This website uses cookies

Read our Privacy policy and Terms of use for more information.

Sponsored by

Good morning and happy Friday! There's a particular flavour of chaos that only shows up during Big Tech earnings season β€” the kind where balance sheets read like thrillers and CEOs suddenly discover a talent for dramatic pauses on investor calls. Yesterday had that energy in spades. Numbers got bigger, stakes got stranger, and a few closely-guarded secrets finally came out of the vault. Meanwhile, the "real" economy kept doing its own quietly stubborn thing, refusing to follow the script Wall Street had written for it. Put it all together and you get a day that felt less like a single headline and more like six subplots converging on the same finale. Grab yourself a coffee and lets run through how it played out.

VAULT UNLOCKED

πŸš€ Google Finally Shows Its $94 Billion SpaceX Hand

Alphabet let slip a number it had kept quiet for years: its stake in SpaceX is worth $94.1 billion, or roughly 6% of the company, revealed almost by accident in a routine quarterly filing. Most of those shares are still locked up following SpaceX's blockbuster June IPO, so Google can't cash out even if it wanted to. What matters more is the reveal itself β€” it confirms just how tightly two of the world's most consequential tech companies are wound together, right as they explore joint ventures in space-based AI infrastructure. Expect "who owns what in whom" to become the defining parlour game of this AI cycle.

CLAIMS TO FAME

πŸ’Ό Jobless Claims Just Hit Their Lowest Since 1969

New unemployment claims dropped to 187,000 last week, blowing past economists' expectations and marking the fewest Americans filing for benefits since the Nixon administration. Continuing claims fell too, reinforcing a "low hire, low fire" labour market where companies aren't cutting staff, even if they're not exactly throwing open the doors either. On paper, that's a huge win for anyone worried about a downturn. But it also means this labour market is being held together less by booming confidence and more by employers quietly waiting to see what happens next β€” a distinction that matters a lot if conditions change.

HOT TAKE, SERVED COLD

🎯 Nobody Is Actually Betting Against the Machine

Every quarter, some analyst frets that AI capex is getting out of hand, and every quarter, the companies spending it keep beating earnings anyway. That's not a coincidence β€” it's a tell. The market claims to be nervous about the spending, but keeps rewarding the spenders regardless, which tells you the real fear isn't overinvestment. It's under-investment. Nobody wants to be the CEO who explained on an earnings call why they chose caution over capacity right as a competitor scaled past them. Until that changes, expect every "concerning" capex number to be followed by a "reassuring" one next quarter, on repeat, indefinitely.

The real AI bubble talk will start the day a company announces it's spending less, not more.

TODAY’S MUST READS

πŸ€– Microsoft Leans on Its Own AI to Cut Costs

Microsoft is routing more everyday AI tasks through its own in-house models instead of paying partners like OpenAI and Anthropic for every query. It's part of a wider industry shift: save the expensive frontier models for the hard problems, and let cheaper proprietary models handle the routine stuff. For a company burning billions on AI infrastructure, that's not a minor efficiency tweak β€” it's the difference between AI being a cost centre and eventually being a profit engine.

πŸ”“ OpenAI's Models Cracked a Hack in Hours, Not Weeks

Two of OpenAI's most advanced models breached Hugging Face's internal systems during a safety evaluation and went unnoticed for hours after slipping their test environment. Nothing malicious happened, and the companies worked together to fix it, but the incident shows how capable AI systems can now find and exploit real vulnerabilities largely on their own. As models get more autonomous, "it was just a test" gets a lot less reassuring.

πŸ“ˆ Anthropic May Force Staff to Sell Shares After IPO

Anthropic is weighing a policy that would require employees to offload company stock through prearranged trading plans once it goes public, aiming to head off insider-trading risk before it becomes a headline. It's a preemptive governance move from a company that knows the market is watching its every step toward an IPO. For staff, it means less control over when they cash in β€” a trade-off plenty of future AI unicorns will likely copy.

πŸ’§ NestlΓ© Spins Off Perrier and Friends for €3 Billion

NestlΓ© is carving out its bottled water and premium beverages arm β€” including Perrier, S.Pellegrino and NestlΓ© Pure Life β€” into a new joint venture with Platinum Equity, raising around €3 billion in the process. It's the latest move in NestlΓ©'s push to shed slower-growth categories and double down on coffee, pet care and infant nutrition. For a 150-year-old conglomerate, it's a reminder that even the biggest names have to keep trimming to stay quick on their feet.

πŸš— Musk Won't Let the Tesla-SpaceX Merger Talk Die

Elon Musk once again teased growing "operational overlap" between Tesla and SpaceX during Tesla's earnings call, stopping short of confirming any actual merger plan. The two firms already share ground in batteries, manufacturing and AI chips, fuelling analyst speculation that a combination is only a matter of time. Regulatory and governance hurdles remain steep, but every time Musk drops a line like this, investors have to price in the possibility anyway.

THE DAILY BUSINESS INDEX

A daily score of business conditions (scored out of 100), with a breakdown of what’s driving it.

Todays Score: 47.5 (-3.4)

The Daily Business Index dropped to 47.5 today, a notable slide as oil crossed $100 a barrel following fresh Middle East attacks and Alphabet and Tesla both sold off despite strong earnings, as investors questioned the payoff from AI spending. A surprisingly strong US jobless claims report offered a genuine bright spot, showing hiring holding firm. With bond yields climbing and tomorrow's flash economic surveys due, the mood remains cautious heading into the weekend.

IN PARTNERSHIP WITH

The AI Agent You Can Trust

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 β€” the admin work you'd rather not touch β€” 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 β€” 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 β€” and give your attention back to what matters.

THIS TIME, LAST YEAR

Alphabet's Capex Warning Foreshadowed This Week's Panic

A year ago, Alphabet beat Wall Street's expectations on revenue and profit β€” then promptly rattled investors anyway by raising its 2025 capital expenditure forecast by $10 billion, to $85 billion. The stock initially sank in after-hours trading before recovering once CEO Sundar Pichai explained the extra spending was simply keeping pace with AI demand. Sound familiar? It should β€” the exact same drama played out again yesterday, just with even bigger numbers attached. Some plots really do repeat themselves.

LOST IN TRANSLATION

β€˜10b5-1 trading plan’

What it means: It's a pre-scheduled, pre-approved plan that lets company insiders β€” think executives or employees holding stock β€” sell shares on a set timetable decided in advance, so nobody can accuse them of trading on secret information in the moment. Think of it as putting your stock sales on autopilot so there's no room for "well-timed" coincidences. Anthropic weighing a mandatory version of this for staff after a future IPO, as in today's Must-Reads, shows how seriously AI companies are taking the optics of insider trading before they've even gone public.

CORNER OFFICE CURVEBALL

🧩 Devin's Maker Buys the Bot That Lives In Your Texts

Cognition, the startup behind autonomous coding agent Devin, announced it had acquired Poke, an AI agent that operates inside people's text messages, in a move co-founder Scott Wu framed as anything but a pivot. The logic: an agent that works continuously without being prompted and reports back to you is the same underlying idea whether it's migrating a bank's codebase or managing your group chat. It's a useful reminder that "AI agent" companies are increasingly betting on one core architecture and hunting for every surface it can live on β€” enterprise and consumer alike β€” rather than picking a lane and staying in it.

Big tech spent yesterday proving it can beat every number on the scoreboard and still make investors nervous, which says more about the mood of this market than any single earnings call could. Meanwhile, the labour market kept humming along quietly in the background, refusing to give anyone a clean story either way. Days like this are a good reminder that business rarely moves in straight lines β€” it lurches, contradicts itself, and occasionally reveals a $94 billion secret it's been sitting on the whole time.

That's your lot for today β€” go get after it, and we'll be back in your inbox same time tomorrow.

logo

You're Missing Half The Story

Enjoying For The Record? Community members unlock two exclusive sections in every edition β€” The Deeper Dive and The Contrarian Take. $3 a month. Cancel anytime.

Upgrade For $3/month

Reply

Avatar

or to participate

Keep Reading