el producto #341 🚀
Apple GPT, Threads: lessons and challenges, AI web builders, Pre-MVPs, Building moats, Netflix & Tesla Q2 results, Tips on scaling, Twitter Articles & more
Hey team,
Happy weekend and welcome to a new edition of el producto!
🎰 The week in figures
$8B: SpaceX expects to double revenue in 2023 to $8B, with $3B of operating profits, as Starlink has started to comprise a significant portion of the company's revenue
$190M: Musinsa, a Seoul-based fashion marketplace with more than 8,000 local and foreign brands, raised a $190M Series C at a ~$2.76B valuation
$100M: Netcraft, a UK-based cybersecurity startup, raised a $100M round; after starting in 1995, this is the company's first raise to date
1.5M US users have signed up for Netflix's ad-supported tier so far; the company also recently eliminated its most affordable ad-free tier at $10/month
-90%: Cameo will cut 80+ employees, leaving the company with under 50 FTEs, down ~90% from its peak in 2022
Q2 results
Netflix: revenue: ~$8.2B (+2.7% y/y), net income: ~$1.5B, FCF: ~$1.3B, global paid memberships: 238.4M (+8% y/y); shares were down ~8% due to missing forecasts
Tesla: total revenue: $24.9B (+47% y/y), FCF: $1B (+62% y/y), vehicles built: ~480K (+86% y/y), deliveries: ~466K (+83% y/y); shares were down ~10% despite the earnings beat
📰 What’s going on
Apple has created its own "Apple GPT" using an internal LLM framework called Ajax and is planning its own AI announcement for some time next year; Apple still lags behind other tech giants in terms of AI employee count
Apple has renamed its XR goggles division "Vision Products Group," or "VPG"; unlike other groups at Apple, VPG does not depend on Apple's software and hardware engineering teams
Twitter is reintroducing its long-form content feature, now called "Articles," according to Elon Musk; it's unclear if or when Articles will be available to the public, as it is still in the experimental stage and limited to select users
TikTok is expanding its closed beta test for its music streaming service to Australia, Mexico, and Singapore; TikTok Music syncs with a user's TikTok account to discover, share, and download music from major record companies
Meta rebranded Facebook Watch to "Video," which now includes tabs for Reels, Live, music, and more
Meta has decided to halt the development of its high-end mixed-reality headset, the Meta Quest Pro, and discontinued work on the Quest Pro 2; this move comes amidst reports of the company's shift in focus toward more affordable VR/AR devices and the upcoming launch of the Meta Quest 3
Meta released Llama 2, an open-source LLM that's free for research and commercial use, with Microsoft as its preferred partner
Microsoft announced its AI-powered Copilot tool for Microsoft 365 will cost $30/user/month; Copilot, powered by GPT-4, allows users to generate content by typing in text prompts in applications like Teams, Word, Excel, PowerPoint, and Outlook
Google is testing an AI tool that can generate news content; the tool, known internally as "Genesis," was showcased to executives from major media outlets including the Times, The Washington Post, and News Corp; Google has pitched the tool as "responsible" technology that could serve as a personal assistant for journalists, automating certain tasks to free up their time
OpenAI announced ChatGPT for Android will roll out next week; its iOS app launched in May
Wix has debuted an AI-powered website builder that can create entire sites from text prompts, including support for events, bookings, and more
G/O media, which operates Gizmodo, the Onion, Jezebel, and others, will double down on AI-generated stories despite internal and external criticism for its botched efforts so far
Uber entered a partnership with Vroom Delivery; the collaboration allows convenience retailers on Vroom's platform to fulfill orders placed via Uber Eats
Amazon will require some remote employees to relocate as part of a 3-day/week return-to-office mandate
📚 Good reads
When to dig a moat, by Packy McCormick. Moats won’t get you product-market fit, but as soon as you have PMF and have proven you’re doing something that (1) works and (2) is valuable, you should be able to protect it from the erosive forces of competition. “Depth of Moat Needed = How Obviously Good Your Idea Is - How Hard it is to Build”
AI means speed and automation. NFX’s James Currier pushes every founder to accelerate, giving precious pointers to mental models, and approaches that can help you unlock your go-to-market, and customer connection capability. Almost tangentially Currier manages to give an outlook on what the landscape looks like for tech in a world where AI becomes widespread, shortcutting gaps and automating tasks: "In the future, we’re going to see smaller teams. Fewer VCs. More software and automation controlling the tasks we do every day"
Tips on Scaling up. Few weeks ago, A16z released a few short posts with suggestions on scaling up organizations - in a growth phase - from the technical (eng), go-to-market, and operational perspectives. Despite not being just-follow-this handbooks, the posts contain quite a few interesting bits to get inspiration
Making every social post and MVP. Greg Isenberg argues that we're witnessing a paradigm shift in how we validate ideas, with social posts becoming the new MVPs (pre-MVPs?) in this 'Lean Startup 2.0' era
Finding your winning idea by selling Products that don’t exist yet. Following on last article’s topic, more on pre-MVP validation. Material Security co-founder Ryan Noon and co-founder Abhishek Agrawal share their path to Product/Market-Fit, and how they ultimately landed on their winning idea:
Start with the team first, not the problem: “Product-market fit was far from my mind at the beginning. I wanted to get the team right first. If you have the right people working together the right way, and you're disciplined in discovery, you can go far,” says Noon
Pick a customer instead of an idea: “One of the best pieces of advice I ever got was that when you're starting a company, you often think you're picking an idea, but you're really picking a customer. It's actually pretty easy to change your idea, but it’s way harder to change the customer you're serving,” says Agrawal
Try the “hot knife through butter test”: With four different software ideas in hand, the founding team started selling. “All of the flight plans boiled down to the same thing. It was essentially find as many people as possible in your network that have anything to do with that idea, and try to sell it to them,” says Noon. “Don't ask for feedback. Don't do user research—just try to sell your idea”
Threads: the problem with the “everything for everyone” approach. An interesting view on Threads potential challenges, and some ideas on future pivots. “Successful social networks have focused on solving specific user motivations through a specific format, unlike Threads’ generic approach.” While the longevity of Threads is not guaranteed, there are a bunch of great Product lessons to be taken from its launch. A great complementary read with 5 highlights on cold starts, distribution, timing and leveraging the new entrant status
How to build a knowledge base to communicate research findings, by Zahra Rahimi. A handy approach to enable easy collaboration across multiple teams
That’s all for today! Let me know what you think by replying back to this email or commenting on Substack
Angel