Flipkart co-founder Binny Bansal leaves PhonePe board
Flipkart co-founder Binny Bansal has stepped down from PhonePe’s board of directors, three quarters after making a similar move at the e-commerce giant.
Bengaluru-headquartered PhonePe said it had appointed Manish Sabharwal, an executive at staffing and HR firm Teamlease, as an independent director and audit committee chair.
Bansal played a key role in Flipkart’s acquisition of PhonePe in 2016 and sat on the fintech’s board ever since. The Walmart-backed startup, which operates the top mobile payments app in India, branched out of Flipkart in 2022 and was valued at $12 billion in funding rounds that netted about $850 million last year.
Bansal still owns about 1% of PhonePe. Neither of the parties shared why he was leaving the board.
“I would like to express my heartfelt gratitude to Binny Bansal for being one of PhonePe’s earliest and staunchest supporters,” Sameer Nigam, co-founder and chief executive of PhonePe, said in a statement. His active engagement, strategic guidance, and personal mentorship have profoundly enriched our discussions. Binny will be missed!”
Keep reading the article on Tech Crunch
New UK crypto regulations will include stablecoins and staking services
The U.K. is set to create an all-encompassing regulatory framework to govern the crypto sector in early 2025.
Speaking on Thursday at the Tokenisation Summit in London, economic secretary to the treasury Tulip Siddiq confirmed that the new rules would include cryptocurrency and stablecoins, which are pegged against a more stable asset such as a fiat currency. The news was first reported by Bloomberg.
The crypto industry had been hoping regulations would exclude so-called “staking” services, which is where investors lock their tokens to support a particular blockchain in exchange for a small yield — much like earning interest on cash savings. But Siddiq suggested there would be no special carve outs.
“For me, it doesn’t make sense for staking services to have this treatment,” Siddiq said. “The government intends to proceed with removing this legal uncertainty accordingly.”
The U.K. also recently introduced a new bill that, if passed, would give greater legal protections to crypto assets such as Bitcoin and NFTs, designating them as “personal property” under the proposed law.
Keep reading the article on Tech Crunch
A company is now developing human washing machines
Forget cold plunges. The new flex could soon be human washing machines.
According to one of Japan’s oldest newspapers, an Osaka-based shower head maker called Science has developed a contraption that’s shaped like a cockpit, fills with water when a bather sits in a seat at its center, and measures the person’s pulse and other biological data via sensors to ensure the temperature is just right. It also “projects images on the inside of [its] transparent cover to help the person feel refreshed,” says the outlet.
Dubbed “Mirai Ningen Sentakuki” (human washing machine of the future), the apparatus might never go on sale. Indeed, for now the company’s plans for it appear limited to an expo in Osaka this April, where up to eight people can experience a 15-minute-long “wash and dry” each day after first booking a reservation.
Still, a home-use version is reportedly also in the works.
Keep reading the article on Tech Crunch
Texas AG opens investigation into advertising group that Elon Musk sued for ‘boycotting’ X
Texas Attorney General Ken Paxton announced on Thursday he is opening an investigation into the World Federation of Advertisers (WFA) to determine whether the trade group’s members conspired to boycott “certain social media platforms.” While the press release doesn’t name social media platforms by name, one of them is likely Elon Musk’s X, which filed an antitrust lawsuit against the WFA in August and alleged that advertisers orchestrated a “systematic illegal boycott” of the platform.
“Trade organizations and companies cannot collude to block advertising revenue from entities they wish to undermine,” said Paxton in the press release. “Today’s document request is part of an ongoing investigation to hold WFA and its members accountable for any attempt to rig the system to harm organizations they might disagree with.”
Several of the WFA’s members – which includes global brands such as IBM, The Coca-Cola Company, and CVS Health – have stopped or significantly reduced the amount they spend for advertising on X since Elon Musk’s takeover of the company. There was an especially large exodus of advertisers, including Apple and Disney, from X in November 2023 following reports from the Center for Countering Digital Hate and Media Matters that suggested Elon Musk’s X had failed to moderate its platform and remove illegal or hateful content. At the time, a White House spokesperson condemned Elon Musk for one of his personal posts, which it called “antisemitic and racist.”
Since then, X has sued many advertisers and ad groups, claiming these global brands were not reducing their ad spend based on individual decisions, but instead collectively conspiring to withhold billions of dollars in revenue from X. Now it appears Texas’ AG is bringing an investigation of his own.
“It’s still a major problem,” said Musk in response to Paxton’s Thursday post on X about the advertiser investigation.
Much like X’s lawsuit, Paxton zeroes in on a since-discontinued, not-for-profit organization within the WFA, the Global Alliance for Responsible Media or GARM. This was a US-based group founded in 2019 that included some of the country’s largest advertisers. It created frameworks and definitions for companies to understand hate speech, brand safety, and misinformation.
The AG’s investigations asks for documents and information from GARM that could reveal whether it told brands to boycott certain social media platforms that violated its brand safety standards.
When announcing her platform’s lawsuit against advertisers, X CEO Linda Yaccarino cited a July report from the U.S. House of Representatives Judiciary Committee looking into GARM’s practices. That report found:
Through GARM, large corporations, advertising agencies, and industry associations participated in boycotts and other coordinated action to demonetize platforms, podcasts, news outlets, and other content deemed disfavored by GARM and its members. This collusion can have the effect of eliminating a variety of content and viewpoints available to consumers.
GARM closed its doors in August, shortly after X sued, noting that it did not have the resources or finances to continue operating.
In the months leading up to this investigation, some advertisers have actually resumed ad spending on X, though at much lower rates than before. Comcast, IBM, Disney and other major brands reportedly returned to Musk’s platform this year. Furthermore, X announced in October that it reached an agreement with Unilever to resume its ad spending, and that the social media platform would drop its claims against Unilever, which X previously named as one of the companies that participated in the alleged boycott.
X and the World Federation of Advertisers did not immediately respond to TechCrunch’s request for comment.
Keep reading the article on Tech Crunch
American Airlines is deploying new tech to shame boarding line cutters
American Airlines has a new tactic for shaming boarding line cutters: A loud beeper.
CNBC reports that the airline is rolling out a system that emits two loud beeps when a traveler tries to get on an airplane before their boarding group is called.
As of Wednesday, the new tech was in over 100 airports around the U.S. following tests at Albuquerque International Sunport, Ronald Reagan Washington National Airport, and Tucson International Airport.
CNBC notes that other airlines have experimented with less, well, audible ways to prevent gate crowding, which can interfere with the boarding process for flyers who pay a premium to get seated early. United, for example, texts customers when it’s time to board, provides a countdown-to-boarding clock via its iOS app, and has digital signs showing which boarding group has been called.
Keep reading the article on Tech Crunch
Hackers break into Andrew Tate’s online ‘university,’ steal user data and flood chats with emojis
Hackers have breached an online course founded by ostensible influencer and self-described misogynist Andrew Tate, leaking data on close to 800,000 users, including thousands of email addresses, and private user chat logs.
The Daily Dot, which broke the news Thursday, reported that the hackers accessed the user data, then flooded the online course’s chatroom with emojis that “included a transgender flag, a feminist fist, an AI-generated image of Tate draped in a rainbow flag, another where his buttocks are enlarged,” among others.
The hacktivists provided the hacked data on the course’s users to The Daily Dot, which handed the records to data breach notification site Have I Been Pwned, and DDoSecrets, a nonprofit collective that stores leaked datasets in the public interest.
Per the BBC News, Tate remains under house arrest in Romania awaiting trial on charges on allegations of human trafficking and rape.
Keep reading the article on Tech Crunch
Apple is reportedly building a more conversational Siri powered by LLMs
Apple is developing a new version of its voice assistant, Siri, powered by advanced large language models (LLMs), according to sources cited by Bloomberg. This more conversational Siri is part of Apple’s attempt to catch up in AI, where competitors have released impressive features, like Google’s Gemini Live, that are far more natural to talk to than Siri.
The new assistant reportedly will fully replace the Siri interface that users rely on today, and Apple is planning to release the feature in the spring of 2026. The feature seems like it will be similar to OpenAI’s Advanced Voice Mode but with all the same access to personal information and apps that Siri has today.
Until then, Apple is relying on third parties to power the iPhone’s advanced AI features. OpenAI’s ChatGPT will become available inside Apple Intelligence in December, and Apple has reportedly discussed similar deals with other AI providers, such as Google and Anthropic.
Keep reading the article on Tech Crunch
Future Google supplier Kairos gets approval to build two small nuclear reactors
Nuclear startup Kairos Power received approval from the U.S. Nuclear Regulatory Commission to start construction on two test reactors in Oak Ridge, Tennessee. The permit marks a significant milestone for Kairos, which in October inked a deal with Google to provide 500 megawatts of electricity for its data centers.
The fluoride-salt cooled, high-temperature reactors are scaled down versions of what Kairos hopes to ultimately build to supply Google with electricity starting in 2030. And while the new reactors are technically test beds, Kairos intends to connect the power plant to the grid, spokesperson Ashley Lewis told TechCrunch.
The Hermes 2 reactors will be capable of producing 35 megawatts of heat each, and they’ll be connected to a 20 megawatt turbine to turn that heat into electricity. Kairos’ commercial-scale power plant will also feature two reactors capable of generating a combined 150 megawatts of electricity.
Kairos’ design differs from existing nuclear reactors in two key ways: The fuel is made of uranium coated in carbon and ceramic shells, which are intended to be durable enough to contain fissile material in the case of an accident. And the reactor isn’t cooled by water but by molten salt.
The small modular reactor (SMR) startup, which has received a $303 million award from the U.S. Department of Energy, has been working for years to refine its molten salt-cooling system. Fluoride salts’ extremely high boiling points allow the coolant to flow under low pressure. That means in the case of an accident, there won’t be any high-pressure, radioactive material waiting to burst forth should pumping systems fail. Plus, Oak Ridge National Laboratory says that, should power to the pumps fail, molten-salt reactors can rely on passive convection to move salt through the reactor to cool it.
Altogether, those features are enough to qualify Kairos’ designs as “Generation IV” reactors, a classification system created by an international organization backed by national nuclear agencies. The classification system is both vague and broad, so it’s hard to tell exactly how Hermes 2 might score on the rubric.
Kairos has been inching toward approval for the reactor design for the last year and a half. Hermes 2 passed its safety review with the NRC in July and its environmental assessment in August. All told, it took 18 months for the NRC to issue the construction permit, a relatively swift timeline compared with previous reactor permits.
Now the pressure is on Kairos to deliver on its promises. The company says it hopes to have the first reactor for the Google deal online in 2030 and the rest completed by 2035. In the world of nuclear power, a decade isn’t much time at all.
Keep reading the article on Tech Crunch
Zepto raises another $350M amid retail upheaval in India
Zepto has secured $350 million in new funding, its third round of financing in six months, as the Indian quick-commerce startup strengthens its position against its competitors ahead of a planned IPO next year.
Indian family offices, wealthy individuals, and asset manager Motilal Oswal invested in the round, which maintains Zepto’s $5 billion valuation. Motilal co-founder Raamdeo Agrawal, family offices of Mankind Pharma, RP-Sanjiv Goenka, Cello, Haldiram’s, Sekhsaria, and Kalyan, as well as celebrities Amitabh Bachchan and Sachin Tendulkar are among the backers in the new investment, which is the largest fully domestic primary round in India.
The funding push comes as Zepto rushes to add Indian investors to its cap table, with foreign ownership currently exceeding two-thirds. TechCrunch first reported about the new round’s deliberation last month. The Mumbai-headquartered startup has now raised over $1.35 billion since June.
Quick-commerce sales — delivering grocery and other items to customers’ doorsteps in 10 minutes — in India are set to surpass $6 billion this year. Morgan Stanley projects the market to be worth $42 billion by 2030, representing 18.4% of total e-commerce and 2.5% of retail sales. These strong growth prospects have forced established players, including Flipkart, Myntra, and Nykaa, to cut delivery times as they lose business to specialized delivery apps.
Even though quick commerce hasn’t made inroads in most pockets of the world, the model seems to be working especially well in India, where unorganized retail stores are ever present.
Quick-commerce platforms are creating a “parallel commerce for convenience-seeking customers” in India, Morgan Stanley wrote in a note this month.
Zepto and its rivals — Zomato-owned Blinkit, Swiggy-owned Instamart, and Tata-owned BigBasket — currently operate at lower margins than traditional retail, and Morgan Stanley expects market leaders to reach contribution margins of 7% to 8% and adjusted EBITDA margins of more than 5% by 2030. (Zepto is currently spending about $35 million a month, according to many people familiar with the figure.)
Zepto, which serves a total of more than 7 million orders in over 17 cities daily, is on track to record annualized sales of $2 billion, according to an investor presentation reviewed by TechCrunch. It projects 150% growth over the next 12 months, CEO Aadit Palicha told investors in August. The startup plans to go public in India next year.
However, the fast growth of quick-commerce has had a devastating effect on mom-and-pop shops that dot thousands of Indian cities, towns, and villages.
Around 200,000 neighborhood stores have closed in the past year, with 90,000 stores shutting down in major cities where quick commerce is more prevalent, according to the All India Consumer Products Distributors Federation.
The federation warns that without regulatory intervention, more neighborhood stores face closure as quick-commerce platforms prioritize growth over sustainable practices.
Zepto said it has created work opportunity for hundreds of thousands of gig workers. “From day one, our vision has been to play a small role in nation-building, create lakhs of jobs, and offer better services to Indian consumers,” said Palicha in a statement.
Regulatory challenges are looming. Unless an e-commerce firm is majority-owned by an Indian company or person, current rules prevent it from operating on an inventory model. Quick-commerce firms are currently not compliant with these rules.
Keep reading the article on Tech Crunch
YouTube Shorts’ Dream Screen feature can now generate AI video backgrounds
YouTube announced on Thursday that its Dream Screen feature for Shorts now lets you create AI-generated video backgrounds. Up until now, Dream Screen could only create image backgrounds. The new capability is possible due to the integration of Google DeepMind’s AI video-generation model, Veo, which can create 1080p video clips in various cinematic styles.
Users can now use the feature to turn text prompts into AI video backgrounds. To access the feature, you need to go into the Shorts camera, tap the “green screen” icon, and then select “Dream Screen.” From there, you can type in a prompt, such as “candy landscape” or “magical forest with a stream.” Next, you can pick an animation style and then tap “Create.”
The Dream Screen tool will then generate a series of video backgrounds that you can choose from. Once you select an AI video background, you can record a video with it behind you.
YouTube says the new feature can be used for things like putting yourself in the setting of your favorite book or creating an animated opening for your Short.
In the future, YouTube plans to give creators the ability to generate six-second-long stand-alone video clips for Shorts through Dream Screen.
It’s worth noting that YouTube is giving Shorts creators access to a feature that’s not available on TikTok, the leading short-form video platform. While TikTok lets creators generate AI-powered background images, the company currently doesn’t let you create video backgrounds.
The new feature is now available in the United States, Canada, Australia, and New Zealand.
Keep reading the article on Tech Crunch
Marissa Mayer just laid out a possible business model for ad-supported AI chatbots
Marissa Mayer has a lot of insights into the promise and problems with online advertising. She played an instrumental role in the early days of Google Search and spent several years as Yahoo’s CEO.
Today, Mayer is the CEO of her own company, Sunshine, which is creating apps to do things like share photos among groups more efficiently, clean up your contacts, and remember your friends’ birthdays. While none of these apps has taken off yet, Mayer’s background makes it worth considering her opinion as it relates to online advertising.
On Wednesday at the Cerebral Valley AI Summit in San Francisco, Mayer was asked how she envisions advertisers responding as AI tools change expectations from consumers about what information is available and how it’s presented.
Her answer: Advertisers will be compelled to turn over more data than ever before in order to give consumers the most precise and detailed answers possible.
She cited the example of concert tickets in the early days of Google Search.
“One of the classic examples we used to talk about how ads make search better was concert tickets. When people search for concert tickets, the fact that there’s an advertiser there that has tickets to sell you and they’re willing to pay to be in your search results is actually a sign of quality, and it’s also where the searcher actually is happy — they don’t want these articles about the concert they want to see; they actually want tickets to purchase. And so there’s a nice meeting of expectations on both the advertiser and searcher side.”
In the AI era, Mayer imagines that when people ask about tickets for a specific concert, “they actually want to see exactly what seats are available, where they are in the stadium, the pricing. They want that information synthesized in much the way they see it synthesized in generative AI. And so I think it means that advertisers are going to have to partner even more closely with Google and other search engines to allow their wares to really be showcased and synthesized with the answer.”
When interviewer Max Child asked Mayer if companies like StubHub or Ticketmaster would be willing to give over enough data to Google to provide this level of detail, she noted, “I think that it’s pretty clear if you look at where search ads were 10 years ago versus where they are today, and certainly where Google Shopping is, there are a lot more advertisers that are giving full information of their inventory and a lot of different aspects and facets of the data, and so I think that trend is ultimately going to continue.”
Although Mayer was talking about search specifically, it’s also an interesting hypothetical business case for pure-play AI providers like OpenAI and Perplexity. It’s possible to imagine, for example, advertisers partnering with these companies to give sponsored answers to specific types of queries, especially where the answers actually match up with what the user is looking for.
As the compute costs for AI continue to rise, AI companies will certainly be driven to seek out new sources of revenue.
Disclosure: Yahoo is the owner of TechCrunch.
Keep reading the article on Tech Crunch
Battery unicorn Northvolt files for bankruptcy, upending Europe’s industrial plan
Beleaguered Swedish battery manufacturer Northvolt announced today that it was filing for bankruptcy in the U.S., striking a blow to Europe’s ambitions for homegrown lithium-ion batteries.
The company reportedly chose Chapter 11 in an effort to right its finances.
Northvolt, which had soared for years on the back of strong fundraising and a string of announcements about new facilities, has stumbled of late. It laid off 1,600 employees, about 20% of its workforce, in September, and unloaded assets in November from its ill-fated purchase of Bay Area battery startup Cuberg.
The company has raised $14.26 billion, according to PitchBook, including a $1.2 billion round in 2023 to expand operations in North America. But that wasn’t enough to keep the cash-starved operation going. The company reportedly was burning through $100 million per month. When BMW pulled out of a $2 billion contract in June after Northvolt failed to deliver on time, bankruptcy became almost inevitable.
This isn’t the first time a battery manufacturing startup hit a rough patch — A123 Systems’s failure over a decade ago stands out as a prominent example in the U.S. — and it likely won’t be the last. Making lithium-ion cells is fiendishly difficult, requiring deep knowledge of chemistry, production equipment, and quality improvement. Even leading companies suffer from costly problems, sometimes to the tune of $1 billion. Northvolt’s bankruptcy is probably more a sign of poor execution than softer-than-expected demand for EVs.
Is this the end for Swedish company? Not necessarily. For one, Volkswagen owns part of the company, and it has placed a big bet on EVs, for which it’ll need millions of cells. Plus, Europe, like other advanced economies, has been rushing to stake a claim in the battery manufacturing space, and Northvolt seemed like its best chance to compete with Asian rivals. It may still be, perhaps with the help of one of those rivals through some kind of partnership, but first it has to get its house in order.
Keep reading the article on Tech Crunch
Brave Search adds AI chat for follow-up questions after your initial query
Brave announced on Thursday that it’s introducing an AI chat mode for follow-up questions based on initial queries on Brave Search. Earlier this year, the company launched “Answer with AI” summaries that appear above search results after you submit a query to give you an easy-to-read answer in response to a question. Now the Answer with AI summaries will include a chat bar that lets you ask follow-up questions regarding your initial query.
The new feature gives users access to an experience that’s not available on Google, the largest search engine in the world. While Google does offer “AI Overviews,” which are similar to Brave’s “Answer with AI” summaries, the company does not offer a way for users to ask additional questions based on an initial query, as users instead have to complete an entirely new search when looking for more information.
If you go to Brave Search and type in “Christopher Nolan films,” you will get an AI-generated summary that gives you an overview of who Christopher Nolan is and a list of some of his notable films. Starting today, you will now see a chat bar under the AI-generated summary that reads, “Ask a follow-up question.”
You could then enter a follow-up query, such as, “Which actors appear most in his films?” The AI chatbot will then give you a list of actors that Nolan has frequently featured in his movies.
Once you ask a follow-up question, you will enter a chat experience that offers sources along the way with each new question to highlight where the chatbot is getting the information.
Brave says this new experience, which is powered by a combination of open and internal LLMs, eliminates the need for users to choose between chat-first or search-first tools. For instance, search engines are good at fetching images, links, and information, while AI chatbots are great for exploring topics through related questions based on context. Brave is now bringing these two types of tools together with its latest feature.
Brave notes that your conversations with Answer with AI remain safe and private, as Brave Search does not store your queries or profile you.
The company revealed that Brave Search serves more than 36 million queries per day and that its Answer with AI tool generates more than 11 million answers daily.
Keep reading the article on Tech Crunch
SpaceX signs second commercial deal for Starship lunar lander with Lunar Outpost
As SpaceX’s Starship test program continues to gain momentum, the company signed its second commercial deal to deliver a payload to the moon using a lunar lander variant of the massive vehicle.
Starship will deposit Lunar Outpost’s rugged rover, called Eagle, to the lunar surface by 2029, the Colorado-based startup said. Lunar Outpost CEO Justin Cyrus declined to provide other details about the mission, such as whether this will be a dedicated flight or part of a ride-share, or the cost. It’s also unclear how Starship will actually transport the rover, which looks like a small pickup truck, from inside the vehicle to the surface.
Lunar Outpost’s rover ambitions got a major boost from NASA earlier this year when the space agency selected it, along with two other teams, for the initial phase of a Lunar Terrain Vehicle (LTV) program. The three teams were selected for the one-year contract to advance their rover concepts for potential use by astronauts under NASA’s Artemis program. The 12-month period will culminate in a subsequent competitive request for proposals, where the three companies will then compete for a demonstration task order.
The LTV program has the potentially to be very lucrative: The total potential value of the task orders over the next 13 years is $4.6 billion, though the agency has said that due to budgeting concerns, it is only selecting a single provider.
Even if the team led by Lunar Outpost, which also includes Leidos, General Motors, Goodyear, and MDA Space, is not selected by NASA, Cyrus said the company plans on proceeding with rover development.
“Lunar Outpost Eagle is our flagship vehicle and will be the backbone of outposts on other planetary bodies,” he said in an emailed statement. “Our company has been focused on mobility since inception and the lunar terrain vehicle has accelerated our truly industry leading capabilities which unlock a sustainable presence on the moon and Mars.”
The startup also closed a Series A round for an undisclosed amount last month to support several initiatives, including the rover development.
The new deal for the Starship cargo delivery follows a similar contract SpaceX made last year with Venturi Astrolab — which is also leading a team selected under the LTV program — to deliver its rover to the moon’s surface.
Keep reading the article on Tech Crunch
Cruise fesses up, Pony AI raises its IPO ambitions, and the TuSimple drama dials back up
Welcome back to TechCrunch Mobility — your central hub for news and insights on the future of transportation. Sign up here for free — just click TechCrunch Mobility!
President-elect Trump has wasted little time picking cabinet members, including naming former Wisconsin Rep. Sean Duffy as the next secretary of transportation. If Duffy’s appointment is confirmed by the Senate, the new secretary will oversee an annual budget of about $107 billion, not to mention a broad range of issues that will touch the tech world, including fuel efficiency standards, advanced driver assistance technologies, and autonomous vehicles.
I took a look at his legislative record and there’s nothing really related to transportation, save for a bill pertaining to logging vehicles. That makes it difficult to predict exactly what his priorities will be. Stay tuned.
Two more fun things. During an episode of TechCrunch’s Equity, Salesforce co-founder and CEO Marc Benioff talked about some of his favorite gadgets. And you guessed it, cars and trucks — many of which are EVs — were on top.
And this is not quite transportation-related, so forgive me. Tesla’s Optimus bot has a new friend: Kim Kardashian.
Before we jump into the news, one housekeeping item: We will not have an edition of TechCrunch Mobility next Thursday due to the Thanksgiving holiday. For all my U.S.-based readers — plus those abroad who celebrate the holiday — stay safe out there on those roads. I know what I will be giving thanks to this holiday: good health and the ability to do what I love, which includes writing this very newsletter.
A little bird
Got a tip for us? Email Kirsten Korosec at [email protected], Sean O’Kane at [email protected], or Rebecca Bellan at [email protected]. Or check out these instructions to learn how to contact us via encrypted messaging apps or SecureDrop.
Deals!
Last month, Pony AI joined the list of Chinese firms going public on the U.S. stock market after a multiyear ban from Beijing on offshore capital raising. Now we know what Pony AI is aiming for.
Pony AI wants to raise around $260 million, up from its previous target of around $224 million, Sean O’Kane reported. To meet the goal, Pony will sell as many as 20 million American depository shares.
That’s still down from the much higher target of $425 million, which Pony was aiming for earlier this year before its board approved a reduction of its minimum valuation from around $8 billion to $4 billion.
Other deals that got my attention …
Ample, the San Francisco-based EV battery-swapping company, raised $25 million from new investor Mitsubishi Corporation.
Class8, a Toronto-based trucking fleet operations startup previously called FleetOps, raised $22 million in a Series A round led by Xplorer Capital. New investor Commerce Ventures and returning backers Inspired Capital and Resolute Ventures also joined.
Ecolectro, which makes hydrogen-producing electrolyzers, raised $10.5 million in a Series A round led by the Toyota Ventures climate fund with participation from Cornell University, New Climate Ventures, Starshot Capital, and Techstars.
ePlane, an Indian-based startup developing electric vertical takeoff and landing aircraft, raised $14 million in a Series B round co-led by Speciale Invest and Antares Ventures. The all-equity round also included participation from Micelio Mobility, Naval Ravikant, Java Capital, Samarthya Investment Advisors, Redstart (from Naukri), and Anicut. The round has valued the startup at $46 million post-money — over 2x its previous $21 million valuation.
Flipturn, a New York-based startup that developed software to help fleets manage their EV charging, raised $11 million in a Series A round led by CRV with participation from Accel.
Notable reads and other tidbits
Autonomous vehicles
Cruise admitted to submitting a false report with the goal of influencing a federal investigation into a safety incident last year. The GM subsidiary agreed to pay a $500,000 criminal fine as part of a deferred prosecution agreement.
May Mobility is starting to test driverless shuttles (self-driving Toyota Sienna minivans) in Ann Arbor.
Nuro started rolling out a test fleet of its driverless, passenger-less R3 vehicles across the Bay Area and Houston, two months after changing its business strategy to license its AV tech to automakers and mobility providers. The reason? Show potential customers what they’ve got.
Xiaodi Hou, the co-founder and former CEO of self-driving trucking startup TuSimple, has urged a California district court to issue a temporary restraining order to stop the company from transferring its remaining U.S. assets to China. This fight is getting messy.
Electric vehicles, charging, & batteries
Hyundai revealed the new Ioniq 9, a three-row SUV that is expected to come to the U.S. in the spring of 2025. Read more from contributor Emme Hall about this new EV and its numerous features.
Remember the Ram Ramcharger? Well Stellantis has released new details about STLA Frame, the underlying vehicle platform that the Ram Ramcharger will eventually be built on. TL;DR: This is designed to be a multi-energy platform, which means it can be the foundation for an internal combustion, hybrid, battery-electric, or even gas-extending vehicle like the Ram Ramcharger.
The $7,500 EV tax credit is on Trump’s chopping block.
In-car tech
Ford agreed to pay a $165 million penalty to federal regulators after moving too slowly to recall vehicles with faulty rearview cameras.
People
General Motors has cut another 1,000 jobs in the U.S., none of which were hourly workers. And Ford said it will slash another 4,000 jobs in Europe in a cost-cutting strategy as the growth rate for EVs has slowed.
Rivian’s head of communications, Sarah O’Brien, who once held a similar role at Tesla, has left the EV maker. She has joined LoveFrom, the creative collective founded by Apple’s former chief design officer Jony Ive.
Meanwhile, Volkswagen has tapped former Rivian exec Kjell Gruner to run its U.S. business.
This week’s wheels
What is “This week’s wheels”? It’s a chance to learn about the different transportation products we’re testing, whether it’s an electric or hybrid car, an e-bike, or even a ride in an autonomous vehicle. Future vehicles include the Lucid Air, more time in the next-gen Rivian R1S, and the Volkswagen ID Buzz. Stay tuned.
Keep reading the article on Tech Crunch
German fintech unicorn N26 just had its first profitable quarter
Ten years after pitching on stage at TechCrunch Disrupt in London, fintech N26 has reported its first ever quarterly (pre-tax) profit. The challenger bank with millions of customers across Europe generated a net operating income of €2.8 million during the third quarter of 2024 (or $2.9 million at current exchange rates).
This is an important milestone for the startup but also significant news for the fintech industry. Challenger banks like Monzo, N26, Revolut and Starling used to be some of the most hyped startups in Europe. They raised billions in funding, expanded aggressively, and overspent to reach that next funding round.
Now, it’s time to sit down and do the math. Large funding rounds are harder to obtain and investors now often require a clear path to profitability.
Revolut is extremely profitable — $428 million in net profits for 2023 alone — while Monzo just crossed the line with a pre-tax profit of £15.4 million for 2023 ($19.4 million). N26 is following suit.
For several years, Germany’s financial regulator BaFin imposed a cap on new signups as a sanction to force the startup to improve its anti-money laundering processes. But it lifted the cap earlier this year and that has had a significant impact on the company’s bottom line.
According to N26, more than 200,000 people currently open an account with it every month. Interestingly, N26 stopped sharing the total number of users it has. Instead, the company focuses on its 4.8 million “revenue-relevant” customers.
The influx of new users has led to a 40% revenue increase for the fintech in 2024 compared to 2023. And N26 is on track to generate €440 million in annual revenue this year.
As a reminder, in addition to free accounts, N26 offers paid subscriptions with access to more financial services and features. The company also offers savings accounts, stock, and crypto trading as well as credit products.
Now, let’s see if N26 manages to stay in the black as 50% of its 2024 revenue comes from interest revenues from customer deposits and the company’s retail lending activities. With interest rates going down in Europe, that source of revenue will also be more difficult to maintain at a high level.
Keep reading the article on Tech Crunch
Teleo wants to help the robotics industry reach its ‘ChatGPT moment’
Teleo describes itself as a construction robotics startup, but its mission is bigger than automating heavy equipment like excavators and tractors. Today, Teleo’s retrofitted machinery allows its customers to operate their existing fleets semi-autonomously. In the future, the startup sees the data it collects as a key enabler for the robotics industry to reach its “ChatGPT moment.”
That isn’t an aspiration to reach the same level of hype surrounding ChatGPT. Instead, Teleo CEO Vinay Shet sees an opportunity for robotics companies — and namely the one he runs — to gather vast datasets similar to amounts used to build ChatGPT in order to make big, game-changing leaps in robotics.
And investors seem keen to help the startup reach that milestone. TechCrunch has learned that Teleo recently raised $16.2 million in funding through two extensions to its 2022 Series A round. The $9.2 million extension closed in April and another $7 million one closed this week, according to recent filings and information from the company.
“The foundation models that led to the creation of ChatGPT relied significantly on, effectively, trillions of tokens worth of data that was freely available on the internet, on language, on videos, on images, and so forth. That data does not exist in robotics,” Teleo CEO Vinay Shet told TechCrunch. “The best data set that we know of in the robotics world is about 2.4 million tokens, whereas in the language world, they train it on trillions of tokens.”
Teleo aims to bridge that gap by logging data from its own daily operations, which Shet says will end up “becoming the basis upon which you can train true robotic foundation models” that can lead to generalized intelligence.
To build that repertoire of data, Teleo needs to deploy rapidly, at scale, and across several industries. And the company’s strategy for doing that is wrapped up in its semi-autonomous approach. Teleo can retrofit any piece of equipment with the necessary self-driving software and sensors – like cameras, lidar, and radar – to drive themselves autonomously in limited conditions. Remote human operators then step in to perform more complicated tasks, like unloading a dump truck, and can typically handle multiple vehicles at a time.
“That combination is what lets us solve the whole use case for the customer while delivering [return on investment] to the customer and making money through a standalone product,” Shet said.
In order to maintain a diverse dataset, Teleo recently expanded beyond construction and is deploying autonomous heavy machinery, like wheel loaders, terminal tractors, and excavators, across a range of industries, including pulp and paper, logging, port logistics, agriculture, and munitions removal. Teleo is also targeting industries like airports, waste and recycling, logistics, and snow removal.
The hope is that the data it collects – including inputs from human operators, video footage, and sensor feedback – will allow Teleo to fine-tune or specialize foundational robotics models. This could eventually enable the replacement or augmentation of the human in the loop with a cloud-based AI agent capable of learning to control different machines, as a human would.
This long-term thinking is no doubt what attracted investors to the company. Teleo’s recent extensions were led by UP.Partners with participation from new investors Trousdale Ventures and Triatomic Capital, as well as returning investors F-Prime Capital and Trucks VC, among others.
Teleo says the funds will be used to scale customer deployments, continue expanding to new industries, and enhance the startup’s AI capabilities, including an integration of large language models (LLMs) to unlock operator efficiency.
“Over the next several years, you will see vertically integrated companies like ourselves actually deploy in the real world in a manner that makes sense economically speaking and grows economically based off that,” Shet said. “But along the way, they’ll collect enough data in the right format so that it unlocks that ‘aha’ moment a few years down the road.”
Keep reading the article on Tech Crunch
WhatsApp rolls out voice message transcripts
WhatsApp announced on Thursday it’s rolling out voice message transcripts. The Meta-owned company says the new feature will come in handy when you’re on the move or in a loud place where you can’t listen to a voice message. A similar feature was announced last year for Apple’s Message app, a top WhatsApp competitor, with the release of Apple’s software update, iOS 17.
With the new feature, voice messages can be transcribed into text to help you keep up with conversations when you’re unable (or perhaps unwilling) to listen to a voice message. WhatsApp claims its transcripts are generated on your device and no one else can read them.
To access the new WhatsApp feature, you need to go into your settings, select the “Chats” section, and then tap “Voice message transcripts” to turn the feature on. From there, you can select your transcript language.
Once you have turned on the feature you can get a transcript for a voice message by long-pressing it and then tapping the “transcribe” option. It will not automatically transcribe every message for you, that is.
Voice message transcripts are rolling out globally over the coming weeks in select languages, the company says.
On iOS, the languages that are supported depend on your OS. If you have iOS 16+, the supported languages include English, Spanish, French, German, Italian, Japanese, Korean, Portuguese, Russian, Turkish, Chinese, and Arabic. If you have iOS 17+, the supported languages include Danish, Finish, Malay, Norwegian, Dutch, Swedish, Hebrew, and Thai.
If you have an Android, you can access transcripts in English, Portuguese, Spanish, and Russian.
WhatsApp plans to add support for additional languages in the coming months.
Keep reading the article on Tech Crunch
AI2’s open source Tulu 3 lets anyone play the AI post-training game
Ask anyone in the open source AI community, and they will tell you the gap between them and the big private companies is more than just computing power. AI2 is working to fix that, first with fully open source databases and models, and now with an open and easily adapted post-training regimen to turn “raw” large language models into usable ones.
Contrary to what many think, “foundation” language models don’t come out of the training process ready to put to work. The pre-training process is necessary, of course, but far from sufficient. Some even believe that pre-training may soon no longer be the most important part at all.
That’s because the post-training process is increasingly being shown to be where real value can be created. That’s where the model is molded from a giant, know-it-all network that will as readily produce Holocaust denial talking points as it will cookie recipes. You generally don’t want that!
Companies are secretive about their post-training regimens because, while everyone can scrape the web and make a model using state-of-the-art methods, making that model useful to, say, a therapist or research analyst is a completely different challenge.
AI2 (formerly known as the Allen Institute for AI) has spoken out about the lack of openness in ostensibly “open” AI projects, like Meta’s Llama. While the model is indeed free for anyone to use and tweak, the sources and process of making the raw model and the method of training it for general use remain carefully guarded secrets. It’s not bad — but it also isn’t really “open.”
AI2, on the other hand, is committed to being as open as it can possibly be, from exposing its data collection, curation, cleaning, and other pipelines to the exact training methods it used to produce LLMs like OLMo.
But the simple truth is that few developers have the chops to run their own LLMs to begin with, and even fewer can do post-training the way Meta, OpenAI, or Anthropic does — partly because they don’t know, but also because it’s technically complex and time-consuming.
Fortunately, AI2 wants to democratize this aspect of the AI ecosystem as well. That’s where Tulu 3 comes in. It’s a huge improvement over an earlier, more rudimentary post-training process (called, you guessed it, Tulu 2); in the nonprofit’s tests, this resulted in scores on par with the most advanced “open” models out there. It’s based on months of experimentation, reading, and interpreting what the big guys are hinting at, and lots of iterative training runs.
Basically, Tulu 3 covers everything from choosing which topics you want your model to care about — for instance, downplaying multilingual capabilities but dialing up math and coding — then takes it through a long regimen of data curation, reinforcement learning, fine tuning and preference tuning, plus tweaking a bunch of other meta-parameters and training processes that I couldn’t adequately describe to you. The result is, hopefully, a far more capable model focused on the skills you need it to have.
The real point, though, is taking one more toy out of the private companies’ toybox. Previously, if you wanted to build a custom-trained LLM, it was very hard to avoid using a major company’s resources one way or the other, or hiring a middleman who would do the work for you. That’s not only expensive, but it introduces risks that some companies are loath to take.
For instance, medical research and service companies: sure, you could use OpenAI’s API, or talk to Scale or whoever to customize an in-house model, but both of these involve outside companies in sensitive user data. If it’s unavoidable, you just have to bite the bullet — but if it isn’t? Like if, for instance, a research organization released a soup-to-nuts pre- and post-training regimen that you could implement on-premises? That may well be a better alternative.
AI2 is using this itself, which is the best endorsement one can give. Even though the test results its publishing today use Llama as a foundation model, they’re planning to put out an OLMo-based, Tulu-3-trained model soon that should offer even more improvements over the baseline and also be fully open source, tip to tail.
If you’re curious how the model performs currently, give the live demo a shot.
Keep reading the article on Tech Crunch
Crusoe, a rumored OpenAI data center supplier, has secured $686M in new funds, filing shows
Crusoe Energy, a startup building data centers reportedly to be leased Oracle, Microsoft, and OpenAI, is in the process of raising $818 million, according to an SEC filing.
The filing indicates that Crusoe has secured $686 million of the $818 million total that it hopes to raise. Seventy investors have contributed to the tranche so far, per the filing.
“A company at our stage of growth is always talking to investors,” a spokesperson for Crusoe told TechCrunch.
The Financial Times reported earlier this year that Crusoe was in talks to raise roughly $500 million in a funding round led by Peter Thiel’s Founders Fund with participation from Felicis Ventures. It seems investors had an appetite for a larger tranche, which presumably would come at a higher valuation than the rumored $3 billion — which was already double Crusoe’s previous valuation.
Should Crusoe successfully raise $818 million, it would bring the startup’s total raised to approximately $1.5 billion in equity and debt. Late last year, Crusoe secured $200 million in debt using its data center chips as collateral to buy thousands of AI processors.
Crusoe launched in 2018 as a cryptocurrency business, powering its data centers with natural gas that would otherwise be “flared off” and wasted. Like many crypto mining operations, Crusoe pivoted as AI rose to prominence, securing deals with AI companies to provide high-performance computing and AI infrastructure.
In early October, Crusoe announced it would enter into a $3.4 billion joint venture with asset manager Blue Owl Capital to build a massive data center in Abilene, Texas. The campus is expected to be leased to Oracle, which will in turn rent it to Microsoft and its close collaborator, OpenAI.
There’s a booming market for “neocloud” startups building low-cost, on-demand clouds for AI.
CoreWeave, the GPU infrastructure provider, has raised more than $12 million over a series of deals in the past ~18 months. Lambda Labs in early April secured a special purpose financing vehicle of up to $500 million. The nonprofit Voltage Park, backed by crypto billionaire Jed McCaleb, last October announced that it’s investing $500 million in GPU-backed data centers. And Together AI, a cloud GPU host that also conducts generative AI research, in March landed $106 million in a Salesforce-led round.
The environmental impact of the build-outs could be substantial. IDC expects global data center electricity consumption to more than double between 2023 and 2028. And according to Morgan Stanley, data center tech suppliers will create emissions equivalent to 2.5 billion metric tons of carbon dioxide by 2030.
Keep reading the article on Tech Crunch