Every week, get the latest curated high signals on news, insights, and ideas around technology, science, and business to help you become a better builder and thinker. Please fill out the survey so we can deliver the best content to your inbox. Thanks for reading!Thanks for reading!
This one came to me from a founder friend who was trying to connect coldly with investors and customers. Think of this as a tool that searches the web and consolidates everything about you into one spot i.e. X, LinkedIn, podcasts you’ve been on, blogs you’ve written, investments you’ve made. It’s quite surreal to see your entire internet bio in one spot but also extremely telling on how much unstructured information there is about us online. I encourage you to do a search on yourself and see what you can find. The tool is really helpful if you want to do a quick background check on someone without having into log into a social platform which they may or may not be on. They also have a paid tool that helps investors find under-the-radar founders, which is most likely their cash cow. You can read their About Page for me info.
Note: I’m not financially affiliated with any of these companies.
Here’s this week’s scoop:
OpenAI Releases GPT-5, but still not AGI (yet)
Microsoft Bets Bigger on OpenAI
Figma finally goes IPO, and why this matters
Cognition offers Windsurf employees an exit route
Apple invests $600B into US manufacturing and production
Why there’s really no one AI to rule them all
Why Chinese universities are encouraging using more AI, not less
🔥 Nuggets for the Road
Anthropic is approaching a $170B valuation and planning on raising a $5B round — it looks like Anthropic is carving out its own LLM path different from other big players [LINK]
Robotics Levels of Autonomy — a really comprehensive breakdown of the five levels of robotic autonomy. Worth a read for anyone trying to understand that path of progress for general-purpose robots [LINK]
Meta is now allowing job candidates to use AI during coding interviews — vibe coding is now becoming normalized in big tech [LINK]
Google now prioritizes building over writing — Madhu Guru (Product Leader) at Google mentioned that the company is now embracing a build-first culture, especially for PMs. This means vibe-coding has entered the chat (again!) [LINK]
Data center construction is now taking over corporate office construction — why build a partially occupied commercial building when you can use the same real estate to build 24/7 data center operating at full capacity? Go figure. [LINK]
The dramatic decline in Chinese startups — a telling sign about the state of affairs in the country, and how government oversight/influence can impact an ecosystem [FT, LINK]
📡 The Signal
OpenAI Releases GPT-5, but still not AGI (yet)
GPT-5 is OpenAI’s latest and most advanced model yet. Obviously, it’s much smarter than its previous models, but how smart? Well, early results show that it’s certainly much more intelligent, with Altman even saying that it’s a “PhD-level expert in your pocket”. But even he admits it’s not quite at AGI level yet. The big problem is that these models still can’t continuously learn from new experiences like human beings. So it’s safe to say that no one will be losing their jobs anytime soon. In fact, even with the most advanced LLM models out there right now, there is something to be said about the value of human insight and ideation — you should read Balaji’s latest blog on ‘Why there’s really no one AI to rule them all’— I wrote a bit about further down below. [OPENAI, GUARDIAN]
I think the way that most of us define AGI, we’re still missing something quite important, many things quite important. But one big one is … this is not a model that continuously learns as it’s deployed from things it finds, which is something that, to me, feels like it should be part of AGI,” he said, adding that, nonetheless, it was a “huge improvement” on its predecessors. — Sam Altman
Microsoft Bets Bigger on OpenAI
Microsoft is creating a new, long-term contract that would lock in its right to use every new OpenAI model, up to and including anything that could be called AGI, and then be able to deploy it across Azure, Windows, Office, and the rest of its ecosystem. The agreement is meant to survive any future shake-ups in the partnerships, giving Microsoft automatic access to the tech (and the underlying code weights) the moment OpenAI ships it. In exchange, Microsoft would keep providing the cloud muscle and cash OpenAI needs as it pivots fully into a for-profit company. The interesting thing here is that Microsoft now has guaranteed first dibs on every GPT upgrade, which gives it a head start in deploying its capabilities at scale, something that Google (and others) will have to replicate or counter. [BLOOMBERG]
Figma finally goes IPO, and why this matters
It’s fair to say that it’s been a few years without any big tech IPOs. Usually, when a company goes public, the plan is to have a big celebration, with lots of hype. But lately, most IPOs have been kind of a letdown. High interest rates and inflation made people nervous, and companies just kept waiting for the right moment. Even when some did go public, their shares didn’t really take off. Now, Figma comes in and totally flips the script. The company’s shares shot up more than 3x their IPO prices on their first day. The big question here is, will Figma become one of those generational tech stocks, or will it become a meme stock like OpenDoor? For a company that was previously on a path to be acquired by Adobe ($20B), it’s done extremely well — now worth $68B (at peak) on the NYSE. It’s got over 13 million monthly users, and most aren’t even designers. The real challenge is how Figma will navigate the AI space as more people start to vibe-code/design with design tools like Bolt, Lovable, v0, etc. The bottom line is that the IPO doors are open again, and it’s a sign that there is a place for venture capital backing companies in the right sector at the right time. Acquisitions are good, and probably fortuitous that the Adobe deal fell through, because if companies can forge their own path, that’s even better! [LINK}
Cognition offers Windsurf employees an exit route
A few weeks ago, the headline was that Windsurf had been acquired by Congition Labs after Windsurf’s founding team got acqui-hired by Google, leaving the rest of the employees stranded without an owner. Now Cognition is doing its own little purge of Windsurf’s employees, but on good terms. Usually, when a company gets acquired, the new bosses try to keep everyone happy with some promises or small bonuses, hoping people will stick around. Most of the time, though, the employees just get overworked or just don’t fit in the new culture. So now, Cognition is making it super clear: either you accept a buyout worth nine months’ salary plus accelerated vesting of your equity, or you stay and work with hardcore intensity. The silver lining here is that the CEO isn’t pretending that this is a chill place to work, either. He straight-up says they don’t believe in work-life balance and that the company’s culture is all about working hard, late nights, and on weekends. This honesty is pretty rare in tech, where companies usually try to spin things up to keep everyone on board, even if it’s not a good fit. In any case, it’s kind of refreshing to see a startup just lay it all out, even if the deal itself is a tough choice. [LINK]
Apple invests $600B into US manufacturing and production
Apple recently announced they are throwing an extra $100 billion on top of the previous $500 billion into US plans, suppliers, and workers over four years. It’s called the new American Manufacturing Program (AMP), and the whole point is to take charge of every step of the manufacturing process of Apple products, starting with the iPhone. So in this case, there’ll be plants taking care of the iPhone glass all the way to the silicon wafers (approx 19 billion Apple chips to be manufactured for 2025) that power the product. The whole point of this is to build out the US Silicon supply chain, which means there’s a lot of collaboration between Apple, TSMC, Samsung, and GlobalFoundries to prove the whole chip pipeline. [APPLE]
To better explain how this works, you should read the excerpt below from Apple’s website on how the supply chain will all come together. Apple is not a foundry, so it needs to build strong partnerships with some heavy hitters. The interesting thing is that other companies like NVIDIA and Tesla will also be leveraging this supply chain as well.
Wafers are the building block of any silicon chip, and Apple is partnering with GlobalWafers America in Sherman, Texas, to produce advanced wafers for use in U.S.-based semiconductor fabs for the first time. American chip fabs like TSMC in Phoenix, Arizona, and Texas Instruments in Sherman, Texas, will use GWA’s 300mm wafers to produce chips for iPhone and iPad devices sold in the U.S. and around the world. GWA uses silicon from U.S. sources, including from Corning’s Hemlock Semiconductor, to produce the world’s most advanced silicon wafers.
Apple is also partnering directly with Applied Materials to boost the production of semiconductor manufacturing equipment in the U.S. The Applied site in Austin, Texas, is a pivotal hub for manufacturing cutting-edge chip equipment.
Fabs take bare wafers and turn them into chips. Apple and Texas Instruments are expanding their partnership to increase future product collaboration and critical U.S. capacity for this work. Apple is making a new commitment with TI, which will support additional tool installations at its facility in Lehi, Utah, and a new facility in Sherman, Texas. These facilities are home to TI’s most advanced process technologies and use American-made chip manufacturing equipment from Applied Materials’ factory in Austin, as well as advanced silicon wafers from GlobalWafers America. These facilities will manufacture critical foundational semiconductors used for Apple products, including iPhone devices shipped in the U.S. and around the world.
Apple is also working with Samsung at its fab in Austin, Texas, to launch an innovative new technology for making chips, which has never been used before anywhere in the world. By bringing this technology to the U.S. first, this facility will supply chips that optimize power and performance of Apple products, including iPhone devices shipped all over the world.
GlobalFoundries and Apple have also entered an agreement to bring more semiconductor manufacturing to the United States, focused on manufacturing cutting-edge wireless technologies and advanced power management — critical technologies that enable longer battery life and enhanced connectivity in Apple devices. The partnership will bring new capabilities, jobs, and technology to the GlobalFoundries semiconductor facility in Malta, New York.
Packaging is the final critical step in manufacturing silicon chips. Apple is investing in Amkor’s new advanced chip packaging and test facility in Arizona, and will be its first and largest customer. This will accelerate the development of packaging capabilities in America, meaningfully strengthening the semiconductor supply chain in the U.S. This facility will package and test Apple silicon manufactured at the nearby TSMC fab, and create chips used in iPhone devices shipped around the world.
Why there’s really no one AI to rule them all
Most people worry about AI being this singular, all-powerful/all-knowing entity that takes over everything. But this couldn’t be further from the truth. Instead, (and I agree) that there’s going to be lots of different AIs, each with their own strengths, made by different companies + groups. The other takeaway here is that AI isn’t as independent or “smart” as people think. It’s actually more like a tool that just amplifies whatever you put into it. You still have to prompt it, check its work, and use your own judgment. Ultimately, I really do think that AI doesn’t really take your job, but rather, it lets you try lots of different jobs, and you still need real skills to do them well. One other insight here is that the real costs and effort are moving to how you ask AI questions (prompting) and how you check its answers (verifying). The below visual of Laffer’s curve shows that AI has the potential to be extremely useful when optimized, but very useless when everything is 100% achieved by AI. [Balaji’s SUBSTACK]
Why Chinese universities are encouraging using more AI, not less
There’s still contention on what to do about AI in colleges/universities. The worry around this is that it’ll make students cheat, be lazy, or become academically dishonest. So the usual fix is to do an AI ban. But let’s be honest: students just find workarounds, so it doesn’t solve much. Now, an increasing number of Chinese universities are encouraging students to embrace AI and use it openly and responsibly. They’re actually teaching students how to get the most out of the AI tools for brainstorming, drafting, and organizing ideas, but always reminding them that AI can’t replace their own judgment. There’s also a push to take AI classes and not just computer science majors, as well as learning AI as young as possible. Apparently, some campuses are setting up AI models like DeepSeek on university computers to provide free access. This makes sense. Considering how many videos, blogs, and tutorials are already out there to help professionals with AI, why shouldn’t there be something for students, who are essentially the next generation of professionals? You want to be getting AI into the hands of people as quickly and as early as possible. Doing this sets up future workforces to be much more effective. Students still need to understand fundamentals/first principles, but using AI to help will only make things better. [MIT]
If you liked this, here are a few more popular posts:
To build the best newsletter for you all, I’d love to get some feedback. If you have some time, please fill out the following survey.
🏆 Alpha Bytes Premium
If you like our content, you can explore our Premium option. You can read more about it here. This includes (for now):
✅ Full access to the weekly newsletter ✅ Full access to podcasts + curated transcripts in Q&A format—this has become widely popular. ✅ Full access to all deep dives, essays, and tactical field guides ✅ Early access to new features and experimental content
🤩 Deals, Discounts & Affiliates
You can find carefully selected offers and deals here.
