A reminder for new readers. That Was The Week collects the best writing on critical issues in tech, startups, and venture capital. I select the articles because they are of interest. The selections often include things I disagree with. The articles are only snippets. Click on the headline to go to the original. I express my point of view in the editorial and the weekly video below.
Content this week from
@kwharrison13, , @alexhern, , @RupertNeate, @adhutchinson, @OpenAI, @jennadillardtx, @KalleyHuang, @SarithaRai, @a16z, @IlyaStrebulaev, @CristinaCriddle, @GeorgeNHammond, @hugh_son, @SubstackInc
Editorial – The Year of the Mask (again)?
Introducing The Information’s Generative AI Database – The Information
AI Canon – A16Z
What is the total value of unicorn exits for US universities? Ilya Strebulaev
Apple MR Headset
Editorial: The Year of the Mask (again)?
For each of the past three years, there have been predictions that Apple was about to release a mixed-reality face mask, oops, sorry, mixed-reality headset.
The rumors are stronger this year and seem plausible; although I am still slightly skeptical that Apple will go in this direction, I am ready to stand corrected if true.
My skepticism is that of a product thinker. Apple has been a world leader in building human-centric products. Products for the ears (airpods and music services), eyes (phones, screens, and video services), mouth (microphones with echo cancellation), and hands/arms (watch, keyboards, mice, phones). It has not yet addressed the brain or the legs (car?).
To release a single device that covers the eyes and head and serves the ears and enables the mouth (microphone), and requires solitude seems counter to its past decisions. If it does, I hope for a compelling problem that this solves. I do not see it.
Meta (Thursday morning Pacific time) announced that Quest 3 is a $399 mask positioned as a competitor, adding credibility to the rumors of Apple’s plans. And it reduced the price of Quest 2 to $250. Rumors say Apple’s device will be over or about $3000.
If it comes, I will probably buy one (I just have to know what it does 😉 ), but I suspect it will soon sit idly on a shelf because there is nothing I want to use it for. Gaming is the only real use case, single and multi-user. And as it is mixed reality, it can extend to the real world, placing images and action there. Meta’s new device promises the same. And then there are enterprise uses.
If Apple can cannibalize the Xbox and Playstation markets, then this may be a large enough business to move the needle for them, but again I am skeptical. And Meta’s failure to get traction, at any price, on the Oculus-based headsets suggests I am right.
But this is Apple. If they do it, I will probably be wrong because that will suggest they have figured out some use case and model that will work. It will be interesting to watch.
The second thing that grabbed my attention this week was a couple of articles about the use of AI in investing. JP Morgan trademarked IndexGPT, hinting at an AI-led public markets portfolio. And the FT penned a piece about the rise of venture investments leveraging AI. Towards the end, they quote Anne Glover, chief executive at venture capital firm Amadeus Capital Partners.
She said generative AI tended towards bias, adding that the tools used limited and historical data:
“It is impossible to assume that you should be making that kind of human decision based on what an AI is doing,” Glover said. “For someone like ourselves, where we are investing at the cutting edge, there isn’t a lot written about what we are looking for.”
This week, Moonfire started investing out of an AI-aided appraisal process, and Angel List announced that its quantitative investing fund was doing likewise.
Most of these approaches in venture use company-specific data for predictions. For example, profiles of founders or hires. That is why Anne Gover’s instinct makes sense. Company-level data tends to be very poor as a predictor of outcomes until much later in the company’s life. SignalRank has a fundamentally different approach.
Here is the table of companies SignalRank AI selected at the Series B round in 2023. It shows the company, the investors, the amount raised, the implied valuation (estimated), and the number of investors. Only 29 of the 646 Series Bs in 2023 were selected. Based on backtests, we can predict an average return of 600% at the 5-year mark and 20-30% unicorns.
AI in venture capital is definitely real at the selection stage.
Essays of the Week
Why Do People Hate Cathie Wood So Much?
KYLE HARRISON, MAY 27, 2023
Everyone has someone (or several someones) who they just do not like. Hearing their voice makes you grind your teeth. Every time they say anything, or do anything, you hate it and everything associated with it. Sometimes it starts to feel like part of your personality is just detesting that person and everything they stand for.
My Mom has that feeling for a particular famous person. Former Secretary of State, blonde hair, loves pants suits. You know who I’m talking about? That’s right. Saturn-award winning actress Tea Leoni (who played the Secretary of State in the TV show Madam Secretary).
The kind of frustration that we feel with certain people is, I think, pretty natural. Sometimes it can be explained (like my Mom saying Tea Leoni is an unemotional chain smoker that doesn’t move her eyebrows). Sometimes it’s just instinctual; we can’t put our finger on why we’re so filled with distaste.
In particular, for individuals who have built their persona in a marketing-rich way, people’s reactions are often even more polarized. Think about politicians (who are basically just boomer marketers), or personalities like Jim Cramer, Tai Lopez, Joe Rogan. People either love them or hate them, there’s very little in-between.
Cathie Wood is no different. As an investor, she’s an incredibly polarizing figure. People are convinced she’s either a misunderstood genius or the dumbest person to ever be allocating billions of dollars. One of the things most people struggle to do with polarizing characters is to separate some valuable takeaways or lessons from what they do, while still being able to acknowledge other aspects of what they do are ridiculous.
I’m not going to unpack the myriad of ways that what ARK Invest does is bonkers. There are much better analysts and shitposters who have done that (and I’ll quote some of them). Instead, I just wanted to spend time reflecting on how something that sounds so interesting and compelling to me on paper can end up being so ridiculous in practice.
Because the concept of ARK Invest on paper really does sound so compelling to me: a research-driven long-term investor focused on transformational technology, that then democratizes access to that thesis through “innovation ETFs” and very public-facing research; open to criticism and learning. Where does it go wrong?
Asset manager Fidelity reports that stake worth $20m in October 2022 is now worth just under $6.6m
Wed 31 May 2023 12.55 EDT
Twitter’s value has plummeted by almost two-thirds since Elon Musk acquired the company in October 2022, one of the social media company’s only remaining external investors has admitted.
Fidelity, an asset manager that held a stake in Twitter worth about $20m after Musk acquired the business for $44bn, said in a corporate filing that its stake was now worth just under $6.6m. That would value the overall company, now officially called X Holdings Corp after Musk’s early venture X.com, at just $14.75bn.
The fund disclosed its holdings in its quarterly reporting on the performance of its blue chip growth fund, which invests in a range of companies in the US and around the world, focusing on household names with stable valuations. It also owns a $386m stake in Musk’s privately held rocket company, SpaceX, and another $849m in Tesla, which is publicly traded.
Twitter’s valuation is of particular interest to the company’s staff, many of whom were employed while it was publicly traded with compensation that included stock options. Musk offered to value those options at about $20bn in March, according to a Wall Street Journal report, an acceptance that the value of the company had fallen by at least half since he took over.
That valuation came with an incentive for Musk to lowball the estimate, however, because the higher it was, the more expensive employee remuneration would be. Fidelity, by contrast, has no such motivation for slashing the valuation further still.
Maverick head of Tesla and Twitter overtakes French billionaire Bernard Arnault, head of the LVMH luxury goods empire
Thu 1 Jun 2023 12.06 EDT
Elon Musk is once again the world’s richest person after leapfrogging French billionaire Bernard Arnault, after a slump in the value of Arnault’s LVMH luxury goods empire.
The 51-year-old maverick head of Tesla and owner of Twitter has seen his fortune recover to $192bn (£153bn) – up $55bn from the start of the year – while Arnault’s wealth has fallen by $5bn in the past 24 hours to $187bn according to Bloomberg’s daily updated billionaires’ index.
Arnault had overtaken Musk as the world’s most wealthy person in December last year after the value of LVMH shares soared amid booming demand for its luxury products. At the same time, Musk’s wealth had fallen as Tesla’s share price slid on investor fears that he might be preoccupied with his recent purchase of Twitter.
Arnault’s wealth then continued to soar, topping $200bn in April as the value of LVMH – which owns brands including Louis Vuitton, Christian Dior and Moët & Chandon champagne – hit a record high. However, since then analysts’ concerns that the luxury goods bubble may be bursting have sent LVMH shares falling by more than 10%, depressing the estimated value of Arnault’s fortune.
The 74-year-old, who co-founded the luxury goods group 35 years ago and is its majority shareholder, recently appointed his children to key roles within the business. In January, his eldest child, Delphine, was named the head of Christian Dior, the second biggest brand in the empire. Her brother Antoine was promoted to run the holding company that controls LVMH and the Arnault family fortune.
Musk’s fortune is directly linked to the share price of Tesla, the electric carmaker, of which he owns about 13%. Tesla’s shares have risen 88% since the start of the year.
Published May 30, 2023
By Andrew Hutchinson, Content and Social Media Manager
One of the key questions posed early on around Elon Musk’s ‘free speech’ led takeover at Twitter was: ‘What will Elon do when government requests come from regions where Musk’s other business, Tesla, has a vested interest?’
Tesla, for example, is heavily reliant on China for both sales and parts manufacturing, while it’s also looking to expand its presence in India and Turkey. The governments in each of these regions have regularly sought to restrict speech, on Twitter and in other forms, which has put them at odds with social platform management in the past.
So how has Twitter 2.0 handled the same?
As it turns out, Elon has opted to comply with pretty much every government request, with a recent report showing that Twitter is now complying with such at a much higher rate than previous Twitter management.
As reported by El Pais:
“Since Musk’s takeover, the company has received 971 requests from governments (compared to only 338 in the six-month period from October 2021 to April 2022), fully acceding to 808 of them and partially acceding to 154. In the year prior to Musk taking control, Twitter agreed to 50% of such requests, in line with the compliance rate indicated in the company’s last transparency report (none have been published since October 2022). Following the change of ownership, that figure has risen to 83%.”
In response, Musk has said that Twitter has no choice but to comply with such, as these are governed by the laws in each region, and Twitter will always align with local laws.
But as noted, Twitter has held firm on such requests in the past. In June last year, for example, the Indian Government called on Twitter to ban various Pakistani-linked profiles due to government dissent. Twitter challenged those bans in court, putting it in conflict with the Indian government – which also put the app at risk of a local ban, something that’s been held over Twitter various times in the past, when it’s refused to cede to requests from the Indian government.
Twitter was briefly banned in Turkey in 2014 for the same, after refusing to comply with censorship requests from the Turkish government, which has since seen Turkish authorities implement new laws on what social platforms can and cannot publish.
So while Musk’s right that Twitter is aligning with local authorities, it does have the capacity to take a stronger stance on such, especially where such requests are in opposition to local laws, which some of these more recent requests have reportedly been. And given Musk’s public campaigning for free speech, you might expect Twitter 2.0 to be more inclined to fight back for the rights of its users – but thus far, that doesn’t appear to be the case.
Of course, Twitter doesn’t have to. As Musk notes, the risk here is that if Twitter doesn’t comply, it could face bans in each region, and in most cases, according to Elon, it’s better to censor a little content in order to ensure ongoing access for all users.
Video of the Week
AI of the Week
We’ve trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”). In addition to boosting performance relative to outcome supervision, process supervision also has an important alignment benefit: it directly trains the model to produce a chain-of-thought that is endorsed by humans.
May 31, 2023
In recent years, large language models have greatly improved in their ability to perform complex multi-step reasoning. However, even state-of-the-art models still produce logical mistakes, often called hallucinations. Mitigating hallucinations is a critical step towards building aligned AGI.
We can train reward models to detect hallucinations using either outcome supervision, which provides feedback based on a final result, or process supervision, which provides feedback for each individual step in a chain-of-thought. Building on previous work1, we conduct a detailed comparison of these two methods using the MATH dataset2 as our testbed. We find that process supervision leads to significantly better performance, even when judged by outcomes. To encourage related research, we release our full dataset of process supervision.
Process supervision has several alignment advantages over outcome supervision. It directly rewards the model for following an aligned chain-of-thought, since each step in the process receives precise supervision. Process supervision is also more likely to produce interpretable reasoning, since it encourages the model to follow a human-approved process. In contrast, outcome supervision may reward an unaligned process, and it is generally harder to scrutinize.
In some cases, safer methods for AI systems can lead to reduced performance3, a cost which is known as an alignment tax. In general, any alignment tax may hinder the adoption of alignment methods, due to pressure to deploy the most capable model. Our results below show that process supervision in fact incurs a negative alignment tax, at least in the math domain. This could increase the adoption of process supervision, which we believe would have positive alignment side-effects.
Solving MATH problems
We evaluate our process-supervised and outcome-supervised reward models using problems from the MATH test set. We generate many solutions for each problem and then pick the solution ranked the highest by each reward model. The graph shows the percentage of chosen solutions that reach the correct final answer, as a function of the number of solutions considered. Not only does the process-supervised reward model perform better across the board, but the performance gap widens as we consider more solutions per problem. This shows us that the process-supervised reward model is much more reliable.
We showcase 10 problems and solutions below, along with commentary about the reward model’s strengths and weaknesses.
May 26, 2023
As every investor and their French Bulldog scrambles to get in on the AI action, we look to the data for answers.
Ironically, we can’t even define “AI” as a sector per se since almost every startup looking for some coin or decent press suddenly identifies as an “AI-centered-something-or-other.”
But for the purists, the numbers don’t lie — or in the very least they tell a clearer story than the prevalent “AI will save the world” narrative.
In fact, early this week investors poured $700 million into two AI startups — Builder.ai and Anthropic — and followed up mid-week with another $105 million to AI marketing platform Insider. It seems we’ve reached another level of the AI craze that has dominated the private markets since late last year.
Overall, $20 billion has been raised by startups using “AI” in 2023. (Fun fact: $20 billion could pay the salaries of 307,266 teachers for one year.) But let’s definitely keep making those AI-generated selfies.
So much buzz
Are startups worried about the downturn? We guess it depends on who writes their checks.
In fact, Bessemer Venture Partners, one of the oldest and more established venture firms in the U.S., earlier this year said it is earmarking $1 billion of its most recent fund solely for investments in artificial intelligence.
And that’s just one firm.
This week we published an interview with Bessemer partner Sameer Dholakia, who aptly said of the AI movement: “Literally trillions of dollars of value gets created when you have these massive tectonic shifts.”
And what about AI IPOs?
But it’s not all unicorns and rainbows for AI. Just because funding to the sector is hot, that doesn’t mean the appetite on the public markets is at the same level.
If we look at public markets (and we did) it’s clear that an AI focus hasn’t been a recipe for stock market gains. This is evident looking at recent performance of the most highly valued AI-oriented companies to go public in the quarters leading up to the market peak.
By Kalley Huang
June 1, 2023 9:54 AM PDT
OpenAI’s ChatGPT launched six months ago, igniting a boom in generative artificial intelligence. Since then, all manner of startups have emerged, building technology to compete with ChatGPT and developing services that use generative AI. Despite an otherwise cool funding environment, investors are jockeying to join the action. The most recent sign: The Information reported Wednesday that Google led a roughly $100 million round in video-generation company Runway.
Runway is one of 39 startups in The Information’s Generative AI Database. Collectively, they have raised more than $15 billion from more than 50 investors. Our database includes information about who provides the AI models and the cloud services that undergird these companies. It also tracks their revenue, valuation and employee count. Our data comes from our reporting and PitchBook, which provides private market data.
• The release of ChatGPT in November ignited a boom in generative artificial intelligence
• The Information’s Generative AI Database lists 39 startups that create AI models or build services on them
• Collectively, the startups in the database have raised more than $15 billion from more than 50 investors
Here are three high-level takeaways from The Information’s Generative AI Database. We will update it regularly with more companies and more details about their trajectory.
OpenAI Is Central to the Generative AI Ecosystem
OpenAI has raised $11 billion from investors including Microsoft and Khosla Ventures and is valued at $27 billion. It has raised the most money and is the most valuable company in our database. But OpenAI also invests in other startups and provides the models their products are built on.
OpenAI has provided models for at least 14 startups in our database. For example, Glean.ai uses models from OpenAI in its service to manage accounts payable. Its CEO, Howard Katzenberg, said the company plans to develop its own large-language model this year.
OpenAI has backed at least four companies in our database. In November alone, it led a $50 million round in Descript, which creates tools to edit audio and video; a $27 million round in Speak, a language-learning app; a $23.5 round in Mem, a workplace and note-taking app; and a $5 million round in Harvey, which builds large-language models for lawyers.
Coatue Management and Khosla Ventures have each invested in four companies in our database. Tiger Global Management, Lux Capital and Insight Partners are the next most common investors, with three investments each…..
Venture firms scrutinize startups to identify weak links
Homework-help provider Chegg plunged after disclosing AI risk
May 25, 2023 at 4:55 PM PDT
Global investors at the world’s largest venture capital firms including Accel and Sequoia Capital India are asking their portfolio companies the same question: How vulnerable are you to artificial intelligence?
The chaotic power of AI was on display earlier this month when online tutoring startup Chegg Inc. saw its stock price slump 48% because of competition from OpenAI chatbot ChatGPT. Now, venture capitalists around the world are evaluating their investments to find out whether AI tools could upend their industries. Chegg’s stock plunge further accelerated such reviews.
“An internal team at Accel is very deeply focused on this,” said Barath Shankar Subramanian, an India-based partner at the firm, discussing AI’s impact to its portfolio of more than 400 startups in South and Southeast Asia. The team is “spending time on the ground both in India and outside,” meeting founders, he said over video. Such an exercise is also ongoing at Accel’s home base in Silicon Valley, he said.
The reviews underscore how generative AI has grabbed the attention of investors, founders and executives, who see it both as a massive opportunity as well as disruptive force. While VC firms are trying to identify the big winners, they’re also trying to weed out the ones whose business models risk being rendered outdated by the new technology.
Sequoia Capital, which similarly has invested in more than 400 startups in India and Southeast Asia, is taking the AI risk “very seriously,” said Anandamoy Roychowdhary, a Singapore-based partner of Sequoia’s Surge, an accelerator for startups in India and Southeast Asia, and the fund’s specialist on AI and deep tech.
The firm has gone through each of its early-stage investments in the region to assess the startups’ preparedness. Meanwhile, more than 75% of new deals made by Sequoia India, the firm’s unit in the region, are related to AI, with the topic dominating its investment meetings.
Table of contents
Research in artificial intelligence is increasing at an exponential rate. It’s difficult for AI experts to keep up with everything new being published, and even harder for beginners to know where to start.
So, in this post, we’re sharing a curated list of resources we’ve relied on to get smarter about modern AI. We call it the “AI Canon” because these papers, blog posts, courses, and guides have had an outsized impact on the field over the past several years.
We start with a gentle introduction to transformer and latent diffusion models, which are fueling the current AI wave. Next, we go deep on technical learning resources; practical guides to building with large language models (LLMs); and analysis of the AI market. Finally, we include a reference list of landmark research results, starting with “Attention is All You Need” — the 2017 paper by Google that introduced the world to transformer models and ushered in the age of generative AI.
A gentle introduction…
These articles require no specialized background and can help you get up to speed quickly on the most important parts of the modern AI wave.
Software 2.0: Andrej Karpathy was one of the first to clearly explain (in 2017!) why the new AI wave really matters. His argument is that AI is a new and powerful way to program computers. As LLMs have improved rapidly, this thesis has proven prescient, and it gives a good mental model for how the AI market may progress.
State of GPT: Also from Karpathy, this is a very approachable explanation of how ChatGPT / GPT models in general work, how to use them, and what directions R&D may take.
What is ChatGPT doing … and why does it work?: Computer scientist and entrepreneur Stephen Wolfram gives a long but highly readable explanation, from first principles, of how modern AI models work. He follows the timeline from early neural nets to today’s LLMs and ChatGPT.
Transformers, explained: This post by Dale Markowitz is a shorter, more direct answer to the question “what is an LLM, and how does it work?” This is a great way to ease into the topic and develop intuition for the technology. It was written about GPT-3 but still applies to newer models.
How Stable Diffusion works: This is the computer vision analogue to the last post. Chris McCormick gives a layperson’s explanation of how Stable Diffusion works and develops intuition around text-to-image models generally. For an even gentler introduction, check out this comic from r/StableDiffusion.
Foundational learning: neural networks, backpropagation, and embeddings
These resources provide a base understanding of fundamental ideas in machine learning and AI, from the basics of deep learning to university-level courses from AI experts.
Deep learning in a nutshell: core concepts: This four-part series from Nvidia walks through the basics of deep learning as practiced in 2015, and is a good resource for anyone just learning about AI.
Practical deep learning for coders: Comprehensive, free course on the fundamentals of AI, explained through practical examples and code.
Word2vec explained: Easy introduction to embeddings and tokens, which are building blocks of LLMs (and all language models).
Stanford CS229: Introduction to Machine Learning with Andrew Ng, covering the fundamentals of machine learning.
Stanford CS224N: NLP with Deep Learning with Chris Manning, covering NLP basics through the first generation of LLMs.
Tech deep dive: understanding transformers and large models
There are countless resources — some better than others — attempting to explain how LLMs work. Here are some of our favorites, targeting a wide range of readers/viewers….
News Of the Week
Ilya Strebulaev, Stanford
Our data covers 1,110 US-based VC-backed companies that became unicorns between 1997-2021 and 2,975 of their founders. For each university, we calculated the total combined value of all exited sample companies at exit time. A company is associated with a university if at least one founder graduated from that university (either undergraduate or graduate).
Stanford University is first at around $400 billion, followed by Harvard University at $300 billion and Massachusetts Institute of Technology at $230 billion.
University of California, Berkeley, UCLA, and Cornell University follow behind, with $160 billion, $130 billion, and $130 billion, respectively.
Note that this takes into account only the exited unicorns (which is about half of our sample) at their exit time and excludes companies that are currently still private.
Thank you to the Stanford University Graduate School of Business Venture Capital Initiative for support.
Note: Exit means going public, sale, liquidation, or bankruptcy.
Accountancy firm KPMG, hedge fund Coatue and VC firm Headline among those incorporating the technology
Venture capital funds, private equity groups and accountancy firms are using the latest artificial intelligence to pick acquisition targets and start-ups for investment, betting the technology can give them an edge over rivals.
Big Four accountant KPMG, hedge fund Coatue and venture capital firm Headline are among those using the latest AI tools to advise clients and help guide their dealmaking.
With investors under pressure to identify the next high-growth start-up at a time when few companies are going public, some argue that dealmakers can benefit from using generative AI for tasks such as assessing a company’s growth potential based on financial analysis.
“If you can train or use a model that gets a lot of efficiency first, you will get an advantage in that particular area of the business that is harder for a second mover to do,” said Pär Edin, who leads innovation in KPMG’s US deal advisory and strategy business. “It is about getting there first for each and every particular use case.”
The pace of artificial intelligence development over the past six months, triggered by the release of OpenAI’s popular ChatGPT — a chatbot that provides humanlike answers to queries — has spurred investors to use the tools to identify fast-growing companies and acquisition targets.
KPMG has used the technology behind ChatGPT to create a system based on its own data to help advise its staff. The company said the tool had seen high take-up over the course of the month it was in use, adding that recent advances in AI had made it “practically useful . . . particularly in M&A”.
Coatue’s software Coatue Brain integrates generative AI into its data platforms, using the technology to sift through sellside research, earnings transcripts and pitch decks to extract and condense key points into clear and concise briefings.
PitchBook’s AI-driven “VC exit predictor” evaluates how likely a company is to go public or be acquired. The data provider has claimed that the two-month-old tool had a 75 per cent accuracy rate.
Meanwhile, venture capital firms Headline and Moonfire Ventures have used generative AI to assess and compare investment targets based on measures such as web traffic and new users, so as to isolate those with the biggest growth potential….
PUBLISHED THU, MAY 25 2023
JPMorgan Chase is developing a ChatGPT-like software service that leans on a disruptive form of artificial intelligence to select investments for customers, CNBC has learned.
The company applied to trademark a product called IndexGPT earlier this month, according to a filing from the New York-based bank.
“It’s an A.I. program to select financial securities,” said trademark lawyer Josh Gerben. “This sounds to me like they’re trying to put my financial advisor out of business.”
Jamie Dimon, chief executive officer of JPMorgan Chase, is planning his first visit to mainland China in four years as the American bank prepares to host three conferences in Shanghai at the end of May.
Giulia Marchi | Bloomberg | Getty Images
JPMorgan Chase is developing a ChatGPT-like software service that leans on a disruptive form of artificial intelligence to select investments for customers, CNBC has learned.
The company applied to trademark a product called IndexGPT this month, according to a filing from the New York-based bank.
JUN 1, 2023
At Substack, we believe great writing is valuable. We’re focused on building simple tools that help you grow your audience and earn an income directly from subscribers, on your own terms.
Learn about the latest tools we’ve built for and with writers to help you do your best work. Please chime in, in the comments section, with your feedback.
Communicating with your subscribers may go beyond the written word. Writers and creators on Substack use images, audio files, video, and more to tell stories and share information. So we’ve made it simpler than ever to publish videos directly to your subscribers with native video embeds.
The new video embeds allow you to insert original video anywhere in any post type. This expands our video offerings beyond video posts, which offer a theater experience for viewers, putting a single video front and center, plus using a video icon for recognition on the homepage.
When creating a new post, drag a video file from your computer into the editor. After the video is loaded, select a thumbnail, and then insert the video into the post.
Read more: How do I embed a video in a Substack post?
Cross-post to web
Cross-posting is a simple way for you to share another publication’s post with your audience. Add your commentary on a post published by another Substack writer, share the post with your subscribers via email, and now publish to your publication homepage. Before this update, cross-posts were only sent via email, but now you can display these posts on your homepage next to your archive posts.
Cross-posts are great for resharing posts you were featured in, launch posts from your favorite new writers, or simply writing you enjoy and want to share with your audience.
Image Credits: WordPress.com
WordPress.com is taking on Substack and others with today’s news that its Newsletter product will now support paid subscriptions and premium content. First launched in December, WordPress.com Newsletter allows writers to automatically send out posts via email to connect directly with their audience, while still being able to leverage WordPress.com’s other capabilities. Writers can opt to use the feature solely for newsletters or they could add the option to their blog to cater to readers who want to receive new posts via email instead.
While, for years, there have been plug-ins and third-party services that allow blog owners to send out their posts via email, WordPress.com’s decision to move more directly into this space was a reflection of how people now prefer to read news and information. As the state of websites has worsened — dominated by clutter, ads, overlays, pop-ups, and cookie acceptance banners — many have turned to email as an easier way to stay connected to writers, journalists, essayists, and other publishers they want to follow.
Given its sizable footprint — WordPress powers 43% of the web, including its open-source version — WordPress.com’s shift into the newsletters market is significant.
Startup of the Week
Apple’s new goggles aren’t for normals. Not yet, anyway. So why does Apple want to show them off?
By Peter Kafka May 31, 2023, 6:00am EDT
Every big, new Apple Product Launch follows a template, one the company pioneered and perfected with the iPhone and then the iPad.
First, long-running rumors and speculation about a mystery device — a version of existing products made by competitors but presumably much better because Apple is making it — percolate among the Apple-obsessed tech set. Then a somewhat clearer picture emerges, courtesy of reporting by mainstream media outlets. The hype crests as Apple unveils The Product at A Big Deal launch event, and then customers flock to buy The Product by the millions.
And that’s kind of what’s happening with the new “mixed-reality” headset the tech world expects Apple to unveil at its developer conference on June 5, in what would arguably be its most ambitious launch since the iPad in 2010. There has been reporting for years about Apple’s efforts to make the devices, and now outlets like the New York Times and Bloomberg have given us a pretty good idea of what to expect.
Meta Quest Blog, Jun 1, 2023
The countdown is on for today’s Meta Quest Gaming Showcase, but there’s some juicy news we just couldn’t wait any longer to share.
Today, Mark Zuckerberg announced our next-generation virtual and mixed reality headset, which launches later this year. It’s called Meta Quest 3. It features higher resolution, stronger performance, breakthrough Meta Reality technology, and a slimmer, more comfortable form factor. Quest 3 will ship in all countries where Meta Quest is currently supported this fall. The 128GB headset starts at $499.99 USD, and we’ll offer an additional storage option for those who want some extra space. Mark your calendars because we’ll have lots more to share at Meta Connect, which returns this year on September 27.
Quest 3 is the supercharged all-in-one headset you’ve been waiting for—no wires required. Sign up to be the first to learn about Meta Quest 3.
Our Most Powerful Headset Yet
Quest 3 combines our highest resolution display yet and pancake optics to make sure content looks better than ever. To power those extra pixels, this will be the first headset to feature a next-generation Snapdragon chipset developed in collaboration with Qualcomm Technologies. That next-gen Snapdragon chipset delivers more than twice the graphical performance as the previous generation Snapdragon GPU in Quest 2—meaning you’ll get smoother performance and incredibly crisp details in immersive games. Keep an eye out for more details on the chipset that’s powering Quest 3 later this year.
Immersive VR + Breakthrough Meta Reality in a Single Device
On Quest 3, our best-in-class Meta Reality technology lets you seamlessly blend your physical world with the virtual one. These new experiences go beyond today’s mixed reality by intelligently understanding and responding to objects in your physical space and allowing you to navigate that space in natural, intuitive ways that were nearly impossible before. High-fidelity color Passthrough, innovative machine learning, and spatial understanding let you interact with virtual content and the physical world simultaneously, creating limitless possibilities to explore. Now you can play a virtual board game on your kitchen table with Demeo, decorate your living room with virtual art courtesy of Painting VR, or dive into a fully immersive world to do things that are simply not possible otherwise. The choice is yours.