📰On The Record – 26th July 2026

South Korean President Lee Jae Myung spent yesterday turning a San Francisco AI summit into Seoul's trading floor, unveiling partnerships worth a combined $950 billion between Samsung, SK Group and American giants including Nvidia and Broadcom. The deals span chips, data centres and the unglamorous infrastructure that actually makes AI work, not just look good in a keynote. This matters because it isn't Silicon Valley writing the cheques this time — it's a government treating AI leadership like a national security priority. When a country bets nearly a trillion dollars on one technology, it's telling you exactly where the next decade's power is going to sit.

Anthropic's head of economics, Peter McCrory, spent yesterday politely undercutting his own CEO's most famous warning. Dario Amodei has spent months predicting an AI-driven wipeout of white-collar jobs. McCrory's research says it hasn't shown up yet — unemployment remains low, and the occupations most exposed to AI aren't losing jobs any faster than anyone else. The catch: he isn't saying it won't happen, just that legal, technical and organisational friction is slowing the rollout. The apocalypse got stuck in the approvals process. For any founder wondering when to start automating, that's either relief or a countdown, depending on the industry.

If AI can technically already do a large chunk of your team's work but hasn't been forced into most workplaces yet, is waiting the smart move or just delaying the inevitable?

Worth sitting with both sides before you scroll on — getting ahead of disruption is savvy, but rushing change nobody's asked for yet has its own costs.

There's a trend quietly reshaping global business that doesn't get its own headline often enough: infrastructure spending has become foreign policy. Governments aren't just regulating technology companies anymore, they're bankrolling them, treating chip plants and data centres the way earlier eras treated ports and railways. That changes the rules for everyone building nearby. Your competitors aren't just other companies now, they're companies with a state's balance sheet quietly standing behind them. If you're planning five years out and not accounting for which governments are subsidising your rivals, you're planning with half the map missing.

Business plans that ignore geopolitics aren't strategies, they're wishes.

Many of the same voices who predicted an AI-driven jobs bloodbath are now softening their tune, Forbes reports. Adoption has proven slower and more complementary than feared, shifting the conversation from mass layoffs toward gradual augmentation. It's not a victory lap — more tasks will still be automated — but it's a meaningful reset of the timeline everyone was bracing for.

Microsoft, Nvidia, Meta, OpenAI and dozens more signed an open letter urging regulators to protect open-weight AI models rather than restrict them broadly. Their case: openness drives US leadership, transparency and security, and targeted rules beat blanket ones. The letter lands as Washington weighs tougher curbs on Chinese AI developers, exposing the tension between security instincts and an industry that wants its playground left open.

Trump's latest tariffs, replacing ones struck down by the Supreme Court, are expected to raise costs on cars, clothing and household goods, per the Wall Street Journal. They'll boost government revenue and nudge some manufacturing home, but economists doubt they'll meaningfully shrink the trade deficit or bring back factory jobs. The real bill lands with consumers and businesses navigating yet another round of uncertainty.

Chinese AI startup DeepSeek has reportedly paused its second funding round after comments from founder Liang Wenfeng leaked and went viral, Bloomberg reports. The company had been chasing fresh cash at a valuation north of $70 billion after raising $7 billion in June. Talks are expected to continue, but the timing is awkward — a stumble right as US-China AI competition sits under a microscope.

Bloomberg reports IBM's turnaround has been knocked sideways as customers redirect IT budgets from its software and mainframe business toward AI infrastructure. The shift forced IBM to cut its 2026 revenue outlook and triggered its worst-ever single-day stock drop. CEO Arvind Krishna insists the spending is delayed, not lost — investors are increasingly betting the AI boom is eating IBM's lunch faster than it can adapt.

The best assistants don't multitask their attention across a hundred tools. Neither does Catch. It's an AI agent that focuses on one thing — the admin work you'd rather not touch — and does it exceptionally well.

Scheduling, flights, restaurants, follow-ups, vendors, clients. You hand it over; Catch handles the back-and-forth and comes back with it done.

No context-switching. No dropped balls. Just your admin, quietly cleared — so your focus stays on the work only you can do.

Meet the agent built for admin, and it'll be ready to work before your next meeting.

Get started at catchagent.ai — and give your attention back to what matters.

Start now for free

What it means: It's an AI model whose underlying settings — the actual numbers that make it work — are published for anyone to download and build on top of, unlike a "closed" model you can only reach through an API. Think of it as open-sourcing a recipe instead of just selling the finished dish. Today's tech coalition letter to Washington was really a fight to keep that recipe public.

Around a year ago today, the FCC voted 2-1 to approve Paramount's $8 billion merger with Skydance, capping a saga that mixed Hollywood drama with political pressure after Paramount settled a lawsuit brought by President Trump. The deal closed the following month. A year on, with governments still shaping who gets to own the next era of media and technology, it's a reminder that even the biggest corporate deals often hinge less on economics than on who's willing to make peace with power.

Founders make dozens of calls a week on gut instinct alone, which is fine until the stakes get bigger than "which font." This prompt forces you to slow down and pressure-test a decision before you commit to it — useful before a hire, a pivot, or any move you can't easily undo.

Copy and paste this into your AI tool of choice:

Run it before your next big call — it's a lot cheaper than finding out the hard way.