Three screenshots of the Blush dating app

Invite-only dating app Blush launches with $7M in funding to redefine online dating

Three screenshots of the Blush dating app

Image Credits: Blush

Invite-only dating app Blush has raised $7 million in seed funding to scale and expand its exclusive platform. Blush aims to redefine the online dating experience through community events, virtual courting and bonding over mutual favorite local spots. The app is currently only available in Los Angeles, but is accepting applications for early-access members in Miami, New York, Chicago and Austin.

Blush was founded in October 2023 by Matthew Alfin, who came up with the idea for the startup based on his interest in the intersection of local places and connecting people.

“I felt there was an opportunity to do something interesting in dating,” Alfin told TechCrunch. “I felt there was also an opportunity to reward our users in the process while driving them to local businesses on dates. We wanted to bring back the nostalgia where you spent time courting and getting to know someone.”

Blush lets users send their potential matches digital roses, which cost $3.50 each and can be redeemed for gift cards to popular places in your community, including restaurants, grocery stores, cafés and spas. In Los Angeles, which is where Blush is currently available, roses can be redeemed at places like Erewhon, Alfred Coffee and Craig’s. Alfin says the feature is designed to mimic courting from back in the day when you would send someone roses to show that you were interested in them. The company says users are four times as likely to match when sending roses.

Once users create a Blush profile, they are asked to share their favorite local café, restaurant, gym and more. The app is focused on bridging the gap between the online and offline world by using algorithms to match people based on their favorite local places in order to create a bond over common interests and favorite spots. Blush believes that this makes online dating more personal and grounded in real-life experiences.

As for the company’s community events, Blush holds private parties where members can enjoy food and music while connecting with other Blush users. The idea behind the community events is to curate real-life experiences beyond the online world.

For people who are into astrology, the app features dynamic zodiac compatibility charts that give users insights into potential matches.

When asked about what sets Blush apart from the numerous dating apps currently on the market, Alfin said the combination of real-life events and turning roses into rewards are the biggest differentiators. 

“The ability to match with someone on the app and then see them at an event is game changing in the dating space,” Alfin said. “Blush is all about bridging the gap between the online and offline worlds, from our mutual locations feature and in real-life events to rewarding users with gift cards for popular local places. Other dating apps are designed to maximize revenue and do not reward its users; by turning roses into rewards at local businesses we are bringing back this traditional gesture of courtship in a modern way while rewarding users.”

Blush plans to use the funding to expand the app to more locations and introduce more features. The company’s seed funding round included participation from Angellist founder Naval Ravikant, DFJ Growth founder Mark Bailey, former Greylock partner Simon Rothman, music industry manager Larry Rudolph and more.

The company is also in the process of creating an AI dating coach that would act as a personalized advisor to help you on your dating journey. It would let users to ask dating-related questions and receive guidance on dating matters.

Plex Rentals Grid

Streamer Plex launches its long-promised movie rentals store

Plex Rentals Grid

Image Credits: Plex

Fresh on the heels of its $40 million fundraise, streaming media company Plex is today announcing its expansion into a new business: a movie rentals storefront. The addition, which will initially be offered to U.S. customers, will give the streamer another means of generating revenue beyond its subscription products and ad-supported streaming — a diversification that will prove critical as the ad market continues to be unpredictable.

At launch, the marketplace will offer movies from top studios, including WB, Paramount, MGM, Lionsgate and A24, which means Plex users will be able to rent titles like “Barbie,” “Wonka,” “Aquaman and the Lost Kingdom,” “Mission: Impossible — Dead Reckoning,” “The Color Purple,” “Expend4bles,” “PAW Patrol: The Mighty Movie,” “Hunger Games: The Ballad of Songbirds and Snakes,” “Mean Girls” and others.

Image Credits: Plex

Plex says there will be just over 1,000 titles available to rent starting at $3.99, but the number of titles will grow over time. Titles will also move in and out of windows, so the number of rentals will fluctuate over time, as well.

The company had teased its plans for a TV and movie rental marketplace for years, but never quite got to the finish line as other priorities took over. Initially planned for just ahead of the COVID-19 pandemic, things changed as Plex, like many other companies, saw an influx of streaming customers which shifted its focus. Other technical concerns also slowed things down at times.

However, at the Consumer Electronics Show (CES) this year, Plex confirmed a rentals marketplace was nearing launch. But as it turns out, the storefront will feature just movies for the time being — not TV shows. It also won’t offer movies for purchase alongside the rentals, because Plex’s user surveys found that rentals were in more demand.

Image Credits: Plex

“We’re looking at the purchase use case because that creates some additional wrinkles — now you’ve got to keep this locker for people long-term and does that really make sense [for us]?,” explained Plex CEO Keith Valory. He noted that the lack of TV show rentals also came from user surveys, which found that most would still turn to streaming services for watching TV, but would be more interested in renting movies.

Once users rent a movie, they have 30 days to watch. After starting the rental, you’ll have 48 hours to finish viewing it, similar to other marketplaces. The movie will also appear in the “Continue Watching” section on Plex’s home screen if you don’t finish watching it upon your first go. The company plans to add more studio partners to its movie rentals store over time, it says.

While there are already plenty of other places where consumers can rent movies today, Plex believes its recommendation capabilities will be a differentiator for its business. Today, Plex’s customers use its software to organize their home media, stream live channels, watch ad-supported TV and movies, and discover new shows across their services or via their friends’ viewing habits through newer social networking features. (The latter got Plex into hot water recently as customers discovered they had shared their viewing of content they’d rather have kept quiet. Plex says it has adjusted this sharing feature to make it more obvious to users that they were opting in.)

That plethora of user data could help Plex drive more users to movie rentals, the company says. For instance, if you saw that your friend had just watched a new movie that you had on your watchlist, you might be inspired to rent it and watch it, too.

Image Credits: Plex

“I think we’re well ahead of the market from a technology standpoint on our personalization and recommendations,” Valory says. “And we use a fair bit of machine learning and other heuristics to kind of get that. I think one of the advantages we have when we build our recommendation engine is that it’s across all content,” he adds.

Of course, Plex also has to contend with being a partner to studios when, in fact, its software solution is used to store people’s home media collections — including movies ripped from DVDs in years past. The company stresses that’s not an issue, however, because it works with the studio partners to provide protection capabilities and watermarking to shut down illegal use and protect their copyright in the case that something gets compromised.

The new movie marketplace will launch across platforms, Plex notes, including its apps on Amazon Fire TV, Apple TV, Android TV/Google TV, Roku, smart TVs (LG, Hisense, Samsung, Sony, VIZIO), game consoles and Apple and Android smartphones and tablets.

 

OpenAI logo and Sam Altman

OpenAI releases Sora, a credit score–based dating app launches and an anti-Tesla ad comes under fire

OpenAI logo and Sam Altman

Image Credits: OLIVIER DOULIERY / AFP / Getty Images

Welcome, folks, to Week in Review (WiR), TechCrunch’s regular newsletter covering noteworthy happenings in the tech industry.

This week, OpenAI stunned the blogosphere with the release of Sora, a new AI model that can generate videos in impressively high fidelity. We’ve seen video generators before. But what makes Sora unique is its understanding of time and physics, which enable it to create not only more coherent videos than previous video generators, but also 3D worlds. Wild stuff!

Elsewhere, startup Score released a dating app exclusive to people with good to excellent credit scores. And an anti-Tesla Super Bowl ad came under fire from the National Transportation Safety Board for using its seal without authorization.

Lots else happened. We recap it all in this edition of WiR — but first, a reminder to sign up to receive the WiR newsletter in your inbox every Saturday.

News

Cleaning fees begone: Airbnb is slowly killing the cleaning fee as it aims for more transparent pricing, Amanda writes.

Silenced before its time: Layoffs at Spotify have put an end to Glenn McDonald’s beloved Every Noise at Once project, a musical encyclopedia of sorts — and fans are pissed.

Mozilla downsizes: Following downsizing, Mozilla plans to scale back its investment in a number of products, including its VPN, Relay and Online Footprint Scrubber, Frederic reports.

Google upgrades Gemini: Google expanded the range of its Gemini AI models available to developers across its platforms. And it’s previewed a new Gemini model capable of analyzing whole books, hours-long audio and hour-long videos.

Slack gets GenAI: Slack introduced a couple of new features designed to make information more accessible, including a new AI-fueled search tool and the ability to summarize information inside channels.

Variston folds: Spyware startup Variston is losing staff — and some say it’s closing up shop entirely. The Barcelona-based startup’s malware has been used to target iPhones, Android devices and PCs, Lorenzo writes.

Analysis

Bluesky vs. Mastodon: Amanda writes about the fight over how — and whether — to bridge the two decentralized social networks Bluesky and Mastodon and how it could shape the future of the internet.

Fortnite and the Mouse: With Disney’s magic (“magic” here referring to IP and a $1.5 billion investment), Fortnite is poised to win the metaverse, Taylor writes.

Podcasts

On Equity, the crew talked about how Bret Taylor’s new startup, Sierra, is turning heads. Taylor — known for his work at Facebook, Salesforce and OpenAI — says that Sierra is about building conversational AI agents.

Found profiled Tigran Sloyan, co-founder and CEO of CodeSignal, a skills assessment platform used by many tech companies to hire engineers based on their engineering chops rather than their résumés.

And Chain Reaction had on Yat Siu, executive chairman of Animoca Brands, which has invested in over 400 web3 projects across a range of sectors like DeFi, education, infrastructure, blockchain gaming and the metaverse.

TechCrunch+

Flash in the pan: Rebecca writes how, for a variety of reasons, VCs are no longer gun-shy about firearm startups.

Demand for ethics: Regenerative community organism (RCO), a novel organizational model that aims to be a practical approach to integrating sustainability at the core of operations, is gaining steam in the startup world, Haje reports.

Bonus round

Foundry Group shutters: Foundry Group, an 18-year-old venture firm with nearly $3.5 billion in assets under management, has quietly decided to shut down and not raise any more funds.

Playing hardball: Apple has confirmed that it’s breaking iPhone web apps in the European Union (EU) on purpose — blaming the new EU regulation, the Digital Markets Act, for the change.

Walmart's PhonePe launches India app store in challenge to Google

Image Credits: PhonePe

Microsoft, Amazon and Meta are among the top brands whose apps are listed on the Indian fintech PhonePe’s Indus Appstore, an Android app store launched Wednesday in challenge to Google’s monopoly in its largest market by users.

To fight Google Play Store, PhonePe has armed Indus Appstore with a range of unique and personalized features and developer-friendly terms. Indus Appstore supports 12 regional languages (including English) and around-the-clock support service, the $12 billion Walmart-backed fintech said.

The app store — which today features over 200,000 apps, including Flipkart, Spotify, and Paytm across 45 categories — supports third-party payments and will not levy any listing fee for one year, it said. If developers use third-party payments services and gateways, PhonePe will still not levy any fee, said the startup’s co-founder and chief executive Sameer Nigam at a press conference.

Google apps were missing from the store, which started courting developers in September last year. PhonePe, which leads the mobile payments market in India, is working with many third-party aggregators to on-board the apps and many big brands have also signed up to Indus Appstore, it said.

Indus Appstore features several other localized features, including app discovery via short videos and the ability to sign in to the store with a mobile number instead of an email address. It also offers “smart updates” that beam updates to customers’ phones during hours when their phones have greater data availability.

PhonePe aims to be a top Google Play alternative in India — but it has a challenging road ahead

Indus Appstore is the latest of a series of attempts by app developers globally to push back against Google and Apple’s duopoly on mobile apps and what many allege unfair terms. The two tech giants are facing growing backlash from app developers globally over the commissions charged on their respective app stores. A plethora of complaints has emerged accusing the companies, which control over 99% of the global consumer spend on mobile apps, of abusing their dominant position to impose excessive fees of up to 30% on app transactions.

India, the world’s largest market by app downloads, is a nation of 750 million to 800 million connected smartphone users. Indus Appstore exists because as a nation “we want to have a point of view on what we can and cannot do,” Nigam said, asserting that what the nation consumes should not be controlled by one or two companies.

“We will offer an alternative with a different set of features and capabilities. Some will like us, some will like others. We are just asking for a chance to compete on merit,” he said, noting that competition will push Google to improve its Play Store terms in India.

PhonePe is making the app store available to download to consumers from its website (users will have to sideload the app on their Android phones). Nigam said the firm is in talks with “multiple” phone makers and hopes to close some deals in the coming weeks.

Indus Appstore also “promises” to offer broader transparency — including explaining to the developers when their app has been pulled, and why.

PhonePe plans to monetize Indus Appstore by charging developers for app discovery and by showing ads, Nigam said.

Ivan Mehta contributed to this report.

Close-Up Of Gemstones in the hands of a woman

Google launches two new open LLMs

Close-Up Of Gemstones in the hands of a woman

Image Credits: Catherine Delahaye / Getty Images

Barely a week after launching the latest iteration of its Gemini models, Google today announced the launch of Gemma, a new family of lightweight open-weight models. Starting with Gemma 2B and Gemma 7B, these new models were “inspired by Gemini” and are available for commercial and research usage.

Google did not provide us with a detailed paper on how these models perform against similar models from Meta and Mistral, for example, and only noted that they are “state-of-the-art.” The company did note that these are dense decoder-only models, though, which is the same architecture it used for its Gemini models (and its earlier PaLM models), and that we will see the benchmarks later today on Hugging Face’s leaderboard.

To get started with Gemma, developers can get access to ready-to-use Colab and Kaggle notebooks, as well as integrations with Hugging Face, MaxText and Nvidia’s NeMo. Once pre-trained and tuned, these models can then run everywhere.

Contact us

You can contact Frederic via Signal at +1 (860) 208-3416 ‬, or by email at [email protected]. While Google highlights that these are open models, it’s worth noting that they are not open source. Indeed, in a press briefing ahead of today’s announcement, Google’s Jeanine Banks stressed the company’s commitment to open source but also noted that Google is very intentional about how it refers to the Gemma models.

“[Open models] has become pretty pervasive now in the industry,” Banks said. “And it often refers to open weights models, where there is wide access for developers and researchers to customize and fine-tune models but, at the same time, the terms of use — things like redistribution, as well as ownership of those variants that are developed — vary based on the model’s own specific terms of use. And so we see some difference between what we would traditionally refer to as open source and we decided that it made the most sense to refer to our Gemma models as open models.”

That means developers can use the model for inferencing and fine-tune them at will and Google’s team argues that these model sizes are a good fit for a lot of use cases.

“The generation quality has gone significantly up in the last year,” Google DeepMind product management director Tris Warkentin said. “Things that previously would have been the remit of extremely large models are now possible with state-of-the-art smaller models. This unlocks completely new ways of developing AI applications that we’re pretty excited about, including being able to run inference and do tuning on your local developer desktop or laptop with your RTX GPU or on a single host in GCP with Cloud TPUs, as well.”

That is true of the open models from Google’s competitors in this space as well, so we’ll have to see how the Gemma models perform in real-world scenarios.

In addition to the new models, Google is also releasing a new responsible generative AI toolkit to provide “guidance and essential tools for creating safer AI applications with Gemma,” as well as a debugging tool.

Whoopi Goldberg–backed BLKFAM launches with 1,000+ hours of kids’ programming

Image Credits: BLKFAM

BLKFAM is a free, ad-supported, Black-focused family streaming service launching today to give Black Americans access to more than 1,000 hours of new kid-friendly animation titles, as well as content for the entire family, including sitcoms and reality shows, fitness and wellness series, news, music-driven content and more.

Whoopi Goldberg is an equity investor in BLKFAM and will also be the creative director. There are 10 new original live-action and animated series currently in development, which are expected to roll out throughout 2024.

“I’m excited to be involved with BLKFAM from the ground floor. I’m proud to be a part of something my great-grandkids can enjoy, and see characters who look like them onscreen, created by people who look like them, for them,” Goldberg said in a statement provided to TechCrunch.

BLKFAM considers its platform the first and only Black-owned and Black-focused family streaming service. The company’s goal is to serve Black audiences who often feel misrepresented or underrepresented in mainstream media even though it’s the demographic that watches the most TV. According to Nielsen, Black audiences consume over 81 hours of media on a weekly basis. Plus, 80% of Black Americans use free over-the-top services, compared to 69% of all consumers, per Horowitz Research.

BLKFAM aims to release content that “entertains, educates, and celebrates Black American family-friendly content,” the company wrote in its press release. Instead of following rival streamers that take the “more is better” approach, BLKFAM is curating a collection with inclusive stories that hopefully spark discussion from the audience. The company has teamed up with content partners Playwatch Kids and Candle Media’s ATTN to deliver educational shows like “Kembe,” “Gabby Galactic and the Orbiteens” and “12th Street,” among others.

The company also believes its co-viewing strategy — providing content that appeals to the entire family rather than individual age groups — will help it stand out from competitors.

“Professionally deeply understanding the streaming landscape, I knew there was a gap in co-viewing platforms,” founder Larry Adams, former chief brand officer for HBO Max, tells us. “But as a Black parent, when my kids and I would sit down to watch a movie or share a show, the options for families to watch together were mixed together with adult-oriented content. The content that was targeted to Black audiences was not necessarily created from the point of view of wholesome family co-viewing.”

“We know that families watch together and only resort to separated viewing experiences when there are limited options,” adds Adams, who also helped launch DirecTV Now in 2016, AT&T’s live TV streaming service.

BLKFAM is now available on Amazon Prime Video Channels, Amazon Fire TV, Apple TV, Roku, YouTube TV, Samsung Smart TVs, Vizio, LG, and iOS and Android devices.

Founded by actors, Mansa brings its free streaming service for global Black culture out of stealth

Superhuman Instant reply feature

Superhuman launches AI-powered instant replies

Superhuman Instant reply feature

Image Credits: Superhuman

Email startup Superhuman is launching an AI-powered instant replies feature today with the company looking to double down in AI-fueled functionality this year. The feature gives you three contextual replies as options to quickly respond to an email.

Superhuman has been built around the idea to get you to inbox zero in the fastest way. Instant replies are one more feature to quickly get emails out of your way. You can think of Instant replies as equivalent to Gmail’s smart replies, but Superhuman said its feature is more contextual and personal.

The core concept is the same though. You will get three potential reply options when you hit reply to an email. While the prompts are short, when you click on any of the options, Superhuman writes out a full-sentence reply to you.

As the demo video shows below, when you tap on “Interested” to an email asking you to speak on a panel, the client types out a full sentence with the sender’s name. You can press the tab to switch between replies and press enter to send one. The company said that users who were part of the beta test of instant replies are sending emails twice as fast.

Adding AI to email

Superhuman has already launched a few AI-powered products. Last year, it launched Superhuman AI, which helps users draft better emails with options to change the length or tone.

In January, the company extended the functionality to iPhone and iPad apps as well. In November, the startup launched an auto-summarize feature, which shows a one-line summary on top of your emails.

Some of Superhuman’s competitors have also included AI-powered features in email workflows. Shortwave, a startup founded by ex-Googlers, has launched things like asking AI questions about emails, instant summaries and a way to use AI to reply to emails quickly. Readdle’s Spark email client rolled out an AI assistant last year. Major tech companies like Microsoft and Google have introduced AI-powered features for Outlook and Gmail.

Superhuman co-founder and CEO Rahul Vohra told TechCrunch that the key differentiator for the company in terms of AI-powered features is its stickiness. He noted that an average user uses AI 25 times a week to write emails.

Superhuman’s roadmap

Superhuman says that the first phase of the AI-powered product was having an assistant for the email; the second phase is to have AI work automatically without invoking it with features like auto-summary and instant replies; and the third phase will replace certain parts of a user’s daily workflow like using an AI agent for scheduling or writing full emails on their behalf.

Vohra said that the company could be profitable anytime it wants and it is a strategic decision. The company’s CEO mentioned that it has five years of runway. So a new funding round is not something he is thinking about on a “day-to-day” basis.

Superhuman has kept its price of $30 per month consistent since the launch. But Vohra mentioned that the company might revisit its pricing later this year.

He said that apart from investing in AI products, the company is focusing on making its offering for teams stronger.

“Imagine instead of constantly forwarding or writing an email, internally you can just share a thread. And then on that shared thread, you can put comments and reactions. We are thinking about transforming email from something that’s pure productivity into a platform for smart collaboration. And that’s a big deal because historically, email has been single-player. It was designed in an era where tools were inherently not collaborative,” Vohra said.

The company is also thinking about building special edition email clients starting with Superhuman for sales directed toward account executives and customer managers with integrations like Salesforce and HubSpot.

New Speechify iOS release

Text-to-speech app Speechify launches Gmail integration and voice cloning

New Speechify iOS release

Image Credits: Speechify

Text-to-speech startup Speechify is launching a new version of its iOS app (v3.0) with features like a redesigned home page, the ability to scan documents, Gmail and learning platform Canvas integration, and an explore page for in-app content.

Speechify is revamping its homepage to provide shortcuts to import files with options like iOS Files, Google Drive, Kindle, Gmail, Text, Scan and Link. The new update has a scan feature, which lets users scan a physical document and import the text to listen to it later. Users can simultaneously scan two pages of a book as well.

Speechify revamped home screen
Image Credits: Speechify

The remodeled home page also has an Apple Fitness Ring-styled daily reading goal, which you can adjust.

The company is introducing a feature to let users read their emails by integrating their Gmail accounts. The app can also help users import their PDF attachments into the Speechify app. The startup is also launching a Canvas integration with the learning management system so students can access their homework assignments. Plus, there is a new iCloud integration that automatically brings your files to Speechify.

Speechify is taking advantage of interactive widgets features in iOS 17 and launching new widgets to easily import sources, track reading progress and continue listening to an imported document. The new app will also let users listen to tracks offline in premium voices.

The company said that the app now has new pre-set AI voices with improved speech models. It has also updated the celebrity voices of Snoop Dogg and MrBeast. Users can also get an AI-powered summary of a document and article, and the app can read it out loud.

Another AI-powered feature that is part of Speechify 3.0 is users’ ability to clone their voice and have it read out text. Last year, Apple released an accessibility-related feature called Personal Voice, which creates a voice that sounds like you. Earlier this month, Clubhouse also released a feature that reads your chats in your voice.

Image Credits: Speechify

Speechify says it has 23 million users when you are signing up on the app. The company didn’t clarify if this number was indicative of registered users or active users. The startup has forged partnerships like Instagram co-founders’ app Artifact, which was shut down last month, and Medium. The company’s co-founder and head of AI Tyler Weitzman said that it plans to release a public API of its text-to-speech solutions for more business-related partnerships.

Apart from developing support for external reading sources, the company has made a redesigned explore page to put its own content in areas like productivity, self-improvement and learning differences. This content is in a format akin to Instagram Stories.

Image Credits: Speechify

The company said it observed that once users listen to around 500 words on Speechify, they are likely to love the service. With the preloaded content, the company wants to give users a head start. The startup noted that it wants to bring more content to the app. It notably allows users to read content on different websites through an in-app browser. Plus, it has a separate audiobook business vertical.

Speechify has different types of competitors in the space. For instance, read later app Pocket, which is owned by Mozilla, has an in-app text-to-speech feature. Last year, The New York Times launched its own audio app based on the Audm product it acquired in 2020. Speechify wants to stand out in this market and wants to become a one-stop shop for anyone thinking about reading experiences by offering all kinds of integrations.

Servers in dark data center room with computers and storage systems

Alphabet spin-off SIP launches Verrus, a data center concept built around battery 'microgrids'

Servers in dark data center room with computers and storage systems

Image Credits: Jasmin Merdan / Getty Images

Sidewalk Infrastructure Partners (SIP) — the Alphabet spinout that focuses on building and backing new approaches to complicated infrastructure problems in areas like power, broadband and waste management — has launched its latest project, a new concept for more flexible data center energy management called Verrus.

Verrus incorporates “microgrids” based on advanced, high-power batteries with software to understand and allocate energy to specific tasks and applications, and it is designed to address some of the power challenges posed by modern computing needs. These include peaks of cloud computing usage and larger projects — such as AI training — that could be bundled into batches distributed across time frames where there is less demand.

Jonathan Winer, the co-founder and co-CEO of SIP, said that the first three data centers designed using Verrus’ architecture will likely be located in Arizona, California and Massachusetts.

The aim is to have these operational in 2026 or 2027. There are as of yet no signed customers, although Winer said that a number of “hyperscalers” — one of whom, Alphabet, remains a primary backer of SIP after spinning it out several years ago — have shown interest in the project for when it does come online and would likely become one of its target segments when seeking investment. (Alongside the new business, SIP is also launching the Data Center Flexibility Initiative to bring stakeholders like energy companies, tech giants and regulators together in the meantime.)

Winer described Verrus as having “gigawatt scale ambitions.” A data center to meet that scale, he estimated, could cost $1 billion to put together, with hundreds of millions of dollars of equity needed to get it off the ground. That is only likely to come after building has started and customers begin to sign up, he said. In addition to Alphabet, others that back SIP currently include Ontario Teachers’ and StepStone.

Winer said that SIP has been developing the project in stealth mode for almost two years already, and that it was an offshoot of other research that it had been doing into electricity grid management, coupled with SIP’s work with companies focused on load shifting to better manage energy consumption. Observing the strain that data centers in particular have on the electrical grid, SIP turned its attention to those data centers themselves.

The explosion of cloud computing and AI data computations present “a real challenge on the grid,” he said, and typically data centers are at capacity. “In order to add what we’re going to need both of the AI challenge and just general cloud compute there’s going to have to be a new approach to energy management,” he noted. Simply building more data centers, whether run by third-party data center operators or by the hyperscalers themselves, will not keep up with demand. 

Today, extra power comes from diesel generators and using redundant electrical systems within data centers themselves. Verrus’ proposal is to instead use what Winer refers to as a “micro grid” that will include a high-capacity battery, which will mean it will be more flexible to deploy power to specific areas or even projects within a data center.

In turn, this means that an AI training job, as one example, could effectively be paused and batched and run at a different time, versus, say, an enterprise cloud service that might require “five nines” demand response.

As SIP sees it, simply adding more data centers — which has been the approach up to now — is not a sustainable approach longer term.

“The challenge with adding data centers to the grid is not the 340 days a year that the grid isn’t maxed out. The grid is very happy to supply power on the days when it’s not at capacity,” he said. “The real challenge is the 20 days a year where for few hours a day they can’t serve the load.” A flexible energy management system would allow for what he described as “islands” in the data center during those hours.

The approach that Verrus proposes to take underscores how energy usage and consumption continue to evolve in the tech world, but also how energy continues to be a persistent, expensive and ultimately resource-intensive issue that needs as much focus and innovation as the software and hardware that rely on it to evolve.

Verrus is not the only tech company that is eyeing up how to build and use super-capacity battery architecture to manage electricity distribution. Instagrid, a startup out of Germany, recently raised funding for its startup, building batteries to help users with power management in use cases where they are off the grid altogether.

Cypher’s inventory drone launches from an autonomous mobile robot base

Cypher’s inventory drone

Image Credits: Brian Heater

Doing inventory sucks. I feel the need to reiterate this any time I discuss the topic here. Having performed a bit of it during my years of retail work, I can personally attest to the fact that it’s a mind-numbing bore. It is, therefore, a prime candidate for automation. After all, two of the main things robots are good at are counting and doing the same job over and over again without complaining.

The halls of Modex this week are lousy with solutions to the problem of inventory. It’s a particularly tricky issue when it comes to warehouses, due to both the big numbers that need to be catalogued and the (literal) heights that have to be reached to collect those figures. Recent years have given us drone solutions and Dexory’s massive AMR (autonomous mobile robot) with a telescoping scaffolding structure.

A new system unveiled by Cypher Robotics this week at Modex splits the difference. At its core is a tall (but not Dexory tall) AMR that serves as a launching and landing pad for a drone. The idea is pretty simple: The mobile platform moves the across the warehouse floor, and then the drone takes off, giving it a lot of vertical reach.

I encountered the drone at GreyOrange’s booth. The robotics firm has been increasingly embracing third-party robotics in its bid to fully automate warehouses. The company tells me the Captis robot was added to its offering at the behest of a customer who needed cycle counting, which involves cross-checking inventory records with what’s actually on the shelves.

You can’t really see it from the image I took, but the drone is actually tethered to the inside of the AMR platform. When it lifts off, it exposes a hole with the cable inside. Hard wiring means you can eke out far more battery life than untethered systems. The Ottawa-based firm says the drone can operate for up to five hours before needing to return to the base for a recharge.