Techcrunch

Apple’s services business remains at record levels, but it’s facing growing pressure. The central part of that business, Apple’s App Store, has faced regulatory pressure and a slump in mobile gaming, which have hurt services growth.

However, on Monday’s earnings call, Apple signaled that it may look to boost services revenue through third-party apps. What does that mean, exactly? Unfortunately, it’s not entirely clear.

When discussing how AI could potentially “contribute to revenue growth over time,” Kevan Parekh, Apple’s chief financial officer, said: “On the product side, you know, we feel AI offers a compelling reason for customers to upgrade and switch in order to get the latest capabilities, and we also feel there’s opportunities across third-party apps and partnerships as well as first-party opportunities.”

Parekh did not elaborate on what sorts of opportunities might be available via third-party apps. However, Reuters reported Monday that investors are wondering whether Apple could profit by distributing third-party AI agents through the App Store and taking a cut of the revenue they generate.

Regulatory and legal changes have cut into App Store commission revenue in some markets, even as Apple’s services business as a whole continues to grow. Third-party estimates in August, for example, suggested the company had seen an 18% drop in U.S. commission revenue, with declines also in Brazil and Japan after those countries introduced new regulations. Apple has traditionally controlled payments within its app marketplace tightly, taking a cut of all App Store transactions. But a lawsuit by Epic Games ultimately forced the company to allow apps to redirect users to outside payment systems. Other legal challenges throughout the world have further hampered the business.

During the earnings call, Parekh referenced Apple’s App Store woes and returned to the idea that the company’s service business could expand into new territory.

“We’ve also seen some factors that impacted the performance of the App Store, and during the September quarter, the performance on the App Store was similar to the prior year, and we had a record number of paid subscriptions with double-digit growth for third-party apps on the App Store,” Parekh said. “At the same time, we did see some headwinds in mobile gaming that we’ve talked about before, and expect to see those continue into the summer quarter.”

Parekh added: “As we move ahead, we expect services growth to to be between 8 to 9% [in the December quarter]. This outlook contemplates our best estimates on sequential headwinds, primarily from third-party licensing and the additional half percentage point of foreign exchange I just mentioned.”

TechCrunch has reached out to Apple for more information and will update this story if it responds.

View Bio

Latest in Hardware

Y Combinator CEO Garry Tan has, in the past, jokingly admitted that he’s suffered from “cyber psychosis” and sometimes barely sleeps because he’s so excited to be working with his AI agents. On a new episode of Walter Isaacson’s podcast The Builders, set to publish on Tuesday, Tan put a price tag on his passion.

“I spend about maybe $100[,000] to $200,000 a year on tokens,” he told Isaacson. He uses agents for countless work projects and coding tasks and has freely shared his set of 23 Claude Code skills, called gstack, on GitHub — an effort to help teach others what agents can do, he said.

On the one hand, he uses his agent the way many people use AI: to help him research Isaacson’s podcast and prep for the kinds of questions he might be asked. On the other, his agent also helps him run his life.

“It’s read all of my email and all of my Dropbox. It’s read, like, all of my Apple Notes, even my therapy notes going back 20 years,” he said.

He describes using AI agents as getting help with “almost anything that you do in your day-to-day life, and for it to do it somewhat more autonomously. Think of it as a Jiminy Cricket on your shoulder at all times. Like, almost anything you want to do, it knows you, and it’s there to help. And it will actually sort of run ahead and say, ‘What about this?’ Or it’ll help you see around corners.”

As an early OpenClaw adopter who is now a power user, Tan says one thing he’s learned is not to treat AI agents like traditional software. Some people are so worried about “repeatability and safety that they actually trade off all of the beautiful parts of working with an AI agent, which basically can be so similar to a human,” he said.

For instance, Y Combinator is also using an agent that is “rebuilding the way we operate at YC,” he said. “We now have an internal AI agent, for instance, named Stray Light that kind of has a personality, and then it has a company brain. Like it actually has the total context of every founder we’ve ever backed, anyone who’s ever emailed us, like all the applications.”

Tan, who is, naturally, an avid lover of technology (as anyone running Silicon Valley’s famed startup accelerator would be), believes personal AI agents will eventually become mainstream, if one major problem is solved: Costs have got to come down.

“It’s too expensive right now,” he said, adding that he knows the amount of money he spends on AI agents is “a crazy privilege that not everyone can do.”

But freebie agents controlled by Big Tech maybe aren’t the answer, either. When discussing the future he’d like tech to build, and how Meta as a social media company has impacted us, he said, “This is a huge company that controls the algorithm, and we all wish that they didn’t at some level,” he said. “So you could imagine a personal AI that you control, that you program.”

View Bio

Hinge is testing a cheaper subscription plan as its parent company Match Group looks for new ways to get users to pay for dating apps at a time when people are increasingly tired of subscriptions and, in some cases, dating apps altogether.

During its third-quarter 2026 earnings call on Monday, Match Group said the new Hinge tier will include features such as additional filters and custom location boundaries, giving users more control over who they see and where they meet people.

“It’s designed for women in particular, but also for price-sensitive users,” said Steve Bailey, chief financial officer of Match Group. “Remember that Hinge is one of the highest-priced products we have. The revenue per payer, for example, at Hinge is almost twice that of Tinder, around $30. It also has one of the lower payer penetrations. Its payer penetration is only slightly more than half that of Tinder.”

Match Group didn’t share how much the new subscription will cost. TechCrunch has reached out to the company for more details.

The move comes as dating apps continue to struggle to convince people to pay for services they’re growing tired of using. Between subscription fatigue, endless swiping, and frustration with the overall dating experience, users are increasingly questioning whether premium dating features are worth the money.

Interestingly, Hinge is making this move despite being one of Match Group’s strongest performers. The app had a much better quarter than Tinder, with global monthly active users growing 12% year over year, direct revenue increasing 19%, and paying users rising 23% to 2.3 million. The app generated $221 million in direct revenue and had 2.3 million paid subscribers as of the end of the quarter. Direct revenue is money earned from subscriptions and in-app purchases.

By comparison, Tinder brought in $480 million in direct revenue but experienced a slight 2% decline year-over-year, although it has 8.9 million paid users, a 4% decline from a year earlier.

The lower-priced subscription could help Hinge build on that momentum by converting more users into paying customers without requiring them to spend as much upfront.

The move also marks a shift in Hinge’s subscription strategy. In 2023, the company experimented with a more expensive tier aimed at users willing to pay extra for features designed to help them get more out of the app. Now, it’s testing whether a more affordable option could appeal to a broader audience.

Still, cheaper subscriptions alone are unlikely to solve the bigger problem facing dating apps: user fatigue. Match Group says it’s seeing some signs of improvement, partly thanks to efforts to make its platforms feel less like traditional dating apps.

Recent initiatives include Group Hangouts and Events, which offer users alternative ways to connect beyond the usual one-on-one matching experience. Match Group plans to ramp up those efforts in the fourth quarter. 

Another initiative on the horizon is real-time adaptive recommendations, which the company plans to begin testing in the fourth quarter. Announced during its CEO Connection presentation in September, the feature will allow Tinder to respond to user actions within seconds rather than hours, with the goal of making recommendations more relevant and personalized.

View Bio

Apple Chief Operating Officer Sabih Khan acknowledged the shipping delays plaguing the new iPhone Duo during the company’s fiscal fourth-quarter earnings call on Monday, but offered little clarity on when customers can expect relief.

The delays were just one point of concern during an earnings call filled with questions about the rising costs of memory chips, driven by soaring demand from the AI industry.

“iPhone Duo is constrained as of today. The ramp is going well, and we’re working hard to get this revolutionary new product into the hands of our customers as quickly as possible. But at this time, it’s really hard to predict when supply and demand will be in balance,” Khan said.

Customers who didn’t wait in line at an Apple Store or immediately jump on preorders are now seeing ship times stretch into next year. Meanwhile, Verizon has been apologizing to its customers for the delays by offering gift cards.

In an October report from Financial Times, Apple described demand for the Duo as “nothing short of extraordinary,” after the device sold out on the company’s website within hours of going on sale, pushing delivery dates into next year.

The prolonged shipping delays are notable given Apple’s reputation for its highly efficient supply chain. But the company is hardly alone in facing supply pressures. Soaring demand for memory chips from the AI industry has driven up component costs for hardware makers across the board, threatening to push device prices even higher.

Apple, too, has raised prices on a number of products, Khan noted on the call, adding that those decisions were made “reluctantly.”

In another sign of the company’s efforts to meet demand for the Duo, Apple said it’s “deepening” its partnership with Perryman, a titanium producer for supplying materials for the new foldable device.

“We’ve been thrilled with the early response from reviewers and customers, and we are working hard to get iPhone Duo into customers’ hands as quickly as we can,” Apple CEO John Ternus said during his opening remarks.

View Bio

Whether developed by OpenAI or Anthropic or built on open source alternatives, current AI models share a critical limitation: They don’t learn from their mistakes without further training.

Businesses adapting these models for specialized fields, such as law and finance, typically get around this by fine-tuning general models. However, that often requires hiring internal research teams or relying on expensive consultants, commonly known as forward-deployed engineers (FDEs).

Trajectory, a startup founded about a year ago by three Stanford classmates with experience at Google DeepMind and Apple, wants to automate that process, allowing AI models to learn from feedback and continuously improve at specialized tasks.

The 18-person company just announced that it raised $45 million in a Series A round at a $300 million valuation, led by Sequoia, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures.

Trajectory co-founder Ronak Malde compared foundational models to a very smart PhD student who’s always starting a new job. “But what the economy values is someone who worked on the job for 30 years,” he told TechCrunch.

The startup aims to solve this problem through an approach known as continual learning. Businesses using Trajectory can start with an open source AI model and connect the platform to their applications, allowing it to collect feedback on how the model performs. Trajectory then uses that information to retrain the model, helping it improve at specialized tasks over time.

With Trajectory, lawyers, doctors, and other specialists can improve the models near automatically themselves, Malde said. That capability could make deploying specialized AI faster and cheaper.

Companies like Harvey and Rogo are already using Trajectory’s platform to post-train open source models, helping them perform more like experienced lawyers or financial analysts by learning from previous tasks, such as reviewing a particular type of legal contract or valuing a company using a specific methodology. Trajectory’s other customers include Decagon, Mercor, and Nvidia. The startup is now making its post-training platform publicly available.

The people using Trajectory decide “what good looks like for their organization and can turn that into improvements for their model,” Malde said.

Trajectory’s approach to continual learning shares some similarities with recursive self-improvement (RSI), the idea that AI systems can improve their own capabilities. But co-founder Michael Elabd argues that keeping humans involved in the training process makes Trajectory’s approach safer, since autonomous self-improvement could lead to unintended or even dangerous outcomes.

Trajectory isn’t the only startup trying to automate post-training. It considers Thinking Machines Lab, founded by former OpenAI CTO Mira Murati and the company behind the fine-tuning platform Tinker, a direct competitor, along with Applied Compute, which was valued at $3.25 billion last month, and Prime Intellect, which helps companies build and train their own AI agents.

Malde believes the startup’s biggest advantage is its team. He was among the handful of researchers Google DeepMind acqui-hired from Windsurf. Elabd also worked at DeepMind, while Trajectory’s third co-founder, Arjun Karanam, previously worked on generative foundation models at Apple. The startup has also recruited researchers and product designers from OpenAI, Stripe, and Netflix.

Why does design matter so much to a company focused on AI training? Karanam says the startup has developed an interface that makes post-training more accessible to its nontechnical customers.

View Bio

Apple’s stockpiling of inventory, particularly computer memory, continued in its fiscal fourth quarter, according to earnings reported Monday. The company reported that its inventory increased significantly in the quarter, growing from $11.09 billion in Q3 to $14.54 billion in Q4, a jump of 31.1%. Year-over-year, inventory grew 154.3% from $5.72 billion in the fourth quarter of 2025.

Apple’s forecast of a 45% to 46% gross margin in the December quarter was largely attributed to the impact of higher memory costs. The company reported a gross margin of 47.9% for the September quarter.

“What we’re seeing with memory pricing right now is unprecedented,” said Apple CEO John Ternus, speaking to analysts on the call. “As we said earlier in the September quarter, we did see higher memory costs as compared to the June quarter, which were partly offset by the benefit of carrying inventory and lower costs on certain non-memory components.”

He said that the company expected “even higher memory costs” in the December quarter, which would be partially offset by price increases implemented earlier this year on some devices, including Mac and iPad at the end of June and iPhones in September.

Ternus also said that beyond December, Apple expects memory costs “continuing to increase.”

“As we said before, we do believe we’ll see a decreasing benefit from the carry-in inventory on memory over time,” he noted, adding that the company was increasing its purchases of this key component. “As we work to improve supply, our manufacturing purchase obligations have increased by 33 billion dollars from June to the September quarter, which you’ll see in our annual filing when it comes out tomorrow.”

Apple also brought COO Sabih Khan, who oversees the company’s supply chain and manufacturing, onto the call to address questions about memory. He reiterated the increased purchase obligations but did not rule out future price increases on Apple devices.

The company also blamed supply constraints for weak iPad sales in the quarter. The iPad was one of the only product segments that didn’t grow, even as iPhone and Mac sales beat estimates. iPad revenue was $6.2 billion, below analysts’ expectations of $6.7 billion. The figure was also down 10% from the prior year.

View Bio

While much attention has focused on the updated financials Anthropic revealed in its revised S-1A on Monday, (revenue is soaring, but so are expenses), the AI lab dropped another little bombshell in the fine print.

The filing seems to warn that the company is on the cusp of creating a model capable of recursive self-improvement (RSI).

RSI is the point at which AI grows so good at training itself that it spawns a dramatic acceleration of AI capabilities. To the AI doomers, RSI represents the cataclysm from which there is no return, in which AI builds itself into something powerful enough that it slips beyond human control and may be indifferent to the damage it causes while achieving its goals.

To AI enthusiasts, RSI, or some version of AI reliably using itself to improve, represents the technology’s promise, where it becomes a tool of unimaginable productivity, scientific discovery, and more.

Anthropic seems to believe that RSI is both a promise and a peril. It says in the S-1 that it uses AI in its R&D, but it cautions: “Using AI to help build more capable AI could also pose a considerable safety risk, particularly if it sharply accelerates the pace of progress. This is one of the areas covered by our Responsible Scaling Policy, which sets out our plans to monitor and mitigate the risks as we approach what has been termed ‘recursive self-improvement.’”

When the term RSI became popular earlier this year, many believed it was a buzzword or, at best, a goal too hard to achieve, define, or prove.

But on Monday, Anthropic added new statements to its Risk Factors section, saying 1) it has already used AI to develop new models whose leaps in capability it didn’t recognize at the time, and 2) it may soon meet its own internal definition of RSI (bold emphasis ours, below):

In addition, our increasing use of AI systems to accelerate our own research and development may cause the capabilities of our models to advance more rapidly than our ability to develop, validate, and implement corresponding safety, alignment, and security measures. As a result, we have in the past not identified, and may in the future not identify, a significant acceleration in our rate of development until after it has occurred.

For example, our internal estimates and our expectations regarding future models, rather than our current models, indicate that we may imminently meet our Responsible Scaling Policy’s definition of “dramatically accelerated progress attributable to AI.” We have not designated any of our current models as exceeding this threshold, and we continue to assess the underlying data. 

However, Anthropic also said in the filing that none of its models have crossed this RSI threshold yet and that it is promising to keep humans in the R&D process for the foreseeable future. “We do not anticipate being able to safely use Claude to fully autonomously design and build successive generations of frontier models without meaningful human supervision or understanding, if and when that becomes technically possible. Accordingly, we are considering updates to our Responsible Scaling Policy to keep humans in control of frontier model development at Anthropic.”

Skeptics might point out that an S-1 functions as both disclosure and marketing, and that claims of being close to a model so powerful that it can build itself could fall into the latter category.

Nevertheless, there’s no question that Claude’s capabilities, especially in terms of coding, math, and cybersecurity, have grow in leaps and bounds this year. If Anthropic thinks the possibility of RSI is close, maybe we should believe it and act — perhaps govern — accordingly.

View Bio

The German government is warning politicians and public figures that they were targeted by government-backed spyware, days after Apple began sending out notifications warning victims that their iPhones may have been hacked.

Germany’s federal cybersecurity office, known as the BSI, said in a note over the weekend that it had become aware that members of the Bundestag, Germany’s parliament, and other unnamed notable individuals had received warnings from Apple that they may have been targeted by cyberattacks.

Per the Financial Times, German government spokesperson Stefan Kornelius said the government in Berlin was “taking the incident and the warning very seriously,” but he did not say who in the federal government had been targeted, including whether German chancellor Friedrich Merz was among them.

Apple periodically sends out threat notifications to people it thinks may have been targeted by spyware, which can steal messages, location data, and other sensitive information from their phones. This spyware often relies on zero-click exploits, which can hack individuals almost invisibly as they don’t require a victim to click on a link or open a file.

Apple does not attribute attacks to specific governments or say whether someone’s phone has actually been compromised. For that reason, the BSI said that anyone who receives an Apple threat notification “should therefore assume that their device may have been compromised.”

Others outside of Germany also received the recent threat alert from Apple. Radwan Mortada, a Beirut-based filmmaker, said in a post on X that he received a new threat notification warning that Apple had detected “another attack against your device,” after an earlier spyware attack targeted him in 2021.

It’s unclear who is behind this latest round of spyware attacks, or which government may have orchestrated the targeting. Per the notice received by Mortada, Apple said its latest batch of threat notifications went out to users in 99 countries.

The episode highlights again the murky business behind a growing number of attacks against members of civil society. Governments rarely build these tools themselves. Many buy them from private vendors, often called “mercenary spyware” makers, the best known being Israel’s NSO Group, which sells the controversial Pegasus spyware. Germany’s own central federal law agency, the BKA, reportedly bought access to Pegasus in 2019.

View Bio

Squarespace, the web company that helps customers build and host their own websites, is laying off about one-fifth of its workforce, in a move that will affect hundreds of jobs.

Jeff Seedman, a public relations specialist representing Squarespace, confirmed the layoffs to TechCrunch. The Verge initially reported the layoffs.

“Squarespace is restructuring its organization and reducing its global workforce by approximately 19%, representing around 300 employees,” said Seedman. “The changes are intended to simplify how we operate, sharpen our priorities and enable the company to move faster as we position Squarespace for its next phase of growth.”

Website building has changed immensely over the last few years as a result of the generative AI boom. “Vibe-coding” website builders, which only require simple prompts from users to spin up functional front-end designs, have largely upended the industry, making it possible for users to create sites with significantly less effort.

These kinds of sites have become an increasingly central part of the website-building business. Last year, web host rival Wix bought Base44 for $80 million. Similar offerings, like Lovable, have generated significant investor interest; Lovable raised a $400 million Series C round in August that valued the company at $13.3 billion. Google also recently launched Stitch, a feature that allows users to swiftly generate user-interface designs.

Squarespace has also begun a big push into AI. In 2024, the company launched an AI tool called Blueprint. The company also recently announced Forma, an upcoming AI native platform, whose waitlist users can join.

Squarespace was founded in 2004 and went public in 2021 at a $10 billion valuation. However, the company struggled in the ensuing years to generate enthusiastic public market interest. Squarespace ultimately decided to go private again by way of a $7 billion private equity deal in 2024.

An earlier version of this report incorrectly stated that Squarespace had not yet launched AI tools; the story has been updated.

View Bio

Latest in AI

Latest in Apps

Latest in Enterprise

Readwise, the maker of a read-it-later app and digital reading companion that keeps track of your highlights across services, is now taking on Amazon’s e-book empire with its latest app, Readwise Books. The app is an e-book store and helps you track what you’ve read, similar to Amazon’s Goodreads.

The bootstrapped startup described the new app as “a cross between Kindle and the Goodreads apps,” as it offers a single place to buy, read, and track books. However, Readwise Books aims to offer more detailed reading stats than the dates you read a book and whether you finished it. It lets readers visualize richer reading data like words read, pace, depth, immersion, and sessions.

Meanwhile, its new store offers “millions” of e-books from the big five publishers and others, initially in the U.S., with more countries to come. Print books (fulfilled by Bookshop.org’s distributor) and some audiobooks are available as well, the company notes.

The launch comes at a time when there’s been an explosion of new reading apps and book trackers, all hoping to challenge Goodreads by introducing new features like book clubs, widgets, improved discovery features, goal-setting, better lists and organizational tools, habit-tracking, and more. However, few indie developers have gone so far as to launch an alternative to Amazon’s Kindle e-books store, which remains the industry leader, with an 83% share of the market when Kindle+ (formerly Kindle Unlimited) is included in the stats.

One of Kindle’s bigger advantages is that it offers an end-to-end solution for everything related to reading e-books, from the hardware devices to the book-tracking log. Readwise isn’t selling its own e-readers, but it is attempting to create its own community around reading, complete with tools for reading purchased and uploaded e-books, tracking them (like on Goodreads), sharing your reading progress with friends, and more.

It also offers a number of ways to track books outside its ecosystem. This includes a way to track reading sessions by using Kindle’s browser-based Cloud Reader and a feature that lets you snap photos of physical books to mark your progress. Plus, you can upload any DRM-free EPUBs (a common e-book file format without copy protection). The app can read your e-books aloud via text-to-speech.

Readwise believes its more in-depth metrics could be a big selling point, too, likening them to a “Strava for reading” — a reference to the metrics-heavy exercise app.

“The desire to read books, like exercising and eating well, is laden with guilt,” the company explained in a blog post penned by co-founders Daniel Doyon and Tristan Homsi. “Many people feel like they should be reading more, but feeling bad about it tends to change nothing. The session tracking taps into a different psychological mechanism. It makes the reading you’re doing legible and, as the adage goes, that which is measured is improved.”

The company adds that the most consistent feedback from users is that the app has led them to read more without willpower tricks like goal-setting, streaks, or scheduling.

With publishers’ permission, the app will support AI-powered features as well, like session recaps, definitions and term lookups, word and phrase translations, and a location-aware ability to ask questions about the part you’re reading. (So far, only HarperCollins has opted in, though.)

Unlike Goodreads’ locked-down ecosystem, Readwise is embracing open technology. This includes a public API and an MCP server. The latter supports AI agents that could be used to research book recommendations on your behalf, use your Readwise data to create a books page on your personal website, or even wire up your hackable e-reader to log your sessions automatically, depending on how involved your AI use cases may be.

The new Readwise Books app will remain free, with some AI features later going behind a paywall. Its prior apps for reading and tracking highlights are either $7.99 per month each or $9.99 per month for both.

The Books app is currently in beta and requires an invite, unless you’re already a Readwise subscriber.

View Bio

SpaceX is looking to buy Boost Mobile from EchoStar, according to a report from The Washington Post. If the deal goes through, it would help the Elon Musk-owned company build a hybrid cellular-satellite network that could help catapult it into fourth place among the top U.S. wireless providers.

The space-launch company has been dabbling in cellular connectivity, rolling out satellite access to T-Mobile’s subscribers in 2024. Last year, SpaceX bought $17 billion worth of wireless spectrum from EchoStar for cellular services, which could help the company’s Starlink service reach inside buildings, for example, something it currently struggles with.

“This is the last critical piece of the spectrum puzzle needed for SpaceX to provide complete phone coverage in America,” Musk said on X at the time of the spectrum acquisition. Yet, if Musk is still looking to buy access to terrestrial networks, it apparently wasn’t the last piece.

Elon Musk, SpaceX’s CEO, reportedly spoke directly with EchoStar chairman Charlie Ergen in October about acquiring Boost Mobile, the mobile virtual network operator. EchoStar has about 7.5 million customers between Boost Mobile and Gen Mobile, its other mobile service offering.

Musk has also approached AT&T, T-Mobile, and Verizon about buying capacity on their networks. All three rejected Musk, and they may still have a say in a SpaceX acquisition of Boost Mobile. Boost doesn’t operate its own towers, instead leasing capacity from AT&T and T-Mobile, which might need to approve such a deal. Sources speaking to The Post said that network leases are usually nontransferable.

The big three providers each have over 100 million subscribers, dwarfing Boost Mobile. But while the acquisition would give SpaceX an immediate bump in subscribers, the true value of the deal would be giving the company access to terrestrial cellular networks. 

Gaining access to towers leased to Boost Mobile would expand coverage significantly — Boost currently covers most of the 48 contiguous states in the U.S. — allowing Starlink to cover devices when they’re outdoors and relying on cell towers to handle devices when Starlink signals can’t reach them.

View Bio

Here’s a question straight out of sci-fi: What if AI could predict what you were going to do before you decided to, even in a completely new situation?

Israeli startup 42AI says it is developing a “behavioral world model” that can do just that, and it has just come out of stealth with $62.5 million in seed funding. The company is initially running its predictive system at a handful of financial brokerages, where it says the model can help predict how individual traders will respond to volatile conditions.

“Behavior is not deterministic; it’s constantly probabilistic,” 42AI’s CEO and co-founder Amit Gal-Or told TechCrunch when I asked him to justify his company’s tall claim. He argues that people have underlying psychological tendencies that influence their decisions even when their circumstances change, and that’s what the startup is trying to measure and convert into data that the model can understand.

“A week from now, you might be a different person, but your tendencies will always play out in certain areas to a probability or distribution,” his co-founder and CTO Ehud Ezra added. And that, the company argues, is what the model does: show the probability of a person making a certain decision.

42AI says the model tries to figure out people’s internal motivations instead of extrapolating patterns based on their previous actions. To do that, 42AI examines how people react to seemingly unrelated events. For an investor or trader, for example, such events might include completing KYC processes, onboarding, support calls, money transfers, or trading stocks.

“When the person is talking to support or sales in the initial process […] we’re looking at the sentiment, we’re looking at the voice, we’re looking at a lot of interactions that they’re doing, how many times do they call, and so forth,” added Ezra. “And all of those data points are going into the model. Later on, you have the interactions: how you buy and sell, and when, and we’re looking at the market at that stage.”

“So we’re trying to look at how they’re looking at the world. In a world model, every individual is the world,” he added.

The company then converts those observations into mathematical representations of traits, which are then fed into a transformer-based world model that lays out probabilities for possible decisions under different circumstances. (The company uses a separate language model to generate explanations for its predictions.)

Ezra said the model is trained to preserve a person’s underlying tendencies during simulations. So if someone is usually risk-averse, but the model suddenly treats them as a risk-taker, “the target is off,” he said, explaining how the model’s training rewards are structured.

42AI spent three years developing this approach, combining psychological research with neuroscience and genetics, as well as advice from experts in psychology and PTSD, and Dan Shechtman, who was awarded the 2011 Nobel Prize in Chemistry.

Gal-Or said the team first studied thousands of people via psych assessments and behavioral scenarios, before adding government and biobank datasets covering more than 2 million people. Its customer datasets have since expanded to cover almost 30 million people, he said.

In live deployments, the model analyzes customers’ existing records, including transactions and support calls. Gal-Or pointed out the system doesn’t need personally identifiable information, arguing that “the model never knows who the person is.” However, it seems the company still creates individual behavioral profiles from linked customer activity.

In predicting market trades, 42AI claims its model achieves 98% accuracy, but Gal-Or clarified that this is more a measure of whether trading flows fall within a predicted range, not if each trade was predicted correctly.

He also said the model consistently achieves an F1 score of 0.8 to 0.85 for predicting trading direction and volume. (F1 score is a metric in machine learning that measures how well a classification model can balance correct predictions against false alarms and missed events.)

And it’s not like the model can work everywhere, given it’s still a machine in a box. “There’s a natural ceiling because you really don’t have access to everything that a person experiences,” Gal-Or said. Ezra explained that the system’s accuracy also suffers when it misses important situational context, like if someone chooses to change their trading choices based on advice from a friend or a colleague.

Gal-Or said 42AI is working with 10 large brokerages, who are using it to manage liquidity, margin requirements, and hedging exposure. The company is onboarding more customers, he said, and is generating more than “seven figures” in revenue, though he wouldn’t specify a number.

Customers pay the startup a flat fee, or a percentage of the value the platform helps them gain. He claimed one brokerage generated approximately $30 million in additional value through dynamic hedging.

The $62.5 million seed funding came from the founders and Infinity Group over three years, and wasn’t entirely raised from external investors, Gal-Or said. Infinity was founded by his father, Amir Gal-Or, who is also 42AI’s chairman and a co-founder. 42AI now plans to raise more money.

Beyond finance, 42AI is now exploring applications for its model in gaming, fraud prevention, and cybersecurity, and one day wants to use it to help robots and AI agents understand people. But that may be a ways off yet: Ezra acknowledged that the model needs to be trained on rules and context for each new domain.

And like with all AI tools, there’s a potentially troubling side to 42AI’s technology: If a brokerage can anticipate what its customers may do, what’s to stop someone from using that knowledge to their own benefit? Ezra admitted that this was “a valid concern” when I raised it, and the founders pointed to regulations, monitoring, and an internal ethics committee, though they didn’t explain what, if anything, would stop their customers from using the technology to gain an advantage over the very people it’s profiling.

Amit Gal-Or, however, was confident that the technology could help address such risks. “We are very proud that actually we are part of the solution; we are the solution to this problem,” he said.

Over Halloween weekend, some Grok Bot users awoke to something especially spooky: Their weekly usage limits were unexpectedly maxed out, even if they hadn’t been using SpaceX’s agentic AI overnight. Some users reported on X that their usage jumped from as low as 50% to 100%.

SpaceX CEO Elon Musk confirmed in an X post on Sunday that this was not a glitch.

“The cost of serving Grok Bot was unsustainably high relative to revenue,” he wrote in a reply to a question from a user. He has not addressed the issue on his main feed but has continued boasting that Grok Bot can help you buy gifts for your family.

Grok Bot has been positioned as a success story for the newly public SpaceX, which said in a filing last month that AI now makes up 52% of company income. Its renamed “Super Intelligence” division actually turned a profit of $431 million, which is a rarity in the AI industry. Its profit would’ve been higher if not for $1.2 billion of one-time costs related to its acquisition of Cursor. Musk recently referred to Grok Bot as having “the fastest growth of a product that I’ve ever seen in my life.”

Musk has not yet shared how subscribers’ usage limits will be impacted moving forward.

Topics

U.S. data centers use 17 to 19 billion gallons of water to cool their hot-running chips each year — and that number is expected to triple in the next five years, according to The Water Foundation.

That’s an expensive and politically tricky proposition for investors throwing billions of dollars at the challenge of scaling up AI. One way to use less water is to do a better job monitoring the cooling fluids, a task that currently involves sending samples away to a laboratory for tests in a multi-day process.

Omen AI was founded in 2024 to solve that problem. The company is building a real-time sensor to determine whether valves were failing, or if bacteria growth in the system threatened to choke off water flow. After raising a $31 million Series A in June, Omen is now acquiring competitor Aquatic Labs in a $16 million cash-and-stock deal.

The deal came together after Omen founder Zach Laberge was introduced to Aquatic Labs’ CEO Allan Adams by an investor in the filtration space. Both startups were building real-time water sensors with complementary technology: Omen’s sensors looked for elements in the water, while the Aquatic Labs sensor tests its alkalinity.

Putting those two technologies in the same sensor creates a “much richer picture of the water and the fluid,” Adams told TechCrunch, while uniting their vision of automating what had previously been a slow, manual process.

The combination unites two classic species of founder: Laberge dropped out of high school to pursue his first company, which did similar fluid analysis for large construction machines that depend on hydraulics. Adams was an academic, trained as a physicist, who spent much of his career developing advanced ocean sensing technology as an MIT professor and Woods Hole Oceanographic Institute researcher before starting his company.

Now, with Adams as the company’s chief science officer, the combined teams are working to integrate both technologies into one sensor, which they expect to have on the market by early next year. Omen’s initial sensors will be monitoring chips pulling in 250 to 300 MW of power by the end of the year, worth about $5 to $7 billion, Laberge told TechCrunch, with the hope of being installed on the equivalent of 1 GW of chips by the end of 2027.

The added value is in uptime: If the fluid system gets out of whack, data center operators typically choose to dump the water rather than risk damage to their expensive equipment. That means shutting down operations until the fluid system is restored, which costs a lot of money. Monitoring lets data center operators use the same fluid for longer, and minimize the costs of a clean up.

Adams hopes to see their technologies have an impact outside of data centers as well, telling TechCrunch that “water goes into data centers, and then it comes out and gets dumped into municipal water, and you want that whole process to happen in a way that’s as safe and responsible as possible — and that means you got to measure stuff.”

AI companies once estimated as much as $4 trillion would be spent on AI infrastructure by 2030, but some investors have grown skeptical of the buildout, sending hyperscaler stocks plummeting in recent days. Omen’s team isn’t worried about shifting sentiment though, since their work comes after a data center has been built.

“We’re working customers on stuff they’re building today or have already built, and so it doesn’t really affect us because all of those projects were financed 12 months ago,” Laberge said. “Our view is, if there’s tightness, this is still going to be the biggest market in human history, even if you sliced it in half.”

View Bio

I never thought I’d get excited about replacing my microwave, but the Ninja Crispi Microwave and Air Fryer has made a pretty convincing case for ditching two appliances and making room for one. Combining a microwave with a built-in air fryer, it’s designed to speed up everyday cooking, give leftovers their crunch back and, perhaps best of all, free up some valuable kitchen counter space.

Priced at $449.99, it’s certainly not a budget-friendly microwave. But this is no ordinary machine for heating leftover soup. Ninja has paired a 1,000W microwave with a 1,700W air fryer system, powered by its FusionCrisp technology, which combines the speed of microwave cooking with the golden, crispy results you’d typically expect from an air fryer.

Think leftover pizza, chicken tenders, fries, and all those foods that tend to turn sad and soggy in a traditional microwave. Instead, the idea is to get food that’s piping hot and juicy on the inside, with a deliciously crisp, golden finish on the outside. In other words, no more choosing between speed and texture.

Of course, microwave and air fryer combinations aren’t entirely new. Brands such as Smeg have already explored the concept, while plenty of modern microwaves come with a range of additional cooking functions. 

What makes the Ninja Crispi Microwave particularly interesting, however, is its FusionCrisp technology, which is designed to bring both cooking methods together rather than simply offering two separate functions. It comes with a 5.5-quart glass cooking dish and a nonstick crisper plate, which sits on top when you want to air fry and crisp up your food. 

Another part that makes it stand out is that there isn’t a rotating plate inside, opening up a lot more space to cook more dishes at once.

It’s not just about reheating leftovers. For example, if you want to make a pasta bake, the microwave can cook the pasta, while the air fryer crisps up the cheese on top. 

There’s one thing to consider before buying it, though. Measuring 20.3 x 17.56 x 14.1 inches, the microwave is hardly compact. It takes up a fair amount of counter space, so while it could free up room if you’re replacing a separate microwave and air fryer, it’s not necessarily the ideal solution for a small apartment kitchen.

Then there’s the $449.99 price tag. That’s a considerable amount to spend on a microwave, particularly when a basic model can handle everyday reheating for a fraction of the cost. But if you’re someone who regularly reaches for both a microwave and an air fryer, the combination could make a lot more sense.

View Bio

Space infrastructure company Northwood Space on Monday said it had raised $375 million in a Series C round co-led by Andreessen Horowitz and DFJ Growth, valuing the company at $2 billion.

Founded in 2022, Northwood builds antennas, software, and other technologies that help satellites communicate better with Earth. Bridgit Mendler, co-founder and former actress, serves as its CEO. 

The company last raised a $100 million Series B in January, also co-led by Andreessen Horowitz, along with a nearly $50 million contract with the U.S. Space Force to upgrade the nation’s “satellite control network,” TechCrunch previously reported.

This latest round also includes a financial agreement with HSBC that, along with the funding, will be used to scale the company, hire, and fulfill existing contracts, the startup said.