Vital Details Regarding the US Prohibition on Routers Manufactured Abroad

Vital Details Regarding the US Prohibition on Routers Manufactured Abroad

In March, the Federal Communications Commission prohibited the introduction of new consumer internet routers manufactured outside the US, referencing national security issues. Current routers in American residences or presently available for purchase remain unaffected, but all newly crafted consumer routers require approval. Manufacturers can seek exceptions, and some have been authorized, meaning there’s no necessity to dispose of your existing router, and numerous mesh systems are still purchasable in stores.

Updated May 2026: Further details on software and component modifications, the inclusion of mobile hotspots in the ban, and Conditional Approval granted to certain firms are provided.

Why Are Routers Made Abroad Prohibited?

The FCC indicated that malicious entities have taken advantage of security vulnerabilities in routers made abroad to target American households, disrupt networks, facilitate espionage, and promote intellectual property infringement. Routers that are foreign-made played a role in the Volt, Flax, and Salt Typhoon cyberattacks, which focused on crucial US infrastructure. These routers are included on the Covered List, representing an unacceptable threat to US national security. As noted by Bogdan Botezatu, director of Threat Research at Bitdefender, the prohibition aims to bolster cybersecurity in US homes amidst geopolitical tensions, emphasizing that consumer routers represent a strategic risk if compromised on a broad scale. He states that Internet of Things devices, including routers, are a vulnerability across the internet.

Which Routers Are Prohibited?

The prohibition solely targets the sale of new Wi-Fi routers and mobile Wi-Fi or hotspot devices aimed at consumers, excluding existing FCC-approved routers or phones with hotspot functionalities. Previously acquired routers are not impacted and can be sold, utilized, and updated until March 1, 2027. Any new router produced outside the US must receive FCC approval before being brought in, marketed, or sold within the US, affecting US firms with overseas production.

What Does Foreign-Made Entail?

The prohibition pertains to “consumer-grade” routers designed or produced outside of the US or by companies not owned by US entities. Key players like Netgear, TP-Link, Asus, Amazon’s Eero, Google’s Nest, Synology, Linksys, and Ubiquiti fall into this category, as do most routers provided by internet service providers in the US. Manufacturers can request Conditional Approval from the Department of Defense and the Department of Homeland Security.

What is Conditional Approval?

To obtain Conditional Approval, companies are required to disclose their corporate structure, provide details on the manufacturing and supply chain, and outline a US manufacturing and onshoring strategy. This facilitates the ongoing sales and updates for both existing and new devices for a period of 18 months, which includes firmware updates, superseding the March 1, 2027, firmware waiver deadline. New devices will still undergo the standard FCC approval procedure.

Companies With Conditional Approval

Three companies have been granted Conditional Approval: Netgear, Adtran, and Eero, with permissions extending until October 2027. The Department of Defense, Homeland Security, and FCC have not clarified the reasons for these companies receiving approvals. The Consumer Technology Association (CTA) has requested clearer guidelines, raising concerns about software updates and component modifications.

What About Software Updates?

The FCC waiver permits all authorized routers to access software and firmware updates until March 1, 2027. After this date, it is uncertain what will occur should companies not secure Conditional Approval. Joshua Marpet from Finite State points out that routers that cease to receive updates present security hazards. Older devices frequently involved in cyberattacks typically do not receive further updates.

What About Components Made Abroad?

The international supply chain often faces component substitutions in routers, prompting worries. The CTA has advocated for additional waivers where security is not significantly compromised. The FCC clarified that a router manufactured in the US is not classified as ‘covered’ due to foreign components unless the component is a modular transmitter. Companies must demonstrate non-foreign production without precise regulations for documentation.

Netgear Has Approval

Netgear disclosed its Conditional Approval through a letter from the CEO. Although based in the US, its routers are produced in countries like Vietnam and Taiwan. The company has advocated for enhancing US cybersecurity and strategic competition with China. Netgear’s stock saw an increase following the announcement of the ban.

Will TP-Link Be Banned?

TP-Link, holding an estimated 35% of the US market share, must seek Conditional Approval or establish US manufacturing to market new routers. Investigations by US agencies and a lawsuit by the Texas attorney general highlight concerns regarding TP-Link’s connections to China. TP-Link asserts that it is headquartered in the US and manufactures in Vietnam, with its CEO applying for Trump’s Gold Card program for residency.

Will Asus Be Banned?

Asus mainly manufactures routers in Taiwan and China and must pursue Conditional Approval for new router sales. The ban has not influenced its stock price on the Taiwanese Stock Exchange.

Are Any Routers Manufactured in the US?

Some Starlink Wi-Fi routers, part of Elon Musk’s SpaceX operations, are created in Texas, although many components originate from East Asia. The security framework is considered more crucial than the manufacturing site.

How Will the Router Ban Impact Ordinary Folks?

The immediate consequence may be minimal, with numerous

AI-created performers and screenplays are no longer qualified for Oscars

AI-created performers and screenplays are no longer qualified for Oscars

On Friday, the entity responsible for the Academy Awards unveiled new regulations regarding the Oscars, which include several stipulations related to the application of generative artificial intelligence.

The Academy of Motion Picture Arts and Sciences announced that only performances “acknowledged in the film’s official billing and clearly executed by humans with their agreement” will qualify for Academy Awards. Likewise, the academy stated that screenplays must be “authored by humans” to meet eligibility criteria.

The academy further asserted its authority to demand additional information concerning a film’s usage of AI and “human authorship.”

These regulatory updates coincide with an independent film project featuring an AI-generated depiction of Val Kilmer, as AI “actress” Tilly Norwood continues to attract attention, and as emerging video models prompt some filmmakers to express considerable dismay. AI was also a significant issue during the actors’ and writers’ strikes in 2023.

Beyond Hollywood, at least one novel has been retracted by its publisher due to its suspected use of AI, and various writers’ organizations are asserting that AI involvement disqualifies works from receiving awards.

Goodbye, Jeeves: Ask.com ceases operations

Goodbye, Jeeves: Ask.com ceases operations

Ask.com, the search engine and Q&A platform previously known as Ask Jeeves, has ceased operations.

Launched in 1996, Ask Jeeves focused on responding to conversational inquiries made in natural language, making it a forerunner to modern AI-driven chatbots. Despite its 30-year existence, it has largely been eclipsed by other search engines, particularly Google.

The parent company IAC purchased Ask Jeeves in 2005, promptly removing “Jeeves” from its title, and by 2010 had diminished its search offerings to concentrate on Q&A. That same year, IAC Chairman Barry Diller remarked at TechCrunch Disrupt that Ask.com was not competitive with Google and lacked value in IAC’s stock.

A notice on the Ask.com site currently states, “As IAC continues to refine its focus, we have made the decision to cease our search operations, which includes Ask.com. After 25 years of addressing the world’s inquiries, Ask.com officially shut down on May 1, 2026.”

However, the website maintains, “The spirit of Jeeves lives on.”

Tovala Family Meals Evaluation: Delicious Selections, Elevated Sodium Content

Tovala Family Meals Evaluation: Delicious Selections, Elevated Sodium Content

A garlic-herb salmon served with risotto emerged as the top choice among the family dishes I tried. While the chopped asparagus appeared less appealing when topped with garlic butter, it was still flavorful and slightly crisp. The salmon was soft and flaky, and the sweet pea risotto was delightful due to the ample amounts of cheese, butter, and lemon, creating a rich blend of flavors.

The chicken parm was equally rich, evoking memories of family meals from my childhood during the 1980s, featuring cheese-covered chicken, garlic bread, and ricotta-stuffed shells. The main difference was the lack of vegetables, which my mother would have certainly included.

However, Toval’s meals face a nutritional challenge. For example, the chicken parm dish has 2,300 milligrams of sodium per serving, almost the entire daily limit for an adult. This sodium content is similar to comparable servings of Stouffer’s meat lasagna, but the Tovala meal also contains roughly ten times the cholesterol amount.

Other dishes similarly relied heavily on fats and salt for flavor, resembling indulgent restaurant fare rather than home-cooked meals. Whether this is a positive aspect is subjective.

Only one out of the seven meals I sampled was a total letdown: the teriyaki chicken dinner, which felt like a cultural misrepresentation of Japan. The dish featured sweet soy sauce over bland chicken, heavy egg rolls, and plain broccoli, reminiscent of ’90s mall food court “Japanese” cuisine. However, such a significant error was rare.

A larger issue with family-oriented meals is the consistently elevated levels of fat, cholesterol, and sodium. Those who have the means and desire for convenient, hearty meals—such as parents with children or seniors—might look for more nutritious alternatives with reduced fat and salt.

After a few weeks of experimenting with recipes, I was relieved to notice my arteries feeling clearer. Tovala’s culinary strategy effectively bridges the divide between prepared meals and fresh food, showcasing their culinary prowess. It would be advantageous if they provided options that encourage a more sustainable diet.

Netflix postpones Greta Gerwig’s ‘Narnia’ film to 2027 for significant theatrical launch.

Netflix postpones Greta Gerwig’s ‘Narnia’ film to 2027 for significant theatrical launch.

Viewers will need to wait a few more months to catch “Narnia: The Magician’s Nephew,” as the release date has been postponed from Thanksgiving to February 12, 2027.

Alongside the reintroduction of “Narnia” on the big screen and marking writer-director Greta Gerwig’s first project since “Barbie,” “The Magician’s Nephew” appears to signify a progressive step in Netflix’s partnership with cinemas — and this step is growing larger due to the delay.

Netflix had initially announced that “The Magician’s Nephew” would debut exclusively on Imax screens for a minimum of two weeks prior to its streaming debut for Christmas. While this would have been a notable theatrical rollout by Netflix’s criteria, it pales in comparison to numerous other Hollywood blockbusters.

Currently, Netflix reports that “The Magician’s Nephew” will kick off exclusive Imax previews on February 10, 2027, followed by a broad global release in theaters on February 12. (In Netflix’s terms, it will be a “global eventized release.”) The film’s streaming will not commence until April 2.

The company’s announcement does not specify which theaters will screen “The Magician’s Nephew,” but Imax released a statement mentioning that the delay will enable the film to enjoy “a complete theatrical window,” thus reducing the likelihood of complaints from major theater chains.

Indeed, AMC Theatres has recently emphasized the success of its “Stranger Things” finale presentations and indicated its intention for further collaborations with Netflix. Simultaneously, the limited support from the streamer for theatrical releases and its reluctance to agree to exclusive theatrical windows was reportedly a “dealbreaker” during discussions with the creators of “Stranger Things,” who ultimately opted for an exclusive agreement with Paramount.

Featuring a cast that includes Daniel Craig and Meryl Streep, “The Magician’s Nephew” adapts one of the later installments in C.S. Lewis’ beloved fantasy series — a prequel that reveals the origins of Narnia.

Techcrunch event

San Francisco, CA
|
October 13-15, 2026

In their announcement, Gerwig expressed that she first encountered the book in her youth, where she “fell in love with the wonderfully improbable yet entirely brilliant idea of a cosmic lion bringing the world of Narnia into existence through song.”

When you make a purchase through links in our articles, we may receive a small commission. This does not impact our editorial independence.

The top AI dictation applications, evaluated and sorted

The top AI dictation applications, evaluated and sorted

AI dictation applications have made significant progress in a brief period. For many years, they were sluggish and imprecise — unless you spoke with a specific accent and articulated clearly.

Improvements in large language models (LLMs) and speech-to-text technologies have transformed this landscape, resulting in systems capable of understanding speech with greater accuracy while maintaining sufficient context to format the text appropriately. Additionally, developers have implemented features to automatically eliminate filler words, correct missteps, and manage punctuation — generating text that requires significantly fewer modifications.

With numerous such applications currently available on the market, we have compiled our selections for the most effective and practical dictation apps accessible at this time.

Wispr Flow

Wispr Flow is a well-financed AI dictation application that allows you to introduce custom words and directions for dictation. It has native applications for macOS, Windows, and iOS; an Android version is being developed.

The app enables you to tailor how it transcribes your content by selecting “formal,” “casual,” or “very casual” styles for various types of writing, including personal messaging, work, and email. Moreover, if you utilize it alongside vibe-coding tools like Cursor, you can activate a feature that automatically identifies variables or tags files in the chat.

You can transcribe up to 2,000 words weekly for free on the desktop version and 1,000 words monthly on iOS. Paid subscription options provide unlimited transcription and start at $15 monthly.

Image Credits:Wispr Flow

Willow

Willow promotes itself as a significant time-saver for those who prefer not to type. In addition to standard features like automatic editing and formatting, the app leverages large language models to create an entire passage from just a few dictated words.

Willow also adopts a more privacy-conscientious model by storing all transcriptions locally on your device and providing options to opt-out of model training entirely. It enables you to input custom vocabulary to help it adapt to the terminology specific to your field or your local dialect.

Image Credits:Willow

Willow allows you to dictate 2,000 words monthly on its desktop application for free. Personal subscription plans begin at $15 monthly, granting unlimited dictation and enabling the app to remember your writing style.

Monologue

If privacy is your main concern, Monologue provides the option to download its AI model directly to your device for transcriptions, keeping your data entirely off the cloud. Furthermore, the app permits customization of its tone based on the application you are using it with.

Monologue allows for 1,000 words to be transcribed monthly for free; a subscription is priced at $10 monthly or $100 annually. Additionally, the company provides its most engaged users with a physical shortcut device called the Monokey to utilize with the app.

Superwhisper

Superwhisper chiefly functions as a dictation app, yet it can also transcribe from audio or video files. The app allows you to select and download AI models, including several of its own at various speeds and accuracy levels, as well as Nvidia’s Parakeet speech-recognition models.

Additionally, the app lets you create custom prompts to guide the output, and you can access both processed and unprocessed transcripts directly from your system keyboard.

[embedded content]

The basic voice-to-text functionality is available free, and you receive 15 minutes to evaluate Pro features such as translation and transcription. The paid version allows you to use your own AI API keys and connect both cloud and local models without any usage restrictions.

The monthly subscription is priced at $8.49, the annual plan costs $84.99 per month, or you can opt for a lifetime subscription at $249.99.

VoiceTypr

The VoiceTypr application adopts an offline-first, no-subscription model, allowing you to use local models for transcription. There is also a GitHub repository for those interested in hosting and running the open-source version themselves. VoiceTypr supports over 99 languages and functions on both Mac and Windows platforms.

The app can be trialed for three days at no cost, after which you may purchase a lifetime license. The pricing is $35 for a single device, $56 for two devices, and $98 for four devices.

Aqua

Aqua is a Y Combinator-supported voice-typing application for Windows and macOS that claims to be one of the swiftest tools in terms of latency (the delay between speaking and text appearing on the screen).

In addition to managing grammar and punctuation, Aqua enables you to autofill text by simply saying phrases — for instance, saying “my address” will have Aqua type it out.

The app also provides its own speech-to-text API, allowing other applications to integrate with Aqua’s transcription engine.

[embedded content]

The free tier allows for 1,000 words monthly. Paid subscriptions start at $8 per month billed annually, unlocking unlimited words and 800 custom dictionary entries.

Handy

Handy is an open-source, complimentary transcription tool compatible with Mac, Windows, and Linux. The application is relatively straightforward and lacks extensive customization options, but it is a viable choice for those looking to start voice dictation without incurring costs.

The app features a simple settings menu that allows for toggling push-to-talk and changing the hotkey to initiate transcription.

Typeless

Typeless is notable for its high free word count. The company asserts that it does not retain any data or employ it for training AI models. Typeless also provides the option to rephrase sentences that you may have mismanaged.

The app permits dictation of up to 4,000 words weekly (approximately 16,000 words per month) within its free tier. You can subscribe for $12 monthly (billed annually) to unlock unlimited dictation and gain access to new features. Typeless is only available for Windows and macOS.

VoiceInk

VoiceInk is an open-source private dictation application for Mac. The app features global shortcuts for starting and stopping recordings, as well as a push-to-talk mode. It analyzes the context on screen and adjusts its output accordingly.

The app can automatically identify certain applications and URLs, applying custom formatting or rules to each one. It also includes an assistant mode capable of responding to your inquiries. The pricing is $25 for lifetime access for one device, $39 for two devices, and $49 for three devices.

Dictato

Dictato is a dictionary application for Mac priced at €9.99 — roughly $12 — which grants you lifetime access along with two years of feature updates. The app operates with offline models such as Parakeet, Whisper, and Apple Speech Analyzer, utilizing Apple Intelligence for light reading and filler word elimination. Thanks to these local models, the app boasts an impressive 80ms latency, indicating that text appears almost instantaneously after your speech.

AudioPen

AudioPen originated as a web-based voice notes tool, but it has grown over time. The Mac version now allows you to dictate text and modify it in your desired format and style, switching between various styles at will. Beyond live transcription, AudioPen facilitates the storage of audio notes across platforms, the combining of notes for summaries, uploading audio files, and rewriting existing notes using AI. The pricing is $33 for three months, $99 for a year, and $159 for two years.

[embedded content]

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Beyond Lovable and Mistral: 21 European startups to keep an eye on

Beyond Lovable and Mistral: 21 European startups to keep an eye on

BottleCap AI should be the highlight of Europe, rather than bottle cap memes. Sporting a playful name, this AI startup located in Prague is among the firms that venture capitalists recommend you should take notice of.

It’s not that European startups fail to stand out — companies like Lovable and Mistral AI demonstrate otherwise. Yet, numerous others still lack nine-digit annual recurring revenue and are closely monitored by insiders.

This is where our list becomes relevant. In recent weeks, we solicited advice from investors at some of Europe’s most prominent venture funds to suggest two startups each: one that they have invested in and another outside their portfolio, as these experts specialize in startups but cannot back every one. We included a few of our selections as well.

Ranging from pre-launch to unicorn status, these startups vary in their developmental phases and industries. While our selection process might not showcase where the region’s top hubs are located, it does highlight how deep tech talent could enable Europe to leverage its strengths in the AI competition.

Alta Ares

Recommended by Julien Codorniou, general partner, 20VC.

What it does: Alta Ares creates AI-driven counter-drone technology.

Why it’s worth watching: The defense sector has shifted from being marginalized to becoming a focal point, especially in Europe, where the Ukraine conflict prompted military modernization. Alta Ares’ interceptors fulfill the demand for more affordable solutions to monitor and counter drone threats.

Techcrunch event

San Francisco, CA
|
October 13-15, 2026

Apron 

Recommended by Jan Hammer, partner, Index Ventures (investor).

What it does: Apron offers invoice management solutions tailored for small business proprietors.

Why it’s worth watching: Small and medium-sized businesses represent a potentially profitable market for fintech firms; owners are prepared to invest at least some of their funds to save time, and they number in the millions.

Botify 

Recommended by Claire Houry, general partner, Ventech (investor).

What it does: Botify assists brands in enhancing their visibility in AI-driven searches.

Why it’s worth watching: Businesses continue to scramble to transition from SEO to generative engine optimization (GEO) — yet this Disrupt NY 2016 participant has already welcomed the change. Botify faces competition in its emerging field, including rivals like Otterly.AI and Profound, but also boasts major clients such as Macy’s and The New York Times.

BottleCap AI

Recommended by Julien Codorniou, general partner, 20VC (investor).

What it does: BottleCap AI builds efficiency-centered foundational LLMs and applications.

Why it’s worth watching: Featuring a founding team that includes an entrepreneur who sold his previous business to Meta and two researchers in AI, BottleCap embraces a bifocal strategy. The company is not only building its own models but also developing applications based on them, such as Pulse, an AI-driven news application.

Cailabs

Recommended by Flavia Levi, investment manager, Join Capital.

What it does: Cailabs specializes in photonics tailored for aerospace, defense, and industrial uses.

Why it’s worth watching: Rooted in sophisticated research concerning light science, Cailabs now applies its findings to enhance the speed and reliability of data transmission. Supported by both public and private investors, it aims to install 50 optical ground stations to meet the rising demand for laser communication with satellites.

Cailabs' turnkey optical ground station
Cailabs’ turnkey optical ground station.Image Credits:Cailabs

Cala

Recommended by TechCrunch’s Anna Heim.

What it does: Creates knowledge graphs for AI agents.

Why it’s worth watching: Cala aims to develop the knowledge infrastructure that AI agents currently lack. Its founder, Elisenda Bou-Balust, is a notable Spanish entrepreneur and AI specialist who sold her previous venture, Vilynx, to Apple in 2020. 

Flower

Recommended by Pär-Jörgen Pärson, partner, Northzone (investor).

What it does: Manages renewable energy resources.

Why it’s worth watching: As wind and solar energy inherently fluctuate, Flower utilizes AI and battery storage systems to enhance their predictability. The Swedish startup also recently raised over $60 million in bonds to facilitate its expansion.

Fundamental 

Recommended by Jonathan Userovici, general partner, Headline (investor).

What it does: AI foundation for extensive data analysis.

Why it’s worth watching: Fundamental’s foundation model, Nexus, is aimed at assisting enterprises in extracting insights from their data. The company just emerged from stealth mode in February but has already achieved a valuation of $1.4 billion following a $255 million Series A round. 

Gradium 

Recommended by Jonathan Userovici, general partner, Headline.

What it does: Voice models powered by AI.

Why it’s worth watching: Gradium’s AI technologies enable real-time text-to-speech functionality, granting AI agents multilingual capabilities. A spinout from the French AI lab Kyutai, this ElevenLabs competitor secured a $70 million seed round.

HappyRobot 

Recommended by Pablo Ventura, general partner, Kfund.

What it does: AI agents designed for intricate applications.

Why it’s worth watching: HappyRobot, a startup backed by a16z and Y Combinator, stands out among those developing AI agents by ensuring that these can be deployed effectively to deliver ROI. While based in the U.S., its three co-founders and part of its team are of Spanish origin.

Inbolt deployed
Inbolt AI robot in deployment.Image Credits:Inbolt

Inbolt 

Recommended by Claire Houry, general partner, Ventech.

What it does: Physical AI tailored for production facilities.

Why it’s worth watching: By merging AI with robotics, Inbolt enhances and broadens automation across manufacturing sectors, from automotive to electronics and home goods production lines. The startup reports active engagement in over 70 factories.

Legora

Recommended by Pär-Jörgen Pärson, partner, Northzone.

What it does: An AI platform designed for legal professionals.

Why it’s worth watching: With rising competition from mainstream LLMs, the legal tech landscape will increasingly rely on marketing strategies. Anticipate a showdown between Harvey and Legora after Legora secured a lead over its competitor by recruiting Jude Law as its brand ambassador. That’s a win for the Swedish-born startup, now based in New York but still emerging as one of Stockholm’s top AI talents.

Macrodata Labs 

Recommended by Floriane de Maupeou, principal, Serena Data Ventures.

What it does: Infrastructure for AI training data.

Why it’s worth watching: “Every formidable model begins with exceptional data,” Macrodata Labs asserts on its “coming soon” page. Rather than create that data itself, the startup plans to offer tooling for other businesses to develop high-quality training datasets. 

Multiverse Computing

Recommended by TechCrunch’s Julie Bort.

What it does: Provides compressed variants of publicly available weight models such as OpenAI, Meta, DeepSeek, and Mistral AI.

Why it’s worth watching: Multiverse Computing’s technology takes an established model and reduces its size and operating costs, particularly for use on a company’s hardware. Co-founded by CTO Román Orús, a professor at the Donostia International Physics Center, this Spanish company has raised $250 million.

Optics11

Recommended by Flavia Levi, investment manager, Join Capital (investor).

What it does: Systems for fiber-optic sensing.

Why it’s worth watching: Optics11’s innovation enables monitoring of equipment in underwater environments and other challenging conditions. Its capability to avert disruptions to subsea infrastructure and energy systems helped the startup obtain venture debt support from the European Investment Bank.

Pennylane

Recommended by Jan Hammer, partner, Index Ventures.

What it does: A platform for financial management aimed at SMBs.

Why it’s worth watching: Pennylane initially focused on accounting, but the company has broader ambitions. Like many growth-stage fintech startups, this French unicorn seeks to construct a unified financial operations system specifically for SMBs throughout Europe.

PLD Space

Recommended by TechCrunch’s Anna Heim.

What it does: Develops rocket launch technologies.

Why it’s worth watching: As part of Europe’s initiative for space independence, PLD Space recently achieved a successful launch of a suborbital rocket in 2023 and is now creating a reusable orbital launch system for small satellites. Last month, the Spanish company secured $209 million in a Series C funding round led by Mitsubishi Electric, bringing its total funding beyond $350 million.

PLD Space's MIURA 1 space rocket during its presentation in Madrid in 2021.
PLD Space’s MIURA 1 space rocket during its presentation in Madrid in 2021.Image Credits:Eduardo Parra / Europa Press via Getty Images / Getty Images

Proxima Fusion

Recommended by Daria Saharova, general partner, World Fund.

What it does: Nuclear fusion technology.

Why it’s worth watching: The quest for a nuclear fission alternative is intensifying, and Proxima Fusion is among Europe’s leading competitors. This VC-backed company recently gained $460 million in funding from the Bavarian state to facilitate its plans for a fusion power plant in Europe, beginning with a demonstration stellarator in the Munich region.

Roofline 

Recommended by Floriane de Maupeou, principal, Serena Data Ventures (investor).

What it does: Software for deploying AI models on advanced chipsets.

Why it’s worth watching: This university spinout, Roofline, effectively connects AI with an increasingly diversified hardware landscape through its software that enables users to deploy models seamlessly across various chip types.

Space Forge

Recommended by Daria Saharova, general partner, World Fund (investor).

What it does: Manufactures semiconductor components in outer space.

Why it’s worth watching: The trend of in-space manufacturing is gaining momentum — particularly for pharmaceuticals and chips, which are the focus of Space Forge. With additional support stemming from geopolitical factors, the startup is making swift advances: it has recently produced plasma in low Earth orbit.

Theker 

Recommended by Pablo Ventura, general partner, Kfund (investor).

What it does: Robotics offered as a service.

Why it’s worth watching: Theker is among several startups funded by Zara owner Inditex through a dedicated initiative overseen by Mundi Ventures. Theker’s AI-powered robots could assist the retail giant in enhancing its logistics operations, while also exploring applications in waste management and the food and beverage sectors.

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Uber aims to transform its vast network of drivers into a sensor grid for autonomous vehicle firms.

Uber aims to transform its vast network of drivers into a sensor grid for autonomous vehicle firms.

Uber has a far-reaching goal that extends well past transporting riders: the firm ultimately intends to equip its drivers’ vehicles with sensors to gather real-world information for autonomous vehicle (AV) enterprises — and possibly other businesses training AI models in real-world contexts.

In a discussion at TechCrunch’s StrictlyVC event in San Francisco on Thursday night, Praveen Neppalli Naga, Uber’s chief technology officer, unveiled the initiative, characterizing it as a logical progression of a fledgling program the company introduced in late January known as AV Labs.

“That is the route we hope to pursue eventually,” Naga mentioned regarding outfitting human drivers’ automobiles. “But first, we must comprehend the sensor kits and their operations. There are certain regulations — we need to ensure every state has [clarity on] what sensors signify, and what sharing entails.”

At this time, AV Labs depends on a specialized, small fleet of sensor-laden vehicles that Uber manages independently, apart from its driver network. However, the vision is evidently much broader. Uber has millions of drivers worldwide, and if merely a small percentage of those automobiles could be converted into moving data-collection units, the extent of what Uber could provide the AV sector would surpass what any single AV company could create on its own.

The insight propelling the initiative, Naga stated, is that the critical limitation for AV progress is no longer the foundational technology. “The bottleneck is data,” he noted. “[Companies like Waymo] must gather data, capturing various scenarios. You might be able to specify: in San Francisco, ‘At this school intersection, I need data at this particular time of day to train my models.’ The challenge for all these companies is accessing that data, as they lack the capital to deploy vehicles and gather all this information.”

Becoming the data layer for the entire AV landscape is a rather clever strategy, especially considering Uber years ago set aside its aspirations to create self-driving cars (a decision that co-founder Travis Kalanick has openly regretted as a significant error). Indeed, many industry watchers have speculated that, absent its own self-driving vehicles, Uber could potentially become irrelevant as AVs proliferate worldwide.

The company presently collaborates with 25 AV enterprises — including Wayve, which operates in London — and is developing what Naga termed an “AV cloud”: a repository of labeled sensor data that partner firms can access and utilize to train their models. Partners, which Uber intends to invest in more actively, can also leverage the system to test their trained models in “shadow mode” against actual Uber journeys, simulating how an AV would have functioned without actually deploying one on the roads.

Techcrunch event

San Francisco, CA
|
October 13-15, 2026

“Our objective is not to profit from this data,” Naga remarked. “We aspire to make it accessible to everyone.”

Considering the evident commercial potential of what Uber is developing, that stance may not be sustainable for long. The company has already invested equity in several AV players, and its capability to provide proprietary training data at scale could grant it considerable influence over a sector that currently relies on Uber’s ride marketplace to reach clients.

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Replit’s Amjad Masad discusses the Cursor agreement, taking on Apple, and his preference against selling.

Replit’s Amjad Masad discusses the Cursor agreement, taking on Apple, and his preference against selling.

Amjad Masad has been developing Replit for ten years, but the past 18 months have been truly extraordinary. The AI coding assistant firm escalated from $2.8 million in revenue throughout 2024 to moving toward what Masad describes as a billion-dollar annual run rate.

At TechCrunch’s sold-out StrictlyVC event in San Francisco on Thursday evening, we explored a variety of topics quickly, starting with the pertinent question on everyone’s mind: in an environment where competitor Cursor is allegedly negotiating an acquisition by SpaceX for $60 billion, is Replit also destined for a sale?

We also discussed Replit’s net revenue retention — an indicator of how much current clients increase their purchases — which Masad claims has reached as high as 300%, his readiness to take Apple to court over what he terms blatant falsehoods in its App Store conflict with Replit, and the potential for the company to begin investing in its clients.

On the topic of independence, Masad was resolute. Unlike Cursor, which he claimed has been operating at negative 23% gross margins, he contended that Replit possesses the economic framework to make that journey feasible — even though he stopped short of completely dismissing the idea of a sale.

The following has been condensed for brevity and clarity:

TC: Cursor’s rumored SpaceX deal was the buzz of the industry last week. What are your thoughts on it?

AM: It’s quite challenging being an independent, smaller AI firm that’s building on foundational models, especially if you’re spending a lot of money. Part of the reports indicated Cursor has negative 23% margins, and if you’re also looking to invest in training models, it becomes incredibly tough to maintain independence.

Techcrunch event

San Francisco, CA
|
October 13-15, 2026

For us at Replit, partly due to our focus on a different customer demographic, we’ve managed to operate the business more logically. We’ve been gross margin positive for over a year. We’re somewhat pricier, but we offer significantly more. Our target audience mainly consists of non-technical users who previously lacked the ability to create software. We provide a comprehensive platform — from the prompt all the way to a deployed application that can scale. We take care of security, databases, and database migration. And we’ve been doing this long enough to have integrated many of those essentials into the platform.

Is Replit for sale? I would think you are consulting with potential buyers frequently; it’s your fiduciary duty.

Indeed. We have fantastic partners, and they occasionally raise these subjects. However, we aim to maintain our independence. I would love for us to continue as an independent entity. We’ve existed for a decade, long before it was acknowledged that you could develop apps solely from concepts. We discussed creating a billion software creators back in 2018 at YC, and some people actually laughed at that vision. Now that aspiration is attainable, and we sparked this revolution with our agentic coding experience in September 2024. It genuinely feels like we can advance it far beyond.

You collaborate closely with Anthropic, Google, and OpenAI. If you had to rank them — who’s performing the best?

Anthropic remains unmatched on the core agentic loop. They have the finest tool calling; their agent can maintain coherence for much longer. GPT-5 is catching up quickly. Google’s Flash family of models offers incredible price-performance. If you need something quick and economical, they are currently outperforming open source. We utilize all three, and honestly, I wouldn’t disregard the newer labs either. Reflection AI is launching open-source models we’re hearing positive feedback about. And the Chinese models are impressive — Kimi is on par with an Anthropic-generation model from January, making it only about three months behind.

When you’re competing for an enterprise deal, what gives you the upper hand?

Most of our sales come from organic or inbound leads — heavily product-driven. We’ve secured clients like Zillow and Meta solely through users adopting the product and subsequently requesting an enterprise plan. When it goes top-down and there’s a formal competition, we typically succeed because of our product. But even in instances where we might lack a specific feature, when it reaches the C-suite and the IT department, Replit excels in security. Many vibe-coding tools will create a website and link it to an external database — great solutions, but this complicates security significantly since the database is publicly accessible and requires configuring row-level security, which is especially challenging for non-technical builders. Replit, being full stack, with the database incorporated into the project and not open to the public — that enhances the app’s inherent security.

We’ve also spent a decade battling crypto fraudsters and hackers, so our cybersecurity function rivals that of a specialized cybersecurity startup. Each time you deploy an app on Replit, we generate an entirely new isolated project on Google Cloud. We adopt Google’s security framework.

Can we discuss churn? How long do you retain customers if the best prototypes ultimately get integrated into a company’s existing stack?

Churn is exceptionally low, and net retention is remarkably high — 300% in some cases. What we often hear from clients is that when engineers get anxious and attempt to transfer an app into their own stack, they frequently worsen it. Once enterprises become comfortable with the full Replit stack — especially when we establish a single-tenant environment for them — they retain their apps on Replit. For example, Bain & Company swapped out Tableau and Power BI for Replit and Databricks.

There’s rising concern about AI bloat — non-technical users create significantly more code and consume far more tokens. This benefits you [given your usage-based fees]. What about your clients?

We don’t encounter much regrettable expenditure. Enterprises are very conscious of ROI, and they inform us about the returns they are experiencing. Generally, they feel the investment is entirely justified — often yielding one, two, three orders of magnitude. If they spend $100,000 a month with Replit, they typically generate $2 million, $3 million, $10 million in returns.

Let’s discuss Apple. Another competitor, Lovable, just received approval for an app-building application from the App Store this week. Replit has been in App Store purgatory, with Apple blocking your updates for months. How detrimental is that for you?

It’s not a matter of life or death — we could lose the app, and it wouldn’t significantly impact our business. However, it’s an application that people genuinely appreciate. We’ve been on the App Store for four years. Kids in disadvantaged areas learn coding on Replit via their Android devices. Executives utilize it during meetings.

The reason Replit faced restrictions when others didn’t, in our view, is that Replit creates iOS apps. When we introduced that feature in December, charts circulated showing how many apps were being accepted into the App Store through us. We believe Apple perceives that as a threat.

Apple’s stated reason is that you download new code to the device [post-approval], which breaches their guidelines.

That’s untrue. And we can substantiate it in court if necessary.

Is that a possibility?

I hope not. I’m an admirer of Apple, and I would love to partner and create something remarkable together. We’re pleased to direct customers toward Xcode [Apple’s own development environment]. But you cannot manage a marketplace accessible to a billion users and make decisions that are discriminatory or based on whims.

Just curious if, like Nvidia, OpenAI, and others, you’re contemplating investing in your clients in exchange for equity.

We’ve given it considerable thought, and it is on our minds. I’ve personally invested in a few startups that originated on Replit before they earned any revenue. Some of these, like Magic School — a teacher decided to dedicate his time during COVID to learn a bit of vibe coding and developed an AI app for fellow educators. He identified an issue where in America, we stress out a lot of teachers. He aimed to use AI to alleviate the workload. He achieved that, generating $20 million in his first year. Other businesses that began on Replit, I believe, are now valued at half a billion dollars. The entrepreneurial spirit on Replit right now is genuinely thrilling. We integrated with Stripe a few months ago, and the transactions flowing through Replit are seeing triple-digit growth month over month. Before long, our customers might be generating more revenue than we do.

You can watch our entire discussion with Masad below:

[embedded content]

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Musely obtains $360M from General Catalyst while retaining full ownership.

Musely obtains $360M from General Catalyst while retaining full ownership.

Musely, a telemedicine platform providing services directly to consumers, has secured over $360 million in non-dilutive funding from General Catalyst’s Customer Value Fund (CVF).

The business focuses on customized treatments for skin, hair, and menopause. Musely co-founder and CEO Jack Jia mentioned to TechCrunch that when CVF investors contacted him last year, he wasn’t seeking to raise funds.

This is due to Musely, which started in 2014 as a wellness platform before transitioning to prescription skincare in 2019, being cash flow positive for several years, he stated. Jia was reluctant to dilute his ownership by selling a part of the company to venture capitalists. He consistently declined their offers for potential funding rounds, he noted.

However, unlike conventional venture capital, CVF did not intend to obtain equity ownership, nor was it providing a loan with interest charges. Instead, CVF’s unique funding model resembles a small revenue-share agreement: businesses with stable revenue streams borrow funds and then repay the amount along with a capped, fixed percentage of the revenue generated from using General Catalyst’s funds.

Initially doubtful, Jia soon recognized CVF’s terms were more advantageous than a traditional bank loan and significantly less expensive than a dilutive equity round.

“When I calculated it mathematically, I found it incredibly attractive,” he stated.

Although Musely has been increasing its revenue by an average of 50% annually and has assisted over 1.2 million patients, Jia explained acquiring new customers for direct-to-consumer brands like Musely can be quite expensive. “Once you reach a billion-dollar revenue status, you require another billion to achieve the next billion,” he said. “This is why many DTC companies have massive capital burn.”

Techcrunch event

San Francisco, CA
|
October 13-15, 2026

The investment from CVF addresses this challenge, equipping Musely with a financial reserve to enhance its customer acquisition efforts. This funding will facilitate sales, marketing, and other initiatives aimed at attracting new customers.

Musely is now part of a CVF portfolio that features Grammarly, Lemonade, and Ro. The fund has its own unique limited partners, and the resources it deploys were not included in General Catalyst’s recent $8 billion fundraising.

In contrast to many competitors, Musely has demonstrated remarkable capital efficiency. Following a $20 million raise from DCM and other investors in 2014, the company has not sought any additional equity capital since, according to Jia. Musely enables patients to access prescription products via asynchronous consultations with board-certified dermatologists and OB-GYNs.

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.