Lumian Gen AI Newsletter Issue #51

Welcome to the 51st edition of the Lumian Weekly Gen AI Newsletter!

After years of hearing about how AI is going to revolutionize every job, it looks like coding is the first to actually get there. Startups like Replit, Anysphere, Augment, Supermaven, Poolside, Cursor, Cognition, Magic, along with tech titans like Microsoft, Google, and Amazon, have been racing to take over different segments of this market—and they’re putting their money where their mouths are. Over $1 billion has poured into AI-powered coding/software engineering tools since the start of 2023. And if you ask investors, we’re just getting started.

Now, I know what you’re thinking: “Coding? That doesn’t sound like the coolest job for AI to transform.” But the thing is, if you’ve ever spent hours slogging through buggy code or refactoring some legacy mess from 2005, you’d beg for an AI to step in. In fact, the geniuses in Silicon Valley are so excited about this, they’re comparing AI in coding to the word processor for writing. Remember typewriters? Neither do I, but I imagine they were as tedious as coding without AI.

AI coding tools are going to be as ubiquitous as spellcheck. Take GitHub’s Copilot for example, a tool that basically acts as your coding sidekick—like Robin to your Batman, if Robin could write better code than you. Copilot has 2 million paying subscribers and accounts for 40% of GitHub’s revenue growth. That’s not just hype, that’s business. Microsoft, which scooped up GitHub in 2018 for $7.5 billion (its annual revenue run rate (ARR) was $2 billion at the start of this month), is probably doing a victory lap right now. Satya Nadella casually mentioned on a recent earnings call that GitHub’s Copilot revenue is growing like crazy, and if Microsoft is talking about you like that, it means something big is going on.

The thing is, even though these AI tools are killing it in terms of productivity—some companies are seeing 20-35% gains—there’s still a lot of skepticism. You see, no one (yet) is crazy enough to let AI-generated code run production systems without human oversight. Sure, AI might write the lines, but it’s still the engineers who have to make sure those lines don’t break the entire product. It’s awesome for repetitive tasks but still a helper, not a replacement. There’s no AI that understands good software architecture yet (Magic?). That’s still a human thing. Which is a bit of a relief, right?

So why coding? Why not some other job function first? Well, it turns out coding is the perfect playground for generative AI. Programming languages are rule-based, and even the most creative coders operate within some pretty structured guidelines. That makes it easier for AI to swoop in and help out. And unlike some of the other generative AI applications we’ve seen (ahem, AI-generated art), this isn’t just about novelty. Investors are excited because AI coding assistants can be embedded into existing workflows. They don’t just feel useful, they are useful—immediately. The value is visible, and the time to value is short, making it an easy sell to enterprises.

But this is just the beginning. What we’re seeing now is a pivot in the role of software engineers themselves. They’re moving away from being “code writers” to becoming system designers. Engineers will move up the value chain to focus more on design and architecture, while AI handles the tedious grunt work. It’s like going from being a carpenter to an architect—except the hammer is AI and it doesn’t complain when you ask it to work overtime.

Ultimately, what we’re looking at isn’t just AI assisting in coding, but coding itself being transformed into something entirely new. The implications are enormous. Imagine a world where building software becomes so easy that it democratizes development, letting anyone with a good idea put together an app. The easier it gets, the more we’ll build, and the more demand there will be for better tools. It’s a virtuous cycle, one that could leave us all swimming in more software than we ever thought possible.

So yes, AI is coming for coding—and it’s probably the best thing that’s happened to the field since Ctrl+C and Ctrl+V.

Happy reading! 📚🤖🎵

In this week’s issue:

• News Flash: SSI raises $1B, Amazon x Anthropic, Gemini’s New Models

• AI Frontier: AI Coding tools you can use today

• Fundraising: The biggest deals in AI

• Nerd Out: Technical and Business Content for Everyone

🚀 AI in Practice

Cutting-Edge AI Coding Tools You Can Use Today

Last weekend Twitter woke up to a (not new) coding AI startup called Cursor. It is a fork of VS Code but with many new features and UI improvements. Even Karpathy is talking about it.

If you’re new to coding, I’d strongly recommend learning to code with Cursor! If you're already proficient, it may be worth learning to code with Cursor all over again!

They’ve raised $60M in Series A funding from Andreessen Horowitz, Thrive Capital, OpenAI, among others.

• Step-by-Step Guide to Setting Up Cursor AI

• CursorCasts - Free screencasts to help you learn how to build your own apps with Cursor