2nd Order Thinkers.

Microsoft does not need the best AI model. However, it has the ultimate sales machine to still win in the AI era.

The consequential question is who gets paid when a bank, hospital, or government department decides it needs AI without creating a security problem.

That answer makes the cloud race easier to see, whether AI proves hugely productive or not.

More at https://www.2ndorderthinkers.com/

Send this to the person still choosing an AI strategy from benchmark screenshots.

Google is turning its best cash machine into a factory before it has proved the factory can pay for itself.

Alphabet is on course to spend up to $205 billion this year on chips, servers and data centres.

Google Cloud is growing fast.

But none of that tells investors how much demand is paid, durable, or capable of covering power, depreciation and the next GPU replacement cycle.

More at https://www.2ndorderthinkers.com/

When Hugging Face got attacked by OpenAI's own model, the commercial AI it turned to for help refused to cooperate — an open-weight Chinese model fixed it instead.

Days later, Kimi K3 went viral for matching frontier performance while giving away exactly what OpenAI and Anthropic are built to sell.

I'll explain why the regulation Altman and Amodei are asking for isn't serving the US's best interests.

More at https://www.2ndorderthinkers.com/

Send this to the friend who thinks AI safety rules are only about safety.

More at https://www.2ndorderthinkers.com/

Send this to whoever on your team keeps pasting the everything into ChatGPT to "save time."

Llama was mentioned 19 times in Meta's 2023 earnings call. In Q1 2026, it wasn't mentioned once. That's not an accident — it's what happens when a public company runs out of patience with a bet that shows no return.

Zuckerberg used to charm developers by giving away Meta's best AI research for free. Now he's building a cloud business to sell the compute he can't justify keeping idle.

Same infrastructure, opposite instinct: from a founder who bets on ideas nobody else believes in, to a CEO who has to answer to shareholders who want a number.

More at https://www.2ndorderthinkers.com/

Micron's customers just signed $100 billion in take-or-pay contracts: commitments to pay whether or not they take delivery. That's not enthusiasm. That's what real fear of a shortage looks like in writing.

More at https://www.2ndorderthinkers.com/

Musk’s Weekend Decision That Could Reshape AI Coding

Musk had 30 days to decide whether to purchase Cursor. He used only a weekend.

Rewind.

If the only tech news that mattered last week was SpaceX’s largest public offering in history, then the one piece of tech news to know this week is SpaceX’s acquisition of Cursor.

This is pretty much a done deal, something they were already planning since April. Yes, so what’s left to talk about?

A lot.

To start with, this is the one thing that will impact the landscape of the enterprise coding platforms over the next 3 years.

So, I’ll answer the following questions in this one:

* Why even buy Cursor to start with?

* Why was Musk in such a rush to make this decision?

* What Musk actually bought?

* How would this impact Anthropic? (yes, Anthropic)

* Two possible war-room scenarios for all parties over the next three years

---

Subscribe to https://www.2ndorderthinkers.com/ for more.

From village matchmakers to dating apps, the job of the dating 'tech' never changed: gather information, open a channel, verify intent. What's never been for sale is standing out and staying together — until AI companions make both unnecessary, by removing the other person.

More at https://www.2ndorderthinkers.com/

If you've ever explained to someone why you flinched when they asked how you met your partner, send them this.

Dario Amodei says AI could wipe out half of entry-level white-collar jobs.

In this episode, I test that claim against company surveys, layoff data, unemployment figures, and a first-principles framework for how jobs actually work.

The argument: most AI adoption still automates tasks, not whole roles, and much of the layoff narrative is better understood as post-pandemic restructuring wrapped in AI language.

The key question for any job is simple: what pain does it solve, and can AI solve that pain better, cheaper, and end-to-end?

Read more at 2ndorderthinkers.com

Every team that won the Gemini 3 Hackathon built a workaround for something AI can't do. That's not a coincidence — it's the pitch, read backward.

I looked at all three winning products: a supply chain crisis tool, a disaster triage system, and an assistive navigation app. Each one won by designing around an AI failure — and each workaround comes with a cost the sales deck doesn't mention.

More of this in the newsletter → [2nd Order Thinkers URL]

What you'll leave with: A framework for reading any AI pitch in reverse. Stop at the human approval gate, the explainability layer, the memory patch — and ask what it's compensating for. That question is worth more than any demo.

Mentioned in this episode:

Globot (Gemini 3 Hackathon grand prize)

Aegis (2nd prize)

Netra / Memory Palace (3rd prize)

Deloitte / UK Ministry of Defence automation bias report

Carnegie Mellon + Stanford research on human-agent supervision costs