Saturday 30 April 2022

This Week in Apps: Elon buys Twitter, Snap Summit recap and an App Store cleanup

This Week in Apps: Elon buys Twitter, Snap Summit recap and an App Store cleanup

Welcome back to This Week in Apps, the weekly TechCrunch series that recaps the latest in mobile OS news, mobile applications and the overall app economy.

The app industry continues to grow, with a record number of downloads and consumer spending across both the iOS and Google Play stores combined in 2021, according to the latest year-end reports. Global spending across iOS, Google Play and third-party Android app stores in China grew 19% in 2021 to reach $170 billion. Downloads of apps also grew by 5%, reaching 230 billion in 2021, and mobile ad spend grew 23% year over year to reach $295 billion.

Today’s consumers now spend more time in apps than ever before — even topping the time they spend watching TV, in some cases. The average American watches 3.1 hours of TV per day, for example, but in 2021, they spent 4.1 hours on their mobile device. And they’re not even the world’s heaviest mobile users. In markets like Brazil, Indonesia and South Korea, users surpassed five hours per day in mobile apps in 2021.

Apps aren’t just a way to pass idle hours, either. They can grow to become huge businesses. In 2021, 233 apps and games generated over $100 million in consumer spend, and 13 topped $1 billion in revenue. This was up 20% from 2020, when 193 apps and games topped $100 million in annual consumer spend, and just eight apps topped $1 billion.

This Week in Apps offers a way to keep up with this fast-moving industry in one place, with the latest from the world of apps, including news, updates, startup fundings, mergers and acquisitions, and suggestions about new apps to try, too.

Do you want This Week in Apps in your inbox every Saturday? Sign up here: techcrunch.com/newsletters

Top Story

A billionaire buys himself a social network

Can you believe it’s only been a week since Elon Musk announced he was buying Twitter for around $44 billion? It’s felt like years!

A lot has happened since Elon Musk first signaled his interest in Twitter by snatching up Twitter shares, then later being offered a board seat, declining the seat, then deciding he’d rather just take the whole company. Initially, no one was quite sure how serious Musk’s offer was, but when he rounded up financing and detailed how he planned to pay for Twitter, the offer had to be given a lot more consideration. On April 25, Twitter accepted Musk’s offer, which includes a $1 billion termination fee on both sides. The deal is a go.

There’s a lot of curiosity over why Musk wanted Twitter in the first place. But it’s likely a combination of a power user thinking they can fix the service and a desire to use the network for the market-moving power it’s been shown to have.

Now everyone’s wondering what happens next. Twitter was never a great business in Wall Street’s eyes, so going private is not the worst choice for the company to make. But going private under a free speech absolutionist already has advertisers wary. If Twitter were to lighten its content moderation rules, it could allow more online abuse and hate speech to thrive. AdAge reported the immediate reaction from advertisers was one of anxiety and confusion. Brands began reaching out to agencies to help them understand and prepare, it said. One agency exec said advertisers are preparing to stop spending after Musk’s takeover if things go south. Looking to quell worries, Twitter emailed reassurances to advertisers, the FT reported. But brands know Twitter can’t make any promises about the nature of free speech on Twitter once Musk is in charge.

Advertisers can pull out of Twitter if need be — there are a number of other social networks hungry for their dollars beyond Meta. Snap and TikTok, for instance, could benefit from a potential ad budget shift, as they also reach a younger demographic and have growing user bases. While Musk has ideas about how to grow Twitter’s revenues in other ways in the future, Twitter’s business today is advertising-dependent. To what extent Musk understands the nuances of that complication is less clear. But unless the billionaire wants to self-fund Twitter, he should probably give it some thought.

Weekly News

Global app store revenues remained flat in Q1

Global consumer spending in apps saw relatively flat growth year-over-year, according to new data from Sensor Tower. The company found that worldwide app revenue growth from in-app purchases, premium apps and subscriptions grew just 0.6% from $32.3 billion in Q1 2021 to $32.5 billion in Q1 2022. However, when looked at individually, the App Store and Google Play saw different trends. Google’s Play Store lack of growth pulled the combined growth rate down, as it saw approximately $10.7 billion in consumer spending, down 8.5% year-over-year from $11.7 billion in Q1 2021. Meanwhile, Apple’s App Store revenue, which was double that of Google Play’s, grew 5.8% year-over-year from $20.6 billion to $21.8 billion.

Image Credits: Sensor Tower

Top grossing apps in the quarter included TikTok (iOS) and Google One (Android), as well as streamers like YouTube, Disney+, Tencent Video, HBO Max, and others. Dating app Tinder was also the No. 5 top grossing app overall.

Image Credits: Sensor Tower

Among app categories seeing increased usage in Q1, medical apps led the market with 102% year-over-year growth, followed by navigation (+24%), travel (+19%), business (+15%), shopping (+14%), finance (+13%) and education (+13%)

Snap Summit Recap

Snap held its Partner Summit this week, where the company announced a number of new features in areas like AR Shopping, 3D asset creation, AR development tools and more. It also announced a drone for taking photos, but that’s not really an app!

Among the highlights:

  • Lens Cloud: Snap launched another component for its developer platform, Lens Cloud. This server-side component will help developers build dynamic, multiplayer experiences. Lens Cloud allows developers to take advantage of multiuser services so groups of friends can interact with the same Lens at the same time; location-based services allow developers to anchor their Lenses to places, starting in central London; and it will offer developers the ability to store assets on Snap’s servers and load them up on demand, allowing developers to expand beyond 8 megabytes of storage.
  • AR Shopping: Snapchat added a new in-app destination within the app called “Dress Up” that will feature AR fashion and virtual try-on experiences and said it will now allow retailers to integrate with Snapchat’s AR shopping technology within their own websites and apps, via CameraKit. Retailers will also gain access to a new AR Image Processing technology in Snap’s 3D asset manager, which Snap says will make it easier and faster to build augmented reality shopping experiences. To use it, brands will be able to select their product SKUs and then turn them into Shopping Lenses by first uploading their existing product photography for the SKUs they sell. Snap’s tech will then process these using a deep-learning module that turns them into AR Image assets.

Image Credits: Snap

  • Director Mode: Snapchat will add a new feature called Director Mode, aimed at providing easier access to Snapchat’s native creative tools for publishing video on its platform, including for its short-form video feature known as Spotlight. Director Mode will include access to tools like greenscreen tech, Quick Edit and a new Dual Camera feature that will allow users to record both the front-facing and back-facing cameras at the same time.
  • Live Nation + Snap: Snap said it would partner with Live Nation to launch AR experiences at select concerts and festivals, including Electric Daisy Carnival in Las Vegas, Lollapalooza in Chicago, Wireless Festival in London, Rolling Loud in Miami and The Governors Ball in New York.

Apps and earnings

It was a busy week for tech company earnings. Apple had a stellar quarter, with revenue increasing 9% to $97.3 billion, and quarterly profit growing 6% to $25 billion, but shares slipped on warnings of possible supply chain constraints impacting the business in the future. But the standout news for the app economy was the record revenue reported by Apple’s services division, which includes the App Store and other subscription-based business lines, like Apple TV+, Apple Music, cloud services and more. The company said services revenue grew 17% year over year to reach $19.8 billion and it now has 825 million paid subscriptions. That means services is now bigger than Apple’s Mac ($10.44 billion) and iPad ($7.65 billion) divisions combined.

Other tech earnings of note this week included:

  • Twitter: In what could be one of its last earnings reports as a public company, Twitter reported revenue of $1.2 billion, slightly missing estimates of $1.23 billion, EPS of 4 cents, above the 3 cents expected, and mDAUs of 229 million, above the 226.9 million expected. But the company also admitted a major error, noting it had overcounted its users over the past three years. On a quarterly basis, the overcounting was as high as 1.9 million.
  • YouTube (a part of Alphabet’s earnings): The video site was expected to bring in $7.51 billion, but made $6.87 billion, up 14.9% year-over-year. That’s slower growth than the prior quarter’s 25%, and even smaller than the year-ago quarter when growth was 48.71%. YouTube cited Apple’s privacy rules and others impacting web browsers for the slowing growth. In addition, YouTube’s TikTok rival, Shorts, saw 4x more views than last year, with 30 billion views per day; 2 billion logged-in users visit YouTube monthly.
  • Meta: The company’s family of apps (Facebook, Instagram, Messenger and WhatsApp) grew to 3.64 billion MAUs in the quarter, but ARPU declined from $9.39 to $7.72. Facebook once again grew its DAUs to 1.96 billion, after slipping last year for the first time, when it then dropped from 19.3 billion to 19.29 billion. But much of the money Meta’s making ($27.9 billion in Q1, up 6.6%) is being spent on its VR pivot, Reality Labs. The division lost nearly $3 billion in Q1 after losing some $10b in 2021.
  • PayPal: The payments app and Venmo owner reported revenue of $6.48 billion in Q1, up 8% year-over-year, failing to meet the $6.6 billion analyst estimates. Net income of $1.03 billion, down from $1.46 billion and TPV of $323 billion, was up 15% year-over-year.
  • Robinhood: The trading app reported Q1 revenue was down 43% year-over-year to $299 million, versus $355.8 million estimated, and a net loss of $392 million, down from $1.4 billion year-over-year. The app has 15.9 million MAUs, down from 17.7 million.
  • Pinterest: The image pinboard now pivoting to video reported Q1 revenue of $575 million, up 18% year-over-year, versus $572.5 million expected, and 433 million MAUs, down 9% year-over-year, with a global ARPU of $1.33, up 28%. The company addressed the TikTok threat during earnings, saying it would focus on video content that inspired users to take action, not just be entertained.

Platforms: Apple

  • An App Store cleanup caused controversy this week, as several iOS developers took to social media to report receiving notices from Apple that their older apps without updates will be removed from sale within 30 days if no updates were submitted. Sensor Tower’s analysis of apps that had at least 10,000 installs in 2022 found some 2,966 apps and games could face removal during Apple’s latest purge, as they were last updated before or during 2018. In looking at apps meeting the same level of installs that were last updated before September 2020, it found apps facing removal may number as many as 7,335. Apple then clarified its plan with a developer update, saying apps that have not been updated within the last three years would be the ones it removed. 
  • Apple reminded developers they have until May 20, 2022 to submit a request to receive payment from Apple’s $100 million assistance fund, announced last year in response to a class-action lawsuit from U.S. developers. The fund would pay out between $250-$30,000 to developers making equal to or less than $1 million per year on the App Store between 2015 and 2021. To qualify, developers must have sold paid apps or in-app purchases (including subscriptions) through the App Store between June 4, 2015 and April 26, 2021.
  • The App Store had an outage on Monday which made it appear as of privacy labels were missing from apps.
  • Apple released the third public beta for iOS 15.5, iPadOS 15.5.
  • Apple’s iOS 15.5 beta was found to be blocking sensitive locations for Memories in the Photos app, including places like Holocaust memorials, the Anne Frank house and others.
  • Mobile attribution firm AppsFlyer released a study (via AdWeek) that showed user opt-out rates for ATT have stabilized, based on data from March 2022 across 4,600 apps on its network. Consent was highest when users launched the app for the first time, when 46% selected “Allow” on the pop-up, it found.

Platforms: Google

  • Google launched the first beta of Android 13, also available as an over-the-air update. The beta was first made available to Pixel device owners. During the preview phase, Google had launched features like themed app icons, Bluetooth LE audio, MIDI 2.0 support over USB, per-app language support and opt-in push notifications, among other things. This new beta included more granular permissions for media file access, better error reporting, a new audio API and other smaller updates.
  • Google’s Privacy Sandbox for Android launched in Developer Preview for Android 13. The privacy-focused ad-tracking overhaul provides an early look at the SDK Runtime and Topics API.
  • Google Play officially launched its own version of privacy-related “nutrition labels” for apps. The company said it will begin to roll out the new Google Play Data safety section to users on a gradual basis, ahead of the July 20th deadline that requires developers to properly disclose the data their app collects, if and how it’s shared with third parties, the app’s security practices and more. Compared with Apple’s initiative, Google’s labels put a bigger focus on whether you can trust the data that’s collected is being handled responsibly by allowing developers to disclose if they follow best practices around data security. The company said it would check each Data safety section “using systems and processes that are continuously improving,” when asked how the information developers submit is vetted for accuracy.

Image Credits: Google

Augmented Reality

  • Camera IQ launched support for AR effects within TikTok’s AR development platform Effect House, which allows creators to build AR effects for use in its app.

Fintech

  • The Bored Ape Yacht Club’s Instagram was hacked this week, leading to the theft of millions of dollars of NFTs. The high-profile incident showed how social media apps run by crypto operations could be a source of vulnerability.
  • Trading app Robinhood’s CEO Vlad Tenev announced this week that it was laying off 9% of its full-time employees, potentially impacting some 300 people. The company is coming off two years of hyper-growth, where it grew from 700 to nearly 3,800 employees between 2019-21, and now faces greater competition.
  • WhatsApp is planning to launch cash-back rewards and merchant incentives to attract users in India to its P2P payments service.

Social

  • Instagram began testing a Templates feature, which allows users to copy formats from other Reels. It also started testing a feature that would let users pin multiple Feed posts to their profile, similar to how TikTok lets users pin videos.
  • Reddit officially launched its Community Funds program with a $1 million investment after previously experimenting with funding over a dozen projects nominated by users on the platform, ranging from community-designed billboards to digital conferences. The company will accept nominations for projects needing between $1,000 to $50,000 in funding.
  • TikTok once again became the top app worldwide by downloads in the first quarter, per Sensor Tower. The app had previously surpassed 3.5 billion lifetime installs in Q1 2021, and in the first quarter of 2022 it saw more than 175 million installs. No app has had more downloads than TikTok since the beginning of 2018 when WhatsApp had 250 million worldwide downloads.

Image Credits: Sensor Tower

  • Reels, Instagram’s short-form video feature and TikTok rival, now makes up more than 20% of the time that people spend on Instagram, the company announced during earnings.
  • An NYT review of the Trump-backed social app Truth Social found a number of phony accounts for big brands that were being used by imposters, as well as broken features (including search), and posts that were hidden for using curse words.
  • China’s social app Weibo, similar to Twitter, told users it would now publish the locations of users based on IP address to their account pages in order to combat “bad behavior.” Users will not be able to opt out.
  • TikTok rolls out a feature that lets users opt into seeing who viewed their profile and allowing others to see when they’ve visited their profiles, too.

Streaming & Entertainment

  • YouTube rolled out Super Thanks, its in-app tipping feature that includes animated GIFs alongside the fan’s comments, to all eligible creators across 68 locations in the YouTube Partner Program.

Image Credits: YouTube

  • The HBO Max mobile app just had one of its biggest quarters to date. The app made the top five in terms of most-download apps in the U.S., and also had the second-best month overall in the month of January.
  • NBCU’s streaming app Peacock added 4 million paid subscribers in Q1, up 44% from last quarter. The streamer ended the first quarter of 2022 with 28 million monthly active accounts, up from the 24.5 million Comcast reported for the platform at the end of 2021. Of those, 13 million are paying customers, up from 9 million in the prior quarter.

Gaming

  • Fortnite maker Epic Games filed a motion for a preliminary injunction to stop Google from removing the music storefront Bandcamp, which Epic acquired in March, from the Google Play Store. Google’s policy says Android apps need to move into compliance with the Google Play Billing policy by June 1, or face removal from the Play Store. The filing argues that having to pay Google’s fees would impact Bandcamp’s business model involving paying artists 82% of their Bandcamp revenues in a rev share. But paying Google’s fees could force that biz model to change or force Bandcamp to operate at a loss. Of particular interest: the filing notes Google offered Bandcamp a special deal involving a rev share of 10%, which Epic Games says is still too much.

Health & Fitness

  • The Apple Watch app Gentler Streak, which takes a more compassionate approach to fitness tracking, was updated with improved heart monitoring during exercise, as well as widgets support for iPhone users. The app recently added other non-conventional exercises not tracked by Apple’s Workout app, like dog walking, shoveling and mountain biking.

Government & Policy

  • Japan is now considering new regulations that would require tech giants like Apple and Google to allow for multiple app stores on their platforms. It also viewed the preinstallation of the companies’ own browsers as a potential anti-competitive issue. Apple protested the report saying it disagreed with a number of its conclusions. Google said it will further examine the report.
  • The EU is preparing to lodge an antitrust complaint over Apple Pay on iPhone, as Apple restricts other payment services from having the same access to the iPhone’s NFC hardware.
  • U.S. regulator National Telecommunications and Information Administration (NTIA) is requesting comments on competition in the mobile app ecosystem, including around app store policies, mobile web development and more.

Security & Privacy

  • Google announced it has blocked 1.2 million apps that violated its policies from being published on the Play Store in 2021, banned 190,000 developer accounts for malicious behavior and closed around 500,000 developer accounts that were inactive or abandoned.

Funding and M&A

💰 Copper, a digital banking app aimed at teens, raised $29 million in Series A funding in a round led by Fiat Ventures. The app has grown to more than 800,000 users, up from 350,000 last October, and offers a combination of personalized debit cards, access to 50,000 ATMs, support for digital wallets, parental monitoring of teen spending, P2P, finance advice and more.

💰 Mirai Flights, a London-based app for instant booking of charter flights, raised a $3 million investment round led by Xploration Capital. The Mirai Flights app is available in 63 countries and operates with eight private airlines. Mirai uses a last-mile model, matching supply and demand, utilizing empty flight legs to make it more efficient to return a jet to its base location.

💰 Singapore-based BandLab, an app that lets users create and share music, raised $65 million in Series B funding, at a post-money valuation of $315 million. The round was led by Vulcan Capital. The company says its app is now used by over 40 million creators.

💰 Berlin-based app Taxfix, which helps users with filing taxes, raised a $200 million Series D round led by Teachers’ Venture Growth, valuing the business at over $1 billion.



On putting toothpaste back into the tube

On putting toothpaste back into the tube

Welcome to The TechCrunch Exchange, a weekly startups-and-markets newsletter. It’s inspired by the daily TechCrunch+ column where it gets its name. Want it in your inbox every Saturday? Sign up here

Good news! We’re not talking about crypto, Elon Musk or SaaS multiples today. We’re also not talking IPOs, global venture capital trends or the like. Instead, we’re going to talk about putting toothpaste back in the tube. Sound fun? Let’s go.

China’s technology industry

Since the Ant IPO was pulled and the Chinese Communist Party executed off a flat-wild period of regulatory action in 2021, you have probably heard less about China’s technology. That’s because the companies that tended to make the biggest splash in foreign media were concerns like Alibaba, ByteDance and the like — tech companies that touched lots of individuals, including folks outside the country’s national borders.

China’s government decided that such companies had too much influence, and thus needed to be cut down to size. This meant, variously, the decapitation of the for-profit edtech sector, social media regulation, the effective curtailment of foreign listings, punitive data reviews, video game limits along with a long pause in new titles, new rules regarding algorithms and more.

After a period of comparative freedom to innovate, compete and, yes, at times act anticompetitively, China’s domestic tech industry entered 2022 in a very different state than it kicked off 2020. (This isn’t to discount the impacts of COVID-19 on Chinese tech companies; but the move toward remote work and the like was global, and for our purposes today we care more about the regulatory environments shifts in particular.)

The result of the fusillade of regulatory action, a full nelson of top-down control, was probably about what you expected. Some recent headlines for flavor:

Those should paint a fair enough picture of market sentiment regarding the crackdown. In more monetary terms, the value of many Chinese tech companies fell sharply. After peaking at more than $300 per share in late 2020, Alibaba is worth less than $100 per share today. Didi, which got caught between the Chinese government and the American markets after its IPO, saw its shares peak at a penny over $18 per share. Today it’s worth less than $2 per share.

Stories began to crop up about layoffs and other misery from Chinese tech companies. A few more headlines for context:

Given that this was pretty much what anyone with a pulse might have expected from the Chinese government throwing its absolute control around like gravity in a rollercoaster, pushing to remake one of its key economic engines by autocratic fiat in a short period of time, you are probably not surprised. And yet it appears that the Chinese government is, at least to some degree!

How do we know that? Well, observe:

The context here is that while the rest of the world is largely figuring out a path out of COVID, China’s government is locking down hundreds of millions of its citizens as it chases an impossible goal of zero COVID-19 cases. (The government previously touted its success at keeping the pandemic at bay as evidence of its superiority; such a stance makes any retreat from the goal difficult.) The result of lockdowns and a sharply diminished local tech industry is, surprise, economic malaise.

Not that the Chinese government intends to accept that. After indicating that besting American economic growth is a priority, debt-fueled infrastructure spending is back on the table, along with more real estate speculation, and, it appears, some softening of the rules deluge that its domestic tech market has been forced to endure without complaint.

Good luck?

Can the Chinese government put the tech toothpaste back in the tube? We’ll find out, but if I was an investor or founder I would not build inside the country. Sure, it’s a big market, but not one that you can count on. More when we get Q2 2022 Chinese venture capital data.



Perceptron: AI mixes concrete, designs molecules, and thinks with space lasers

Perceptron: AI mixes concrete, designs molecules, and thinks with space lasers

Welcome to Perceptron, TechCrunch’s weekly roundup of AI news and research from around the world. Machine learning is a key technology in practically every industry now, and there’s far too much happening for anyone to keep up with it all. This column aims to collect some of the most interesting recent discoveries and papers in the field of artificial intelligence — and explain why they matter.

(Formerly known as Deep Science; check out previous editions here.)

This week’s roundup starts with a pair of forward-thinking studies from Facebook/Meta. The first is a collaboration with the University of Illinois at Urbana-Champaign that aims at reducing the amount of emissions from concrete production. Concrete accounts for some 8 percent of carbon emissions, so even a small improvement could help us meet climate goals.

This is called “slump testing.”

What the Meta/UIUC team did was train a model on over a thousand concrete formulas, which differed in proportions of sand, slag, ground glass, and other materials (you can see a sample chunk of more photogenic concrete up top). Finding the subtle trends in this dataset, it was able to output a number of newformulas optimizing for both strength and low emissions. The winning formula turned out to have 40 percent less emissions than the regional standard, and met… well, some of the strength requirements. It’s extremely promising, and follow-up studies in the field should move the ball again soon.

The second Meta study has to do with changing how language models work. The company wants to work with neural imaging experts and other researchers to compare how language models compare to actual brain activity during similar tasks.

In particular, they’re interested in the human capability of anticipating words far ahead of the current one while speaking or listening — like knowing a sentence will end in a certain way, or that there’s a “but” coming. AI models are getting very good, but they still mainly work by adding words one by one like Lego bricks, occasionally looking backwards to see if it makes sense. They’re just getting started but they already have some interesting results.

Back on the materials tip, researchers at Oak Ridge National Lab are getting in on the AI formulation fun. Using a dataset of quantum chemistry calculations, whatever those are, the team created a neural network that could predict material properties — but then inverted it so that they could input properties and have it suggest materials.

“Instead of taking a material and predicting its given properties, we wanted to choose the ideal properties for our purpose and work backward to design for those properties quickly and efficiently with a high degree of confidence. That’s known as inverse design,” said ORNL’s Victor Fung. It seems to have worked — but you can check for yourself by running the code on Github.

View of the top half of South America as a map of canopy height.

Image Credits: ETHZ

Concerned with physical predictions on an entirely different scale, this ETHZ project estimates the heights of tree canopies around the globe using data from ESA’s Copernicus Sentinel-2 satellites (for optical imagery) and NASA’s GEDI (orbital laser ranging). Combining the two in a convolutional neural network results in an accurate global map of tree heights up to 55 meters tall.

Being able to do this kind of regular survey of biomass at a global scale is important for climate monitoring, as NASA’s Ralph Dubayah explains: “We simply do not know how tall trees are globally. We need good global maps of where trees are. Because whenever we cut down trees, we release carbon into the atmosphere, and we don’t know how much carbon we are releasing.”

You can easily browse the data in map form here.

Also pertaining to landscapes is this DARPA project all about creating extremely large-scale simulated environments for virtual autonomous vehicles to traverse. They awarded the contract to Intel, though they might have saved some money by contacting the makers of the game Snowrunner, which basically does what DARPA wants for $30.

Images of a simulated desert and a real desert next to each other.

Image Credits: Intel

The goal of RACER-Sim is to develop off-road AVs that already know what it’s like to rumble over a rocky desert and other harsh terrain. The 4-year program will focus first on creating the environments, building models in the simulator, then later on transferring the skills to physical robotic systems.

In the domain of AI pharmaceuticals, which has about 500 different companies right now, MIT has a sane approach in a model that only suggests molecules that can actually be made. “Models often suggest new molecular structures that are difficult or impossible to produce in a laboratory. If a chemist can’t actually make the molecule, its disease-fighting properties can’t be tested.”

Looks cool, but can you make it without powdered unicorn horn?

The MIT model “guarantees that molecules are composed of materials that can be purchased and that the chemical reactions that occur between those materials follow the laws of chemistry.” It kind of sounds like what Molecule.one does, but integrated into the discovery process. It certainly would be nice to know that the miracle drug your AI is proposing doesn’t require any fairy dust or other exotic matter.

Another bit of work from MIT, the University of Washington, and others is about teaching robots to interact with everyday objects — something we all hope becomes commonplace in the next couple decades, since some of us don’t have dishwashers. The problem is that it’s very difficult to tell exactly how people interact with objects, since we can’t relay our data in high fidelity to train a model with. So there’s lots of data annotation and manual labeling involved.

The new technique focuses on observing and inferring 3D geometry very closely so that it only takes a few examples of a person grasping an object for the system to learn how to do it itself. Normally it might take hundreds of examples or thousands of repetitions in a simulator, but this one needed just 10 human demonstrations per object in order to effectively manipulate that object.

Image Credits: MIT

It achieved an 85 percent success rate with this minimal training, way better than the baseline model. It’s currently limited to a handful of categories but the researchers hope it can be generalized.

Last up this week is some promising work from Deepmind on a multimodal “visual language model” that combines visual knowledge with linguistic knowledge so that ideas like “three cats sitting on a fence” have a sort of crossover representation between grammar and imagery. That’s the way our own minds work, after all.

Flamingo, their new “general purpose” model, can do visual identification but also engage in dialogue, not because it’s two models in one but because it marries language and visual understanding together. As we’ve seen from other research organizations, this kind of multimodal approach produces good results but is still highly experimental and computationally intense.



Affirm’s CTO talks transparency and the tech that makes BNPL possible

Affirm’s CTO talks transparency and the tech that makes BNPL possible

BNPL is not a new concept; it’s just taken off in recent years and become far more mainstream.

Buy now, pay later lets people do exactly what its name suggests — buy something and pay for it later. The difference between BNPL and credit cards is that rather than charge the full amount of a purchase on a card, consumers can choose to pay for an item in installments.

However, there are some that argue BNPL is just another form of debt, which could lead to a discussion on whether companies that enable it are doing it responsibly. In the case of Affirm, one of the space’s largest players, co-founder Max Levchin (who also founded PayPal) has been vocal about what he describes as a “mission-based” approach.

Ukraine-born Levchin started Affirm in January 2012. The fintech went public in 2021, and while it’s trading considerably lower than its 52-week-high (which stock isn’t?), Affirm is today valued at nearly $9 billion, and its executives remain bullish on the company’s future.

TechCrunch sat down with Libor Michalek, president of technology at Affirm, to understand just how the company differentiates itself from its plethora of competitors, what is unique about its technology and strategy, and why he thinks using BNPL is much better than using a credit card to pay for purchases.

(Editor’s note: This interview has been edited for length and clarity.)

TC: I grew up in the era of layaways, where you could pay in installments for an item but had to wait to take it home. So when I heard about BNPL, I was intrigued. In your view, what makes Affirm stand out?

We have this notion of a vertically integrated stack where we are able to handle the full touchpoint — that really gives us a lot of visibility into the customer, in the transaction, and that lets us underwrite accurately.

Libor Michalek: Our main focus is doing right by the customer. And that really translates into this idea of aligning our interests with that of the customer. So if they get the unexpected or unwanted, then we share in the negative outcomes.

The second pillar for us is building modern technology that enables us to do this. How do you deliver a financial product with no late fees, with no gimmicks and no deferred interest tricks? It’s really the ability to have access to real-time data, deliver it on the phone and do it at e-commerce sites in real time, and then bringing all that together to make real-time decisions and deliver those decisions clearly to the customer.

Another advantage we have is the scale of our merchant network. We work with 170,000 merchants, which enhances our ability to provide access to à la carte credit wherever the customer might want it and need it.

I recently learned that Affirm (and other BNPL players) do charge interest at times, but often at a lower rate than traditional credit card providers. Tell us more about how those decisions are made — how do you decide who is charged interest, and who isn’t?

For us, the most important and biggest difference is that unlike a credit card, the customer knows how much interest in dollars they’re going to pay for that purchase. There’s no way for them to pay more for that purchase, and they will know it upfront before they click.

We’ll communicate it to them obviously, as an interest rate as we’re legally required to, but also in dollars and cents. A lot of times people get surprised when I tell them that a $1,000 purchase at 15% for a year actually translates to $83 because of the amortization schedules. A calculator on our website lets you play with all of those numbers.

I think the transparency part is pretty key, because I feel like with credit cards, you do run that risk of — depending on how long it takes you to pay or what your minimum payments are — how much you pay in interest potentially ranging wildly. With us, it’s a fixed amount that’s communicated to the customer upfront.

And even if they miss a payment, there are no late fees and nothing gets tacked on in any way that would ever result in a different outcome. In fact, if they pay early, the number can be lower, but it won’t ever exceed the figure we give them.

How many people are usually able to use BNPL through Affirm without being charged interest?



India seizes $725 million assets from Xiaomi unit over illegal remittances

India seizes $725 million assets from Xiaomi unit over illegal remittances

India’s anti-money laundering agency said on Saturday it has seized assets worth about $725 million from Xiaomi India for breaching the country’s foreign exchange laws in a major blow to the Chinese phone maker that commands the Indian smartphone market.

The Indian Enforcement Directorate said it had seized bank accounts of Xiaomi India after finding that the company had remitted $725 million to three foreign-based entities “in the guise of royalty” payments.

“Such huge amounts in the name of royalties were remitted on the instructions of their Chinese parent group entities,” it said. The amount remitted to “other two US-based unrelated entities” were also for the “ultimate benefit of the Xiaomi group entities,” the agency added.

The directorate, which has been investigating Xiaomi as well as several other Chinese firms since December, said Xiaomi has “provided misleading information to the banks while remitting the money abroad.”

Xiaomi, which has yet to comment, commanded 23% of India’s smartphone market share in the quarter that ended in March this year, according to market research firm Counterpoint. Its former India head, Manu Jain, was summoned by the directorate earlier this year for questioning over tax related compliances.

The company has taken a hit in its popularity in recent years following India’s ban on Chinese apps over national security concerns. For optics measures, Xiaomi rebranded several of its shops in India two years ago with “Made in India” banners in a move that analysts said was the company’s attempt to distance itself from its Chinese parent firm.

The company earlier this week updated its smartphone, smart TV and tablet lineups in India.



If the earliest investors keep going earlier, what will happen?

If the earliest investors keep going earlier, what will happen?

There’s a clash happening in the early-stage market.

In one world, late-stage investors are reacting to tech stonk corrections by clamoring toward the early-stage investment world, forcing seed investors to go even earlier to defend ownership and potential returns. This trend was underscored by firms like Andreessen Horowitz launching a pre-seed program months after launching a $400 million seed fund. Even more, Techstars, an accelerator literally launched to help startups get off the ground, debuted a fund to back companies that are too early for its traditional programming.

While all that is going on, early-stage investors are enduring a valuation correction and portfolio markdowns. Some are admitting that they’re telling portfolio companies to refocus on cash conservation, profitability and discipline, not just growth.

Let’s pretend these two vastly different worlds are in the same universe: Early-stage investors are getting more disciplined and cash rich, but at the same time, the earliest investors are going earlier. Investors are pushing founders to be lean but also green, but at the same time, offering them $10,000 to take PTO for a week and try their hand at entrepreneurship. Growth, gross margin and burn are the new top priorities for CEOs, but at the same time, venture capitalists are clamoring to offer more funds, earlier, in newly invented subcategories of early-stage investment.

The tension between these two worlds looks different depending on if you’re a Stanford founder starting a SaaS company, or if you’re a bootstrapped, first-time entrepreneur trying to disrupt agtech. Regardless, the growing spotlight, and discipline, on the early stage just makes me wonder one broad thing: What’s left for early-stage investors to focus on?



Friday 29 April 2022

China’s logistics robot maker VisionNav raises $76M at $500M valuation

China’s logistics robot maker VisionNav raises $76M at $500M valuation

Industrial robotics has in recent years become one of the hottest tech sectors in China as the country encourages the use of advanced technology to enhance efficiency on the production floor.

VisionNav Robotics, which specializes in autonomous forklifts, stacking vehicles and other logistics robots, is the latest industrial robot maker from China to get funded. The Shenzhen-based automated guided vehicle (AGV) startup has snagged 500 million yuan (around $76 million) from a Series C extension round led by Meituan, China’s food delivery giant, and 5Y Capital, a prominent venture capital firm in the country. Its existing investors IDG, TikTok’s parent firm ByteDance, and Xiaomi founder Lei Jun’s Shunwei Capital also joined in the round.

Founded in 2016 by a group of PhDs from the University of Tokyo and the Chinese University of Hong Kong, VisionNav’s valuation was boosted to over $500 million in this round, up from $393 million just six months ago when it picked up 300 million yuan ($47 million) in a Series C financing, it told TechCrunch.

The new funding will allow VisionNav to invest in R&D and broaden its use cases, expanding from a focus on horizontal and vertical moving to other functions like stacking and loading.

The key piece in adding new categories is to train and improve the startup’s software algorithms, less so developing new hardware, said the firm’s vice president of global sales, Don Dong. “From controlling, dispatching, to sensing, we will have to improve our software capabilities as a whole.”

A major challenge for robots, said Dong, is to effectively perceive and navigate the world around them. The problem with a camera-powered autonomous driving solution, like that of Tesla, is it can be easily affected by bright light. Lidar, a sensing technology heralded for its more accurate distance detection, was still too expensive for mass adoption a few years ago, but it has seen its price slashed substantially by Chinese players like DJI-affiliated Livox and Robosense.

“Before, we were mostly providing indoor solutions. Now that we are expanding to unmanned truck loading, which is often semi-outdoors, it’s inevitable we will be operating in strong light. That’s why we are adapting a combination of vision and radar technologies to navigate our robots,” said Dong.

VisionNav sees Pittsburg-based Seegrid and France-based Balyo as its international rivals but believed it has a “price advantage” for being in China, which houses its manufacturing and R&D activities. The startup is already dispatching robots to clients in Southeast Asia, East Asia, as well as the Netherlands, the UK, and Hungary. It’s in the process of setting up subsidiaries in Europe and the US.

The startup works with system integrators to sell its robots, meaning it doesn’t collect detailed customer information, making data compliance in foreign markets simpler. It’s expected to derive 50-60% of its revenues abroad in the next few years, compared with the current share of 30-40%. The US is one of its main target markets, said Dong, as the forklift industry there generates “greater gross revenues than that of China despite having a smaller number of forklift vehicles.”

Last year, VisionNav pulled in total sales revenues between 200 million ($31 million) and 250 million yuan ($39 million). It currently operates a team of around 400 people across China and it’s expected to reach 1,000 staff this year by hiring aggressively overseas.



More TradFi in DeFi as Sienna Network launches SiennaLend, a private crypto lending platform

More TradFi in DeFi as Sienna Network launches SiennaLend, a private crypto lending platform

It was last year that we covered the $11.2M fundraise for Sienna Network, the ‘privacy decentralized’ startup. The network is built on the Secret Network, which allows asset holders to switch to privacy-oriented tokens. Privacy-based financial blockchain projects are crucial if “DeFi” is to work properly, otherwise normal financial transactions – which are normally private in the traditional finance world – will struggle to take off.

Sienna is among several other blockchain startups trying to prevent “front-running”, where transactions on Ethereum can be preempted by someone else simply by them paying a higher transaction fee – just like trumping a trade on the stock market by paying a higher fee to a broker.

Sienna Network has now launched its private crypto lending platform, dubbed SiennaLend. The company claims that crypto users can use the platform to earn interest on their crypto and also borrow crypto from the platform – all privately. The platform will compete in this space with Uniswap and PancakeSwap, but claims to have more features than both.

SiennaLend is (obviously) built on top of the Secret Network, which affords – claims the company – greater security and safety compared to open and non-private blockchains such as Ethereum, Solana and others.

SiennaLend’s pitch to the market is that its lower gas prices will make it more attractive to small investors.

“Paying a transaction fee of $150 to make a loan of $200 makes little sense and this is another major benefit for SiennaLend. Gas fees are counted in cents rather than dollars as the scalability of the blockchains are much higher. We have spent 15 months in stealth, fine-tuning this absolute game-changer for crypto and its ascendance to the next level of mainstream finance,” Monty Munford, Chief Evangelist, Sienna Network, said in a statement.

Sienna Network says “Personal Identifiable Information (PII)” means users can “backtrack” and see their wallet possessions and trading history via their wallet address. It will also offer loans against collateral, as would happen with traditional lending, by allowing users to deposit into a pool and choose to earn interest or borrow based on that deposit. The idea is that traders can defend against the current market volatility more easily.

Lending protocols like this in DeFI are widely observed to be taking off, despite market volatility, because they offer more reassurances to crypto holders. Cryptocurrency-based loans have become a highly utilized aspect of DeFi.

By being built on Secret Network, Sienna Network/SiennaLand is also part of the Cosmos ecosystem, recently upgraded to be a part of IBC. Cosmos competes – after a fashion with Polkadot, also an ‘internet of blockchains’.



Thursday 28 April 2022

Musk sells nearly $4 million worth of Tesla shares

Musk sells nearly $4 million worth of Tesla shares

Tesla CEO Elon Musk sold around 4.4 million shares of the company on Tuesday, according to regulatory filings published on Thursday.

Musk, via Aaron Beckman, his power of attorney, filed a total of five Form 4s with the U.S. Securities and Exchange Commission to cover all 138 individual transactions.

The value of the sales in the filings disclosed is so far is around $4 million, per TechCrunch calculations.

The executive said in a tweet on Thursday: “No further TSLA sales planned after today”.

The filings don’t reveal why Musk sold his shares, something he’s been wont to do. The money could possibly go towards his recent controversial plans to purchase social media platform Twitter, however, $4 million seems to hardly make a dent in the $44 billion the acquisition will cost.



Airbnb commits to fully remote workplace: ‘Live and work anywhere’

Airbnb commits to fully remote workplace: ‘Live and work anywhere’

Airbnb is going all in on the “live anywhere, work anywhere” philosophy that much of the business world has been forced to adopt, committing to full-time remote work for most employees and a handful of perks like 90 days of international work/travel. It’s a strong, simple policy that so few large companies have had the guts to match.

In an email to employees posted to the company blog (or was it a blog post emailed to employees?), and in a Twitter thread for those who can’t be bothered, Airbnb CEO Brian Chesky outlined the new policy, summing it up in five points:

  • You can work from home or the office
  • You can move anywhere in the country you work in and your compensation won’t change
  • You have the flexibility to travel and work around the world
  • We’ll meet up regularly for gatherings
  • We’ll continue to work in a highly coordinated way

They’re pretty self-explanatory, obviously, but just to be clear let’s run them down.

Apart from “a small number of roles” for whom presence in an office or location is required (and who probably already know this), all employees can work from wherever they want.

If you want to move, as long as you stay within the country, your pay won’t change. Wherever you go in the U.S., for instance, you’ll get the same pay, and one hopes it’ll be enough whether you live in a small town in Colorado or midtown Manhattan. Sadly if you decide you want to move permanently to London or Seoul, this is “much more complex, so we won’t be able to support those this year.”

Though workers will need a permanent address, they’ll have dozens of companies and locations to work from for up to 90 days a year — so stay over in Lisbon for a bit and work from that villa for a week or two after your vacation. Why not? Well, possibly because remote work visas may not be available for those areas, but that’s all a work in progress. (They’re adding partners to a big list over here.)

Chesky says they’ll all “meet up regularly,” even though Airbnb probably has about 15,000 employees at this point. That’s even more than TechCrunch! They’ll have “limited off-sites” in 2022, which is probably smart, but next year you can “expect to gather in person every quarter for about a week at a time.” I really don’t understand how they can possibly get any work done over there.

The last point seems kind of superfluous and self-congratulatory, but it is probably good to officially note that the general shape of working at the company, or how people are managed and so on, won’t change due to this new policy.

Many companies have announced tentative policies with the understanding that they would be revisited in a few months. There’s a lot of talk about the “hybrid” or “flex” model where employees work from the office a few days, then from home the rest of the time. Depending on where and how you work, this could be the best or worst of both worlds. But it does suggest a certain lack of decisiveness in leadership. (Among the early adopters of full time remote work was Twitter, which may soon be under new leadership.)

And then there’s the safety and liability question. Activision Blizzard, already kind of fubar, mandated a return to the office, then lifted their vaccine mandate. As someone noted at the time, “do not die for this company,” or any company for that matter.

Perhaps Airbnb will be the guinea pig for this particular type of “fully remote workplace” and all the other companies will be watching and waiting for the company to stub its toe on some huge new tax burden or productivity issue. But the simplicity and flexibility of the policy, international legal restrictions notwithstanding, may outweigh any new troubles it creates.



Revise raises $3.5 million to build rails for programming NFTs

Revise raises $3.5 million to build rails for programming NFTs

A popular criticism of NFTs is that they are just static JPEG files. Technically, they are not, of course. They are pieces of code on a blockchain, which means that they can be programmed to have various qualities. The NFTs that go on sale on marketplaces such as OpenSea are already programmed with instructions on the royalties to be provided to the owner, for instance.

But what if they could be programmed to do much more?

That’s a trend we have seen in recent quarters. Pearpop introduced dynamic NFTs earlier this year that gain value as a social media post goes viral. The NFTs on Axie Infinity change their properties as a user makes inroads in the game.

A new startup, called Revise, is attempting to productize this ability.

It offers developers the ability to make their own NFTs interact with data feeds of their choice, which could be a web3 platform such as Chainlink or a web2 outlet like Weather.com.

The goal, as the startup explains it, is to make the NFTs change their properties based on events. For instance, a soccer NFT could hypothetically interact with data from FIFA and change its property or media content based on the real-world performance on the field.

“What programmable NFT allows you to do is also leverage the user’s interaction or skill to make the properties rarer,” said Raunaq Vaisoha, co-founder and chief executive of Revise, in an interview with TechCrunch.

By programming NFTs based on users’ interactions, developers can incentivize them to participate more in their projects, he said.

Revise is also adding a layer of governance for the storage through its data structure to help developers handle disputes in a trustable manner.

“Currently, an issue with dynamic NFTs is that your data has to be off-chain. Imagine you’re playing a game and you have to wait for the block time for your gaming character to update. Most people end up storing the NFT on an AWS S3 or a different web2 layer,” he said. “That’s what our data structure is built to solve.”

Revise is initially launching with the Polygon blockchain, but plans to expand to other blockchains in the future. Its SDK is live in private beta on NPM, and the startup has amassed some customers already, including Ludo Labs.

The startup said on Friday it has raised $3.5 million in its seed round. Alpha Wave Global and 8i co-led the round. Bharat Founders Fund and a number of entrepreneurs including Polygon founder Sandeep Nailwal, DeFi Pulse founder Scott Lewis, AngelList India head Utsav Somani, The Graph’s Pranav Maheshwari, and Treebo founder Rahul Chaudhary also participated in the round.

“The entire NFT space has seen a massive narrative shift as people have discovered that NFTs can be more than just digital collectables or static assets. We see this shift happen in Gaming most prominently, but the abstract concept is broadly applicable to any real-world asset that can be tokenized. Further, as more complex utilities are built on NFTs, the aspect of traceability and transparency in governance will become front and centre,” said Tushar Behl of Alpha Wave Global, said in a statement.

“We loved Revise and the founders for their deep product insight and a forward-looking vision of the space. What the team at Revise is building can become the most fundamental layer for NFT provenance and programming, much like Chainlink did for Defi!”



Glorang scores $10M Series A to expand its edtech marketplace across Asia 

Glorang scores $10M Series A to expand its edtech marketplace across Asia 

Glorang, a Seoul-based edtech startup that offers after-school classes and extracurricular activities via online for students between the ages of 3 and 18, said Friday it has raised a $10 million Series A funding co-led by Korea Investment Partners and Murex Partners, along with Japan’s Pksha Capital. 

The new funding, which brings its total raised to $18 million, values Glorang at around $40 million, Glorang CEO and founder Taeil Hwang told TechCrunch.  

The startup has aspirations to become Outschool of Asia. Hwang said that its business model is similar to Outschool, the San Francisco-based after-school marketplace for children. Glorang will use the Series A to expand its service to Japan and Malaysia by the fourth quarter of this year and Taiwan, Thailand and Vietnam in the following years, Hwang said. It also plans to increase its headcount. 

“The education market in the English-speaking and North American regions is undoubtedly large, but we [at Glorang] understand that each country’s local D2C education market in Asia can be just as substantial,” Hwang said.

The coronavirus pandemic has forced students in many parts of the globe to become online learners; the education technology industry has experienced a sudden surge and demand globally due to the pandemic. The Asia Pacific is one of the fastest-growing regions in the adoption of Edtech, increasing to $64.5 billion in 2027, from $17.6 billion in 2019.

Glorang was founded in 2017 by Hwang, who started off this company with an AI-powered platform helping students match with study abroad programs. Before the COVID-19 pandemic, Glorang pivoted its main business, the online class platform, and launched Gguge in 2020, Hwang said. The company claims Gguge has more than 100,000 users in South Korea. 

What sets Gguge apart from its peers is providing education services in local languages, Hwang said adding that it currently offers Korean but will add the Japanese language soon. 

Gguge offers a selection of more than 5,000 online classes via Zoom. Instructors help students engage with active learning methods, ranging from reading newspapers through solving puzzles to incorporating Minecraft and Pokemon games into the classes. 

“As a team that understands both the local culture and strategies in Asia, we are confident that our platform will have a strong standing in Asia’s ever-growing D2C education market,” Hwang said. 

Students can take a one-day class or subscription-based semester classes via Gguge. The company has a team of 35 in Korea.