislation that would provide a 30% tax credit to help industries scale their recycled water infrastructure.

“We think it would greatly accelerate the pace with which data centers and other industries entered into this area,” he said.

While there’s some unintentional science behind Liquid Death’s joke, the commercial and its virality serve as a reminder to the tech industry that the environmental demands of data centers have become a mainstream concern. According to a recent Gallup poll, about seven out of 10 Americans oppose data centers in their communities, and AI products continue to face backlash from consumers who feel as though the technology is being forced into their lives.

“I’m glad that people are concerned about water, and anything that raises awareness of water, however crude it may be, could actually be beneficial,” Pigott said. “It gives us a chance to educate the public about what we’re doing today.”

If you are a malicious hacker, cybersecurity professionals may very well be the worst people in the world to try to hack, as there is a very good chance they are going to catch you.

A person pretending to work for a leading crypto news site targeted several cybersecurity professionals around the time of the Black Hat and Def Con hacking conferences earlier this month. The hacker approached attendees on the social media site X, both via public replies and DMs, and then leveraged Google Docs in an attempt to trick the targets into installing malware, according to researchers.

On Wednesday, security firm Huntress published a blog post detailing the hacking campaign, which targeted one of its researchers, who pretended to go along with it to learn what the hacker was trying to do.

In broken English, the hacker asked the researcher if they had plans to attend a conference next, and then mentioned a conference allegedly organized by the crypto news website, according to a screenshot of the conversation.

After that, the hacker shared a legitimate Google Doc that looked like it was a planning document for the fake conference. The document displayed a sidebar designed to make the target think it was encrypted. The goal was to first trick the target into entering a fake decryption key provided by the hacker. That was the first step in a process that would lead to the installation of malware for macOS and Windows, depending on the operating system used by the target, according to Huntress.

To make the sidebar appear real, the hacker used Google App Script, a platform that allows developers to customize the user interface of Google Docs with menus and sidebars, for example.

A screenshot of the Google Doc sent by the hacker to the Huntress researcher. Image Credits:Huntress/Screenshot

The hacker tried to trick Huntress’ researcher into installing an infostealer for Apple computers; a remote desktop viewing tool repurposed as malware for Windows; and a fake installer for the cryptocurrency wallet Ledger.

The person behind the account identified by Huntress researchers as the hacker did not respond when TechCrunch sent them a private message on X.

Hackers of all kinds — whether they are unknown government hackers using advanced spyware or North Korean government hackers using fake Twitter profiles — have targeted cybersecurity professionals before. What made this campaign a bit more believable was the use of a legitimate Google Doc and Google feature.

Google did not immediately respond when TechCrunch reached out asking if the company had seen this or similar hacking campaigns.

In Brief

Mark Zuckerberg just bought a cozy abode close to Meta’s international headquarters in Ireland. By cozy abode, we mean a castle. Zuck bought the 19th-century Strancally Castle and its 440-acre grounds, an estate estimated to have cost him anywhere from $23 million to $35 million.

“Mark and his family are excited to continue caring for this historic home and look forward to spending time in Ireland, where Meta maintains its international headquarters,” a spokesperson told the press about the purchase.

We’ve reached out to Meta for further comment.

The home is stunning: It sits on a river, surrounded by greenery, and the Irish Times reported in 2001 that it has at least 11 bedrooms, four tower suites, a library, a dining room, and a drawing room. Zuck bought the castle from a financier who’d lived there with his family for 25 years.

Meta’s international headquarters is in Dublin, where the company opened a 31,000-square-foot data center in 2017, just 20 minutes from the city. Waterford, where Zuck’s new castle is, is about a two-hour drive from Dublin.

This isn’t Zuck’s first big real estate purchase this year. He and his wife also reportedly splashed out $170 million for a home in the same exclusive Miami neighborhood as Jeff Bezos, joining a slew of tech billionaires leaving California as the state seeks to implement a form of wealth tax.

As AI continues to kill traffic to websites, Google on Thursday threw a bone to those publishers negatively impacted by the change. It’s now allowing readers to push a button on a publisher’s website to indicate it’s a “favorite source” they’d like to see highlighted more often across Google Search, Discover, and Google News.

The tech giant said it’s making this new, interactive “Preferred Sources” button available to online publishers to embed on their own websites.

The launch follows Google’s rollout of Preferred Sources in May to Google’s AI experiences, including AI Mode and AI Overviews. The option was previously available in Top Stories.

The idea is to make it easier for readers to find links from the sites they know and trust when they’re searching for content or interacting with Google’s AI to learn about a topic or read the latest news. As of May’s launch, the company said that people across the web had already selected over 345,000 unique sources through this method.

To add a site as a favorite publisher, you can visit Google’s source preferences page, then search for a publisher by name or website.

Becoming a preferred source can drive more traffic to publishers’ websites, Google said. In earlier studies, it found that people are twice as likely to click through to a preferred source when available. By offering publishers these additional tools, Google is trying to assuage the damage that the rapid growth of AI-powered search features has had on traffic-dependent businesses.

Alongside the new button, Google said that readers will soon be able to customize their Discover feed in Google’s app in their own words. To use this feature, readers will tap any three-dot menu in the feed and then tell Google what topics they’d like to see more or less of, using natural language commands. This helps Google refine the feed in real time.

Image Credits:Google

The search giant is not the only company turning to AI to offer feed-tuning tools powered by AI. In recent months, a number of top social media apps have launched user-controlled algorithms that allow people to fine-tune the content that is recommended to them.

In addition to personalizing the Discover feed, Google says Android users will be able to customize their audio daily briefings in the Google News app, as well.

In Brief

Castelion, a missile startup founded by former SpaceX executives, has raised a $1 billion Series C at a $13 billion valuation. The round was co-led by Andreessen Horowitz, Carlyle, and JPMorgan Chase.

Founded in 2022, Castelion set out to manufacture hypersonic weapon systems at a lower cost and at faster speeds than traditional defense primes. Since then, the startup has secured more than $500 million in U.S. military contracts.

The funding for the startup is timely. The Pentagon has been looking to procure more of these weapons, since the U.S.’s stockpile of hypersonic missiles capable of traveling at speeds above Mach 5 hasn’t kept pace with China’s. Castelion is betting it can help close the gap with its faster manufacturing process.

The Torrance, California-based startup will use the fresh capital to produce its Blackbeard missiles, named after a fierce English pirate, and other hypersonic weapons in its New Mexico production facility.

The funding round, which included participation from existing backers, including Lightspeed, General Catalyst, and Altimeter, comprised $800 million in equity and a $250 million revolving credit facility.

With back-to-school season approaching (or already here in some places), keeping track of everyone’s schedules can get pretty chaotic. Between work, school, appointments, sports, chores, and everything else going on, a regular paper calendar just doesn’t cut it. That’s where Linkdaze’s smart digital calendar comes in — a touchscreen tablet built specifically to organize a household rather than a single person.

One of Linkdaze’s biggest strengths is its calendar compatibility. The system can synchronize calendars from popular services, including Google, iCloud, Outlook, Yahoo, and Cozi, which is a dedicated family-organizing app. This is particularly useful for families where different members use different platforms. Instead of asking everyone to switch to a single calendar app, Linkdaze brings multiple schedules together and uses color coding to make individual family members easy to identify.

Launched last December, Linkdaze is available in 15.6-inch and 10.1-inch models, giving you some flexibility depending on how much wall space you have. Beyond calendars and appointments, you can use it for chores and rewards, meal planning, shopping lists, and other family organization. It can even double as a digital photo frame for displaying family photos.

Image Credits:Linkdaze

The most interesting feature, however, is Linkdaze’s AI meal planner with “Snap-to-Sync.” Instead of manually entering everything into a meal-planning app, you can take a photo of a paper recipe or your kid’s school lunch menu. Linkdaze will turn that information into a digital meal plan and generate a shopping list from it. While not an entirely new idea, it’s a useful feature that helps Linkdaze stand out from a basic digital calendar.

Another big plus is that Linkdaze doesn’t require a monthly subscription for its main features. It’s an interesting choice in a category where recurring revenue has become the default. Skylight, a competing smart-calendar brand, offers additional features through its $79 per month subscription. For a hardware company entering a crowded smart-display market, that decision is either going to differentiate its product or look like a lost revenue stream.

Linkdaze is also less expensive up front, with the 10.1-inch model priced at $119.99 compared with Skylight’s 10-inch model at $159.99.

Overall, this device could make a practical gift for busy parents who are trying to keep everyone’s schedules in one place. It could also be a great fit for college apartments, where roommates can use it to coordinate chores, study schedules, shared meals, and other household responsibilities. It’s also very helpful for those of us juggling interviews, deadlines, meetings, and story assignments.

An unusual glitch has resulted in xAI’s Grok chatbot speaking gibberish to many users.

After asking the model to generate a PDF, one user received the response: “match it without and your they and two for planets can practical and often cheese…” with similar nonsense continuing for several paragraphs.

Checking source links, another user found a string of links to reinforcement learning research sites.

Affected users who spoke to TechCrunch said they were using Grok Lite and noticed the issues as early as Wednesday morning. TechCrunch was unable to reproduce the issue in our own testing, and it is likely affecting only a small subset of users. xAI did not respond to a request for comment.

Still, the result has been a lot of confused Grok users. The chatbot’s Reddit community has been overwhelmed with complaints. Refreshing the session often restores normal function, but some users reported that the gibberish responses continued even after multiple refreshes.

The bug appears to be limited to direct queries on Grok.com. The Grok account on X.com has been unaffected.

Responding to unhappy users, the Grok account confirmed the issues on X:

“That pure word salad is a rare temporary generation glitch. Official status at https://status.x.ai shows all Grok services fully operational with no incidents. Start a fresh chat or regenerate—it usually clears right away. Sorry about the gibberish,” the account wrote Thursday morning.

xAI has seen significant staff turnover in recent months, losing most of its founding team and at least 50 researchers and engineers, according to a report from The Information in May. The company released its most recent foundation model in July, describing it as “an Opus-class model, but faster, more token-efficient and lower cost.”

TechCrunch Disrupt 2026 is built around one question for founders: How do you build an enduring company in the AI era? For investors, it’s the opposite: How do you find that company before someone else does?

At Disrupt, the volume and competition are the point for investors, not a drawback. Year after year, investors who’ve explored the Expo Halls, met founders, and learned from peers at panels and Side Events have proven why you need to be on the ground to get ahead of the curve.

Consider that every founder pitching during the Startup Battlefield competition does so in front of a packed room of VCs. Why? Because they know their next cohort could be sitting right there. All it takes is making the trip to San Francisco this October 13-15.

This guide is for the investors who recognize that three days spent at Moscone West are well worth it and who plan to walk away with the insights and prospects that yield returns.

And speaking of returns, we have up to $300 in savings with our current ticket tiers, which end Friday, August 21 at 11:59 p.m. PT, so act swiftly to get an even greater ROI on Disrupt.

Let’s dive in!

How Disrupt works for different investors

Disrupt features numerous tracks, depending on the check sizes you’re able to write and the stage of startup you specialize in.

Angel, pre-seed, scout

If you’re coming to build out your portfolio of early bets and get in on rounds before they’re priced, prioritize the Startup Battlefield 200 semifinalist pitches, the Builders Stage, and the Expo Hall.

Seed, Series A

Disrupt will be all about volume and speed of qualified deal flow. Prioritize the founder list, our curated 1:1 meetings, and the Deal Flow Café.

Growth, late stage

This stage is best suited for those coming to the event as much for market intelligence and strategic partnership scouting as for deal sourcing. Prioritize the Smart Money and Smart Systems Stages, as well as the exclusive StrictlyVC investor-only session.

Why Disrupt?

More than 20,000 curated meetings take place over just three days, within dedicated environments like investor receptions and structured networking sessions. Investor-founder connections aren’t hallway luck –they’re built into Disrupt’s infrastructure.

You get direct access to 200 pitch-ready, TechCrunch-vetted startups through Startup Battlefield 200. And we’ll put an emphasis on “vetted.” Our Startup Battlefield team, working alongside TechCrunch’s incredibly discerning team of editors and writers, has done a significant amount of diligence for you already.

Past speakers have included investors like Elad Gil and Vinod Khosla, and this year’s lineup puts you in the room with the operators you’re underwriting, too. We’re talking Rivian’s RJ Scaringe, Amazon’s Panos Panay, Replit’s Amjad Masad, and Cerebras’ Andrew Feldman, just for starters.

The Disrupt crowd is your signal. With a mix of 10,000 founders, investors, and operators filling Moscone West, the companies worth knowing will get discovered by someone. This is your chance to make sure it’s you.

Startup Battlefield 200 is a sourcing engine, not just a pitch competition

Image Credits:Kimberly White / Getty Images

For founders, Startup Battlefield is a visibility engine and a trial by fire. For investors, it’s a filtered shortlist that the TechCrunch team has spent months narrowing down for you.

Consider that:

200 pre-Series A startups, handpicked by our Startup Battlefield and editorial teams and sharpened through the SB 200 program, are competing for $100,000 in equity-free funding.

Battlefield alumni have collectively raised over $32 billion and produced 250+ exits, which is evidence we produce fundable companies.

Startup Battlefield is intensely competitive. Thousands of global startups apply, while just 200 make the cut and only a handful reach the finals. That funnel is doing your top-of-pipeline filtering for you before you ever take a meeting. If you’re not circling around these startups, your competitors are.

The exclusive access that an Investor Pass grants

We have several different passes for Disrupt, and by joining the community via an Investor Pass, you get access to perks like…

The Deal Flow Café

This is a space exclusively for founders and investors, fostering impromptu run-ins with founders actively seeking capital. Grab coffee or a beverage of choice, explore opportunities, and start conversations that can turn into your next deal.

The founder list

You’ll get early access to the full list of Disrupt founders looking for connections with investors. The next addition to your portfolio can be identified before the event even begins, giving you more time to find even more opportunities.

Curated meetings

Through the Disrupt app, you can schedule 1:1 and small-group meetings with founders matching your focus areas, with AI helping match you by mutual interests instead of waiting on fate.

The investors who make the most out of Disrupt don’t leave sourcing to chance — they’ve already scanned the available resources and set up their agendas in advance.

The Disrupt programming that matters most to Investors

Image Credits:Kimberly White/Getty Images

Our editorially curated programming, including all three days of sessions and panels, is available here, stretching across six stages, breakout sessions, roundtables, and Side Events. Here are some of the stages that might be of interest:

The Smart Money Stage: Interested in capital markets, embedded finance, and stablecoins? Everything within the fintech realm, especially within the intersection of AI, is included.

The Smart Systems Stage: It’s all about compute, infrastructure, and energy economics. If you need to build a justification for underwriting AI infrastructure bets, this is the place to start.

The AI Stage and Real World AI Stages: Get market intelligence on where the fastest-growing startups are deploying, not just building.

The Disrupt Stage: This is where operators and CEOs set the narrative your current and future portfolio companies will be measured against.

Is Disrupt 2026 worth it?

Disrupt’s value isn’t found in the size of the crowd, though that certainly helps. It’s in the potential, the expertise, the value of that crowd and the feedback it gives you. With so much overlapping investor attention, the companies worth knowing are getting found fast — regardless of whether you’re in the room. The cost of attending is a ticket and three days of your time. The cost of not attending is finding out about your next portfolio company from someone else’s term sheet.

Disrupt 2026 logistics for investors

We’ve gone in depth about the benefits to you, your portfolio, and the opportunities Disrupt offers you. But there are always matters of hotels, expenses, and travel to sort out. Our Disrupt site has the bulk of these issues covered, but to tackle some common questions and pressing opportunities:

Disrupt passes are discounted until August 22, after which our next pricing tier begins. That means up to $300 discounts on tickets relative to their final prices.

For your hotel needs, we have partnerships with several stellar hotels near Moscone West. You’ll get exclusive discounts, easy access to Disrupt and anything else you might want to explore in San Francisco, and yes, you’ll get points for your accommodations.

If you’re bringing a large cohort or are interested in having a company in your portfolio take part in our Expo Hall to get wider visibility, check out our bulk ticket options here and our Exhibit Table options here.

For all other questions, you can explore our full Disrupt page to find more specific FAQs, and we hope to see you in San Francisco this year!

Over one-third of web pages published after the release of ChatGPT show signs of being written by AI, according to a new study from Pew Research released on Thursday. The report corroborates other studies that detail how much of the web’s newer web pages are now either written by or “substantially edited” by AI, the firm says.

It also arrives shortly after internet infrastructure provider Cloudflare reported that bot web traffic had overtaken human web traffic — a milestone that was reached sooner than the company had estimated.

Pew’s data, however, is focused not on who or what is browsing the web, but on what is being browsed. And apparently, much of it is bots reading web pages written by other bots.

To compile the report, Pew said it used the Common Crawl web archive to collect nearly half a million English-language web pages from the past five or so years, starting a couple of years before ChatGPT’s November 2022 release. Pew then used Open Pangram’s technology to detect how many were likely written or heavily edited by AI.

In a random sample of 10,000 web pages collected in July 2026, around 10% showed “significant signs of AI authorship,” Pew said.

However, Pew pointed out that a random sample like this would inevitably include older web pages — ones published before AI writing tools even existed, and therefore couldn’t have been AI written.

% of web pages showing significant AI editing or authorship.

Image Credits:Pew Research

To get a better sense of how much of the web is now being written by AI, Pew filtered out the older web pages and focused only on those published after the release of ChatGPT.

In this same snapshot with the older pages removed, signs of AI authorship were found in over one-third (35%) of the pages.

Image Credits:Pew Research

Drilling down into domains themselves, Pew found that URLs with a .com domain showed signs of AI authorship at around 10x the rate of a .edu or .gov domain (both of which were around 1% AI authored). In addition, .org domains had only a 4.6% rate of AI authorship.

Of course, this analysis is not perfect. Pangram, like other AI-detection tools, can misclassify pages as AI written when they weren’t. But at scale, the data is likely at least directionally correct.

In addition, Pew found that other supposed tells of AI authorship had also increased over the years, like the use of em dashes, Oxford commas, and phrasing like “it’s not X, it’s Y,” among other things.

Adit Singh was a partner at Foundation Capital when he helped source and co-lead the first funding round into then-obscure chip startup Cerebras.

By the time Cerebras held its blockbuster IPO in May, Foundation Capital held a roughly 7% stake, making it the third-largest shareholder in the company, which now trades at a valuation of nearly $50 billion.

Singh now has a new VC firm to call home: On Thursday, he said he’s now joined the 57-year-old Mayfield as an infrastructure partner.

At Mayfield, Singh will invest in hardware, infrastructure software, cybersecurity, and physical AI. He says it was the firm’s strength in semiconductors that drew him to the role.

“If you look at Mayfield’s semiconductor portfolio, it’s probably the best in the market right now,” Singh told TechCrunch, pointing to Mayfield’s investments in Upscale AI, which was recently valued at $2 billion, and Lumilens, which just raised $700 million at a $5.5 billion valuation.

“I’m a chip designer by training,” Singh said. “My secret sauce is being able to look at any workload and see how it works from the application all the way down to the transistor.”

He credits his background in electrical engineering to helping him recognize the potential in Cerebras 10 years ago. However, Singh didn’t stick around as an investor at Foundation. In 2017, he co-founded Neotribe Ventures, and after spending four years leading the firm, he joined early-stage firm Cota Capital.

Singh says he decided to join Mayfield in large part because the size of seed infrastructure deals has ballooned in recent years. While smaller funds often get priced out of mega-seed deals, Mayfield’s $3 billion in assets under management allows it to write seed checks of up to $20 million.

Mayfield managing partner Navin Chaddha told TechCrunch that despite knowing Singh for 15 years and sitting on three startup boards together — including Upscale AI and unicorn AI startup Velaura AI — it still took some time to convince him to join the firm.

“We have been talking for a long time, and this time stars aligned,” Chaddha said.

Corporate expense management platform Ramp is hot on the heels of Stripe in setting up toll houses for AI inference.

Ramp on Wednesday evening launched its own AI model routing service, dubbed Router, that lets users and companies use and switch between various large language models through an API. The company says it has been using the router it built for its own AI usage needs over the past three years.

The service is only available in the United States for now. It’s free to use for the remainder of 2026 (users will still have to pay for AI model inference costs), and it comes with a $26 credit launch offer. The company did not say how much the service will cost next year.

Router functions similarly to OpenRouter, though the latter offers many more AI model options than Ramp’s current offerings.

Router offers access to models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai. It also provides several “strategies” to help its customers route AI requests to models based on their preferences. For example, one lets users set a preference for model providers’ flex usage tiers, while another lets Router choose which model to route queries to based on up to three user-specified benchmarks. Users can also choose to route only difficult problems to expensive models or test models easily without having to switch.

Users get a dashboard, too, that lets them see token spend, cost, latency, fallback attempts, and other details.

Notably, Router has an opt-out data retention policy: It will record model inputs, outputs, and tool calls for one year by default, though the company says it will remove “personally identifiable information before using that content to improve the product.”

For Ramp, entering the model routing business offers a two-pronged opportunity: It gets to tap the booming AI inference market and offer its existing clients a model routing service that fits in neatly with its existing products, which includes AI token usage monitoring and token spend management.

And, if Router proves as attractive of a model testing arena as OpenRouter has, Ramp may also be able to build long-standing relationships with AI labs and inference providers worldwide. That could help Ramp, which raised $750 million at a $44 billion valuation in June, gain new customers and a new point of entry for selling its expense management products.

In Brief

Two senators have given TikTok a deadline to answer questions about an experiment that withheld a safety feature from 15 million U.S. users, including a teenager who later died by suicide.

Senator Marsha Blackburn (R-TN) and Senator Richard Blumenthal (D-CT) sent a letter to TikTok CEO Shou Chew and Adam Presser, CEO of the company’s U.S. business, calling the company’s decision to run the experiment “depraved.”

Bloomberg reported that TikTok intentionally disabled an algorithmic safeguard for 10% of U.S. users, creating a control group for the experiment. The safeguard was designed to prevent users from being overwhelmed by harmful content, but TikTok wanted to determine whether it made the app less engaging, so it decided to conduct the experiment.

One of the people in the control group was 16-year-old Chase Nasca, whose algorithm fed him thousands of videos about sadness, suicide, and loneliness up until he died by suicide.

The senators are demanding answers to 13 questions regarding the experiment, including names of every employee who knew about the experiment and why TikTok allowed minors to be part of the control group. They’re also looking for a list of every algorithmic experiment in the U.S. when the social network disabled or delayed a safety feature. The senators have given TikTok until September 1 to respond.

TikTok did not immediately respond to TechCrunch’s request for comment.

Meta’s experimental vibe-coding gaming app, Pocket, is now rolling out to everyone in the U.S. The app, which arrived quietly last month in the test market of Brazil, allows people to generate small, interactive games using AI prompts, which are published to a scrollable feed.

Based on Meta’s acqui-hire of the team at the vibe-coded gaming platform Gizmo earlier this year, these interactive experiences and games — or “gizmos” as Meta calls them — respond to touch and the tilt of your phone, play sound effects, and can include clips of your favorite songs.

They can also use photos from your camera roll or access your camera. The resulting games can then be shared on your profile, where others can save them, remix them into other creations, or simply repost them.

Pocket. Image Credits:Meta

The app is the latest example of Meta’s push to make AI-creation tools mainstream, following its earlier efforts that included making AI-generated images with its Meta AI app and creating AI videos with an experimental app called Vibes.

Pocket now joins these and several other stand-alone mobile applications from Meta in recent months. CEO Mark Zuckerberg has credited the increased output to AI-enabled software development, which makes it faster for the company to test and ship new ideas.

“Earlier this year, we shipped Instagram Instants. We also just launched Forum, a stand-alone Groups app, and Seller, a stand-alone Marketplace app. I expect it to become a lot easier to ship new apps,” he told analysts on July’s earnings call. “So we are planning to build out more ideas and use our recommendation systems to scale them.”

In addition to Instants, Forum, and the Seller app, Meta has shipped an experiment involving AI bedtime stories, and this week, a Meta AI app for Mac.

With the launch of Pocket, Meta is shutting down the original app acquired from Atma Sciences.

We’re thrilled to unveil the 2026 Startup Battlefield 200 — TechCrunch’s annual, handpicked list of the 200 most promising early-stage startups from around the world.

These founders rose to the top from thousands of applications spanning every industry and geography. Their reward? A coveted spot to exhibit at TechCrunch Disrupt 2026, taking place October 13–15 in San Francisco’s Moscone West. This is the cohort to watch — and this fall, you’ll get to meet them in person.

From AI to climate, health to fintech, robotics to consumer — these are the companies shaping the future across every major category.

A special thank-you to Google for Startups and SAP, our premier sponsors, for helping early-stage startups build, scale, and succeed.

Only at Disrupt can you:

Watch live demos of never-before-seen tech.

See Battlefield companies pitch in front of top global VCs.

Cheer them on as they compete for $100,000, equity-free, and the iconic Disrupt Cup.

Are you a VC scouting the next breakout investment? A tech enthusiast looking to stay ahead of the curve? An early-stage founder who missed the Startup Battlefield 200 application deadline because you were too busy building a robot in your garage?

Step onto the battlefield. Join us at Disrupt 2026. Register for your ticket before prices go up in September.

Now, the moment you’ve been waiting for…

Presenting the 2026 Startup Battlefield 200

Agtech and food

Canopii

Delight Food Inc.

GreenBox

Muju Earth Technologies Limited

NakedPak

Automotive and transportation

Cautio

Laplace Intelligence

Motion Sync Technologies Inc.

MyMonthlyCar Inc.

Neumo

Urbanchain

Biotech and pharmaceuticals

Alithea Biotechnology

Apertum Neuroscience

BioTrillion

BRAIV

ChemT Biotechnology

Efferon

Infiuss Health

Intero Biosystems

Lume Health Technologies

MyoGene Bio

Nanotrace

ORAIN Biotechnologies

OrganNexus

Sit Sense, Inc.

StuffThatWorks

Sunny

Consumer

Biuty.ai (BelleTech, Inc.)

Chaos Audio LLC

Commissary Club

Craze

Genza Inc.

Hiyd AI

OHLA LLC

Prolo Inc.

Richualist

Swippitt

Tingit

Tressfecta

Yopi SAS

Cybersecurity

Apate.ai

Cerberus

Cude Labs

DetectifAI, Inc.

Hardshell

Indora Labs

MagicSword Inc

PsyberCog Labs, Inc..

Rein Security

Spektrum Labs

Straiker

Trotta

Tzun

E-commerce and retail

Crease Beast

DeepLumen

SDX Holding Limited

SyenApp

Veritas

Wear It AI, Inc.

Edtech

BuddyPro

ConfiTalk Ltd (YNova.ai)

Efiwe AI Nigeria Limited

LicenseCare

MuseCool

ROYO

ScrollEd

Video Pro Learning

WeGlobal

Energy and clean tech

Adden Energy

Armeta Inc.

ChargeMate

Climasel Ltd.

CubeNexus

Danu Water

GinnoLab

Helix Carbon

Licube, Inc.

LNK Energies

Noble Carbon

Petra Power

Enterprise tech

AIEEV Inc

Aiphrodite

Ardook Data

Ariso

Arusto

Bounty

Buzz & Beyond

Collar AI

Cotriply

Dosu

EasyMate AI

Expert Scale Inc

GUDEA

GuideAI

Kilo

Latch AI

Marketrix

Mesh

Paraverse Technology Limited

Pocodot

Trinitite

YOUnifiedAI, Inc.

Entertainment and media

Edit on the Spot

Maka Kids

Press Club

Sound AiSleep

Soundcheck Live

Fintech

Allin AI

Aparta

CedarDB

Climat8

Finaive

FirstGlance

Gemba

HODOR

Prismm

Tabby

TaxFigure

Verapath

Zeeh Global

Government, legal, and regulatory

Bakua AI

Legali AI

Lexsy

Pollsee

wellplayd GmbH

Health and wellness

Avedian

Care Mojo

Circuit Breaker Labs

Forela

Greenairy Pte Ltd

Imagine Devices Inc.

Lifescapes

LumenUs

MindMuscle

Mutanex

Nervotec

Ocean Friends Inc

OmicXHealth

Panacea

Predictheon Medical

Sama Fertility

TimeTeller GmbH

Uhura Bionics

Valene Health

Virtuosis

Voythos

WoafMeow

Human resources and recruiting

AskMAE ai LLC

NTangible Inc.

Reel Analytics

SLI.pro

Your360 AI

Logistics and supply chain

Atreus

Destro AI

Dexa

NuPort Inc

Ridelink Inc

SSINDEX | Stakeholders Sustainability Index

Manufacturing and industrial

Airs ML

ArgosVision

Bioroom Mycelium Materials

CIT Co., Ltd.

Danu Robotics Ltd

IndustrialNext

MUI Robotics Co. Ltd.

Nexterity, Inc.

Niobia AI

Ortex

Planck Lab Inc.

Reclaim Metals

STRUCMEDICA

Marketing and advertising

Floqer

Honeyjar

Intuition Intelligence

LOOQ

onSpark

Prodisphere

Snoika OÜ

Real estate and proptech

Blueprints AI

Propra Technologies Inc.

Super Polymer

Titl

Vitrobot

Watad

Ziffy.ai

Smart cities and infrastructure

ConeLabs

Flox Intelligence

Lumai

Mosqitter Inc

Snowbotix

Space and defense

ItsWare Inc

Lola Vision Systems

Longshot Space

MARASA Corporation

Public Access LLC

Rocket Fast Drones

SatEnlight

Satlyt

Symphony Space Inc

xOrbita Inc.

Every year, Startup Battlefield alumni go on to build category-defining companies — including Dropbox, Cloudflare, and Discord. With over $32 billion raised and 250 exits, the Startup Battlefield track record speaks for itself.

The top 20 finalists from this list will be revealed live on Day 1 of Disrupt on October 13, so be sure to follow along.

If you’ve applied for Startup Battlefield this year

If your startup made the list, keep an eye on your inbox — we’ll be in touch shortly with time-sensitive next steps. If you applied but don’t see your company here, you’ll also receive an official update from us soon. Due to the high volume of submissions, we’re unable to respond to individual application inquiries.

As for the rest of you …

Witness Startup Battlefield 200 live at Disrupt 2026

Get out of your inbox and get ahead of the game at TechCrunch Disrupt 2026 on October 13-15 at Moscone West in San Francisco.

Secure your spot at TechCrunch Disrupt. Prices will go up in September.

Startup Battlefield is proudly presented by Google for Startups and SAP, our premier sponsors.

The team at Patreon, the membership platform for creators, has been building like crazy and is now revealing the result. On Thursday, the company debuted dozens of new features to help creators get discovered, build their community, learn from insights, and more.

Patreon says its roadmap is meant to reflect the core philosophy that today’s internet has failed creators by prioritizing fleeting engagement over genuine community, making it hard for creators to be fairly paid.

“The current version of the web is a failed promise for creators and fandoms. As creators, we can’t even reach our followers on the platforms anymore,” wrote Patreon CEO Jack Conte in an announcement about the new features. “Social media has created mass polarization, addiction, and made it increasingly difficult to build a creative community and business. I refuse to believe that this is the best or only way that the internet can work for creators and fans. I believe it can and should be better,” he added.

Patreon’s changes, however, acknowledge the role that short-form video platforms, like TikTok or Reels, play in helping creators find an audience.

To make that discovery easier, it’s rolling out a feature on iOS called Clips, which will turn creators’ videos into short, shareable clips. These can be downloaded and shared off-platform, as on TikTok, or posted as Patreon’s own short-form post format, known as Quips.

Image Credits:Patreon

Plus, creators will be able to leverage new tools that offer potential fans more previews of their work.

The tools, which are still in early testing, include suggested video previews, which will pull out the most engaging parts of their paid videos, and another that lets creators quote snippets of their longer written posts into Quips. The video previews can be shown to fans who are weighing whether or not to pay to unlock the content or subscribe to a membership, while the text-based quote Quips can be shared as previews to the Home feed, offering creators more exposure.

Image Credits:Patreon

Native video and Quips are also now rolling out to creators of adult (18+) content.

Another significant change in today’s update involves Patreon’s discovery algorithm.

Before, the company’s recommendation systems focused on finding creators similar to those the user already liked, but it tended to favor larger creators. Now, discovery will focus more on comparing a particular post to other posts by topic, craft, style, theme, and other factors. Patreon says this will help to highlight smaller creators, too.

Image Credits:Patreon

The company is also rolling out several new features designed to deepen fandoms and help make Patreon more of a destination.

The highlight here is a new communities feature called Niches, which allows creators to participate in dedicated, topic-based communities built around specific interests, subcultures, or fandoms. This could help fans discover more of what they like, without needing to dig around the site to find all the different creators focused on the topic at hand.

Image Credits:Patreon

Patreon will also roll out other tools, like support for Live Q&As on creator livestreams, and tools that let fans make their own profiles. (Those profiles and the communities are now in early testing, while the Live Q&As are now arriving.)

Image Credits:Patreon

Image Credits:Patreon

Another set of tools will help creators track their audience and earnings, which can be key to developing a healthy fandom.

Patreon is launching more granular earnings, membership, and growth analytics, with an earnings tab that breaks down earnings by tier and billing cadence. The platform is also adding membership insights, like member counts that show free trials and members in payment retry, plus filters to break down memberships by trial, gift, active, and retrying status.

Image Credits:Patreon

Membership Insights. Image Credits:Patreon

A revamped version of the Payouts tab will also show creators a complete record of the earnings moving in and out of their account.

Soon, Patreon will also add creator milestones, or key growth moments, to celebrate special moments in terms of the growth of their community.

As part of its longer-term roadmap, Patreon plans to develop security features like anti-AI scraping tools, real-time spam detection, expanded auto-moderation, additional verification measures, and more.

The full list of improvements includes 30 new or revamped features, some of which are still in early testing. Over the next few months, creators should expect to see many more changes arrive, Patreon said.

Fusion startup Inertia Enterprises said it has found a way to cut the time it takes to make its fuel pellets from several days to just minutes.

By slashing fuel filling time, Inertia says it has knocked down one of the 10 barriers it must overcome to deliver the first phase of its commercial power plant ambitions. The startup gave TechCrunch an exclusive first look at the process.

The advance was not guaranteed: Investors gave Inertia $450 million on the premise that it could commercialize technology developed at the National Ignition Facility (NIF), which is an elaborate science experiment that requires painstaking care and feeding to operate. At the NIF, making fuel pellets can take a week or more and cost a small fortune. That’s not a great recipe for a profitable operation.

“But when you really double-click on it, you’re like, wait a minute, they’re only making a handful of them a year,” Jeff Lawson, co-founder and CEO of Inertia, told TechCrunch. “I put on my commercial hat and was like, wait a minute, I know the word for this: Prototypes.”

At Inertia, the team set about turning those prototypes into something mass-manufacturable. “That’s why we’re hiring people from the likes of Apple,” he said. “There’s a bunch of industrial engineers who are like, ‘Alright, I guess I have to go figure out how to make that a billion times over in a factory.’”

The NIF’s fuel pellet is far from an iPhone, though. The outside is a spherical diamond shell, and just inside, there’s a thin layer of frozen deuterium and tritium, the isotopes of hydrogen that can fuel a fusion reactor. Inside the crystalline layer lies a mix of gaseous deuterium and tritium. Each solid layer must be as close to perfectly spherical as possible.

Those fuel pellets are then wrapped in gold casings known as hohlraums, which convert laser energy into X-rays that compress the fuel pellet inside. If everything goes to plan, that compression causes atoms to fuse and release energy. But even small imperfections in the spherical shape can disrupt the ignition process, preventing a fusion reaction from reaching its full potential.

Inertia had to condense the process to the point where it makes commercial sense without straying too far from the physics discovered at the NIF.

“What happens if you try to do it faster?” Lawson said. The team had a head start: Annie Kritcher, co-founder and chief scientist at Inertia, designed the first fusion experiment at the NIF that released more power than it consumed.

After several rounds of development, the startup was able to grow the crystals in about 30 minutes, something that could take up to a week at the NIF. Altogether, a single fuel pellet can be made in about two to three hours, and the process can be ramped to industrial scale. Inertia developed the process with help from the NIF at the Lawrence Livermore National Lab, with which the startup has formed a public-private partnership.

Inertia has an advantage that the NIF does not, though. Because the startup is planning to use a laser that is four times more powerful than the one currently at the NIF, it can tolerate more imperfections in the fuel pellet. This also helped speed the manufacturing process.

“We actually have a lot of margin,” Lawson said. “That’s our strategy, to oversize our driver, our laser, to give us lots of margin to go play with in every other part of the system.”

Reducing fuel filling time has the knock-on effect of reducing the amount of tritium Inertia needs to hold at any given time. Tritium is radioactive, and handling it requires care. It’s also extremely expensive at the moment, about $30,000 per gram, and only about 25 kilograms are stockpiled globally, according to the journal Science.

Inertia, like many fusion startups, plans to make its own tritium with help from the fusion reactions, but it still needs some inventory to get started. Plus, reducing manufacturing time helps keep inventory small. Inertia expects its full-scale commercial power plant will use 10 fuel pellets every second.

“By reducing the latency of this step, you’ve made your facility smaller; you’ve made the whole thing faster, the whole thing more efficient,” Lawson said.