Four human beings are currently hurtling toward the moon. Gas just hit $4 a gallon. It's the one-year anniversary of Liberation Day tariffs. And the AI industry had maybe the most chaotic week in its history.

NASA's Artemis II mission launched last night from Kennedy Space Center — the first crewed lunar mission in 53 years. Victor Glover is the first Black astronaut, Christina Koch the first woman, and Jeremy Hansen the first non-American to travel beyond low Earth orbit. The crew is orbiting Earth right now and will fire their engines tonight to begin the four-day coast to the moon. If all goes well, they'll set a new record for the farthest distance any human has ever traveled from Earth: 252,000 miles.

Oh, and the only toilet on the capsule malfunctioned hours after launch. They fixed it. But imagine being 50,000 miles from Earth with a broken toilet. Puts your Tuesday in perspective.

Meanwhile on Earth: OpenAI closed a $122 billion round at an $852 billion valuation — the biggest private tech financing ever. Anthropic — the company that literally sells itself as the responsible AI lab — leaked their most powerful unreleased model AND 500,000 lines of Claude Code source code. In the same week. Chef's kiss.

Meta cut 700 jobs from Reality Labs to fund AI spending — the metaverse is officially dead. B2B SaaS stocks plunged 25% year-to-date as agentic AI eats the per-seat licensing model alive. Tesla missed Q1 deliveries again. Apple turned 50. And Trump told the nation the Iran war is "nearly over" — again — while gas prices climb past $4/gallon.

And because it was April Fools' yesterday: Dogecoin announced a "corporate restructuring" into DogeCoin Financial Solutions LLC. Traeger dropped AI-powered grilling goggles with thermal imaging and night vision (I actually want these?). Dyson launched a pet styling line including the "Airwrap Fur" for precision curling of — and I quote — "delicate chest fluff." Yahoo made the Scrōll Stoppr, a finger device that physically stops you from scrolling. The most useful product announced all year.

Let's get into the real stuff.

Read between the lines on this one.

That press release wasn't a funding announcement. It was a practice S-1. Revenue metrics. User growth. Direct comparisons to Alphabet and Meta. OpenAI is telling Wall Street exactly what it wants to hear before the actual filing drops.

What they're bragging about: 900 million weekly active users. Enterprise now 40% of total revenue, up from 30% last year. An ad pilot already at $100M+ annualized in six weeks. Sam Altman once called ads a "last resort." That aged well.

What they're not bragging about: they're projected to lose $14 billion this year. They've never disclosed enterprise retention rates. Amazon's $50B came with a requirement for OpenAI to spend $100B on AWS over eight years. Nvidia's investment was mostly compute credits, not cash. These aren't clean equity investments. They're commercial deals wearing a funding round costume.

PitchBook actually ranked OpenAI weakest on business quality among the three big AI IPO candidates — OpenAI, Anthropic, and Databricks — despite commanding the highest valuation.

Meanwhile, GPT-5.4 launched last month in three flavors — Standard, Thinking, and Pro — with a 1M+ token context window and native computer use. It's a good model. But the pricing keeps creeping up. And those ads are coming.

What this means for you: ChatGPT is about to get more expensive and more ad-heavy. If your business runs on the OpenAI API, start diversifying now. Not tomorrow. Now.

Two security lapses. Five days apart. From the AI safety company.

Leak 1: Someone at Anthropic left nearly 3,000 files in a public data store. Including a draft blog post about Claude Mythos — their most powerful unreleased model. A new tier above Opus. Anthropic confirmed it's real and being tested with early customers.

The scary part: they're privately telling government officials this model could make large-scale cyberattacks way more likely in 2026.

Leak 2: Days later, someone pushed a source map to npm containing Claude Code's full source. Not compiled code. The actual TypeScript. 500,000 lines. 1,900 files. Developers had it torn apart before lunch.

What they found is wild: a three-layer memory system, permissions architecture, and something called "KAIROS" — an autonomous mode where the agent keeps working while you're idle. It does memory consolidation in the background. Merges observations. Removes contradictions. Converts vague notes into facts. Claude Code isn't a chatbot in a terminal. It's an operating system for software work.

The strategic takeaway: OpenAI open-sources parts of its coding stack on purpose. Anthropic just revealed comparable architecture by accident. That tells you where each company thinks the real moat is.

And the fallout is still playing out — malicious actors are now typosquatting npm packages from the leak, embedding remote access trojans targeting developers who install them. If you're a developer, double-check what you're installing.

What this means for you: If you're building AI automations for clients (which you should be), study the patterns from this leak. How they handle memory. How they manage context. How they keep agents from going off the rails. It's a free masterclass in production AI architecture.

This one flew under the radar but it shouldn't have.

On March 24, Shopify flipped a switch and 5.6 million stores became shoppable inside ChatGPT, Microsoft Copilot, Google AI Mode, and the Gemini app. No setup form. No new app. Your products are already there. It was opt-out, not opt-in — most Shopify merchants are already in the system.

Think about what just happened. AI assistants are now literal storefronts. Someone asks ChatGPT "find me a waterproof jacket from a sustainable brand" and your Shopify products can be in that answer. They can check out without leaving the conversation. Shopify's catalog uses specialized LLMs to categorize and structure product data so AI agents can parse it.

The numbers behind this: ChatGPT has 880+ million monthly active users. That's a bigger audience than Google Shopping in most merchant categories. AI-driven traffic to Shopify stores grew 7x between January 2025 and the launch. And OpenAI scrapped its previous 4% Instant Checkout fee in favor of a redirect model — so there's no extra cost beyond standard Shopify processing.

They also co-developed a Universal Commerce Protocol (UCP) with Google — an open standard for bringing commerce to agents at scale. And a new Agentic plan lets brands on any platform (Magento, BigCommerce, whatever) add products to the Shopify Catalog without migrating their store.

What this means for you: If you sell anything online, audit your product catalog immediately. Vague titles like "Blue Shirt M/L" won't compete against listings with specific material descriptions and use-case language. Think of it like writing Google Shopping copy, except the buyer is asking a conversational question and the AI decides which products answer it. If you consult for DTC brands or recommend products in a newsletter, this is the play to understand right now.

Meta just killed the metaverse with a pink slip. 700 jobs cut from Reality Labs on the same day they announced $921 million in stock options for four executives. Zuckerberg renamed the company, spent $36+ billion on the metaverse, and is now gutting it to fund AI. The lesson for builders: when a CEO bets the company identity on a narrative and then abandons it three years later, pay attention to what they do next — not what they say.

The SaaS pricing model is dying in real time. B2B software stocks dropped 25% year-to-date. A CIO survey found 40% of IT budgets are being pulled from legacy SaaS subscriptions and redirected to agentic platforms and LLM token usage. Gartner reports 40% of enterprise SaaS contracts now include outcome-based pricing — up from 15% two years ago. If you're building the next thing? Don't charge per seat. Charge per outcome. The market is screaming this.

Tesla can't sell cheap cars either. 358,023 Q1 deliveries, missing Wall Street's 365,645 target. That's two consecutive years of declining deliveries from the 2023 peak. Cheaper Model Y ($39,990) and Model 3 ($36,990) aren't moving the needle. But the Cybercab with no pedals and no steering wheel enters production at Giga Texas this month. Musk has burned every timeline credibility point he ever had. But if even one of these ships, the narrative flips overnight.

Apple at 50 needs to figure out AI or admit it can't. Revenue is great. iPhone 17 is selling. But they're falling behind on AI and everyone knows it. Siri's big AI upgrade? Delayed again. They had to partner with Google Gemini just to have a credible story. At some point "we do AI on-device" stops being a strategy and starts being cope.

Here's the opportunity hiding in this week's news.

Anthropic — one of the most well-funded, technically sophisticated AI companies on Earth — leaked their unreleased model AND their flagship product's source code in the same week. By accident. If it can happen to them, it is absolutely happening at small businesses everywhere.

The play: offer AI safety and compliance audits.

What you'd actually do: check if employees are pasting client data into personal ChatGPT accounts (they are). Vet the AI tools the company uses for SOC 2 and GDPR compliance (they aren't compliant). Write a custom AI governance policy. Set up private API connections with zero data retention.

A Dark Reading poll found 48% of cybersecurity pros now rank "shadow AI" as the #1 attack vector this year. Above deepfakes. Above everything else.

This is a $3,000-$5,000 per engagement service. Maybe more in healthcare, legal, and finance. You don't need to be a cybersecurity expert. You need to understand AI tools better than the average IT department — and right now, that bar is underground.

Build a landing page this weekend. Post on LinkedIn about the Anthropic leaks. Position yourself as the person who keeps companies from becoming the next headline. The awareness just spiked. Go get it.

That's the week. Four humans on their way to the moon. $122 billion for a company that's never turned a profit. The "safety" AI lab leaking everything. And a commerce platform that just made every AI chatbot into a store.

If you're building — keep building. The people who move right now are going to look like geniuses a year from today.

— Michael

P.S. Speaking of building — I'm doing 60-minute Work-Life Balance Audits where I dig into your operations, find where your time is going, and hand you a clear roadmap for automating it. If you're successful but still working nights and weekends on stuff that should run itself, let's talk.