🚦 Market Signals
Anthropic launches Claude Opus 4.6 with 1m context
The all new Opus 4.6 "plans more carefully, sustains agentic tasks for longer, can operate more reliably in larger codebases, and has better code review and debugging skills to catch its own mistakes." according to Anthropic. They also upgraded Claude in Excel, and released Claude in PowerPoint in preview.
Super Bowl AI Ads
An average of $8 million per 30-second spot. Yes, you read that right. Since it's average, then obviously there were slots for $10m+. Crazy!
The Super Bowl is usually where mainstream narratives are tested, and this year AI dominated the narrative. Anthropic took a swing at OpenAI and sparked debate around ads in AI products. Then Google leaned into Gemini in a way that framed the model race as a consumer story, not just a developer one.
Why it matters: distribution is becoming as important as model quality. The next phase of competition is not only "who has the best model?" but "who gets embedded in daily behavior first?"
Stock market sell-off
Software stocks sold off hard this week after Anthropic's Cowork launch reinforced a fear that AI is moving from "assistive" to "substitutive" faster than expected. If agents can execute longer workflows with less supervision, revenue tied to seat-based productivity tools gets repriced.
The Cowork compliance disclaimer also highlighted a second-order issue: enterprise adoption will not be linear. Regulated buyers still need audit visibility, but the market reaction says investors are already pricing in future displacement before those blockers are solved.
SpaceX's acquisition of xAI
Elon Musk's decision to merge SpaceX and xAI is less about branding and more about stack control: compute, distribution, and narrative under one roof. If this structure holds, the company can train, deploy, and market faster without the usual coordination tax across separate entities.
For builders, it is another reminder that frontier AI competition is not just model-vs-model. It is ecosystem-vs-ecosystem, where ownership of infrastructure and user channels can matter as much as benchmark scores.
Vibe-coding startup Anything hits $100M valuation after $2M ARR in two weeks
We've had a plethora of vibe-coding apps out there by now. Some are great, some not so great (prompt quality matters!), but the real question is: what do you do next as a non-technical creator once you vibe-coded your prototype? Enter vibe-deployment. Technical people struggle with cloud docs, why would you expect non-technical creators to do better? I think there is a real space for hassle-free, production-grade deployment - the market is going to be massive in the future.
Worth watching, not because the tool is revolutionary, but because the consequences might be.
What Is ChatGPT Doing … and Why Does It Work?
I don't know about you, but I don't like to just use tools and don't understand how they work. Stephen Wolfram’s deep dive is still one of the clearest mental models for how LLMs map statistical structure to coherent language. It’s useful reading for builders who need to explain “why this works” to stakeholders or design prompts that align with how the model actually behaves.
đź“° In the News
The AI race is normalizing 70+ hour workweeks
This was one of the most revealing reads of the week: the AI boom is pulling parts of tech toward "996"-style expectations dressed up as ambition. My take is simple: intensity can win short bursts, but sustained product quality comes from clear priorities, leverage, and systems that do not burn out the people building them.
If you are shipping in AI right now, this is the right question to ask: are we compounding capability, or just compounding exhaustion?
đź“š Books
How To Think About AI: A Guide For The Perplexed
A compact, non-technical primer that reframes the conversation around outcomes, institutions, and the second-order effects of AI. I have really enjoyed this book. Key insights for me are:
Process-Thinking vs Outcome-Thinking
In almost every AI discussion, I try to separate two camps: process- oriented thinkers and outcome-oriented thinkers. Process-oriented people focus on imperfections in today's technical approach. Outcome-oriented people focus on results and care less about implementation details. Framing the conversation this way makes it easier to understand where people are coming from.
Automation, Innovation and Elimination
AI affects work in three ways: Automation, Innovation, and Elimination. People protective of the current state of their profession usually focus on automation and ignore the other two. That blind spot leads to bad strategy.
Elimination, not automation, is the real threat
Lawyers think their jobs are safe because legal reasoning is hard to automate. They're right, but they're asking the wrong question. The question isn't "can AI do a lawyer's job?" It's "does society still need lawyers?" Most people don't want adversarial legal proceedings. They want fair, predictable outcomes. What if minor crimes don't go to court at all? "You did X under circumstances Y, therefore consequence Z" - a deterministic system, not a trial. No judge's discretion, no lawyer's persuasion, just algorithmic justice. Sounds dystopian? Maybe. But also: faster, cheaper, more consistent. The demand for lawyers doesn't decline, it collapses.
Developers aren't immune.
We think our jobs are safe because "AI can't architect complex systems." True. But what if AI makes systems so simple they don't need architecting? What if generated code is so reliable that your debugging skills become irrelevant? Elimination, not automation. That's the risk nobody's pricing in.
GoodreadsThat's it - see you next week!