Daily Briefing
SUNDAY, 1 MARCH 2026
It's March 1st, 2026, and somewhere between a Hollywood megadeal, a war people are actively betting on, and Olivia Dean winning everything in sight, there's a lot to untangle.
Let's start with the biggest number in the room. One hundred and eleven billion dollars. That's what Paramount is paying to acquire Warner Bros. Discovery — a deal that, if it closes, reshapes Hollywood more dramatically than anything in a generation. Two studios, both of which have spent the last few years lurching from one crisis to the next, deciding that the answer to their problems is each other. Maybe it works. Maybe you've just combined two struggling companies into one enormous struggling company. The cynical read is that scale is the last competitive advantage left when you're getting eaten alive by Netflix. Which brings us, neatly, to Netflix — because Netflix was apparently in talks to buy Warner Bros. themselves, and then... walked away. The reason, reportedly? Co-CEO Ted Sarandos told Donald Trump, quote, "I took your advice." That sentence contains multitudes. What we do know is that Netflix, after years of being the existential threat, is now apparently taking business direction from the Oval Office. Make of that what you will.
From Hollywood to the Pentagon. Anthropic has had a strange week. They've been in fraught negotiations with the US military — details are murky, but the friction was real enough to make headlines — and the weird outcome is that Claude, their AI assistant, shot up to number two in the App Store. Nothing markets a product like a good controversy. But there's a deeper story here, and it's one worth sitting with. Anthropic, like OpenAI and Google DeepMind, has spent years making serious, public promises about responsible AI governance. The implicit deal was: trust us, we'll police ourselves, we don't need heavy regulation because we take this seriously. And that argument held some water when it felt like regulation was coming anyway. But now, in this political climate, external oversight looks increasingly unlikely. Which means the self-governance promises are all that's left — and those promises have no enforcement mechanism whatsoever. The trap Anthropic built for itself is the trap they all built: make enough noise about safety to sound credible, acquire enough power that it matters, and then discover there's nothing actually stopping you from doing whatever you want. That's not an accusation. It's a structural problem. And it's getting harder to ignore.
Meanwhile, Polymarket — the prediction market platform — is defending its decision to allow people to bet on war. Specifically, on when the US would next strike Iran. That has now happened. People have died. And Polymarket is calling its service "invaluable." I'll be honest with you: prediction markets do have genuine informational value. Aggregating what people are willing to put money on can tell you things polls can't. But there's a version of this that starts to feel less like a forecasting tool and more like a casino with better branding. Betting on the timing of an airstrike — one where real human beings are killed — and then defending it while the rubble is still warm is a particular kind of tone-deafness. The site will probably be fine. That's the more unsettling part.
On a completely different note — and I mean that as a genuine gear shift — the Brit Awards happened last night in Manchester, and Olivia Dean won four of them. Artist of the year, album of the year, song of the year. Twenty-six years old. If you haven't listened to her yet, this is your third or fourth warning. Beyond the trophies, what was interesting was the mood backstage. Artists including Wolf Alice, Wet Leg, and Loyle Carner used the platform to talk about Reform UK's rise — "scary times" was a phrase that came up more than once. British musicians have always had a complicated relationship with political commentary, but right now that reticence seems to be dissolving. Whether that translates into anything beyond red-carpet quotes remains to be seen.
And finally, Neil Sedaka died this week at 86. Breaking Up Is Hard to Do. Calendar Girl. Laughter in the Rain. Solitaire. A career that spanned from Brill Building songwriter-for-hire in the 1950s to genuine pop star to emphatic 1970s comeback — which almost never happens. He was a genuine craftsman, the kind of songwriter who understood that a great melody is an act of generosity toward the listener. Eighty-six years. Not a bad run at all.
That's the week's first Sunday. Hold it loosely.
Archive
It's the last day of February, and the world is somehow both falling apart and quietly getting more interesting at the same time.
Let's start with Anthropic, because this one is genuinely strange. Pete Hegseth — yes, the Defense Secretary — has designated Anthropic, the AI safety company behind Claude, as a supply chain risk. And Trump had already announced on Truth Social that Anthropic products were being banned from the federal government. Nearly two hours between the Truth Social post and the official designation, which tells you everything you need to know about how this administration handles policy. Anthropic, for the uninitiated, is the company founded by former OpenAI researchers explicitly to make AI safer and more interpretable. The irony of the "safety-focused AI lab" being flagged as a national security risk by an administration that has shown very little interest in AI safety is thick enough to cut. OpenAI, by contrast, has been cozying up to Washington. There's a lesson in there somewhere about how proximity to power works.
Speaking of the federal government and technology — India has moved to block Supabase, the open-source database platform that's become something of a darling among developers globally. India is one of Supabase's biggest markets, so this isn't a rounding error. The block appears to be an administrative order, the kind that often sweeps up legitimate platforms alongside whatever the actual target was. It's patchy right now — some users can get through, some can't. But it's a reminder that the open internet is really more of a patchwork quilt than an open field, and that any company building infrastructure on the assumption of unrestricted global access is making a bet that governments keep finding new ways to complicate.
Here's a piece of genuinely good technical news, which feels worth savouring on a February Saturday. Google has figured out how to make HTTPS quantum-resistant by compressing what used to be 2.5 kilobytes of cryptographic data down to 64 bytes using something called Merkle Tree Certificates. It's already in Chrome. The reason this matters is that quantum computers — while not yet capable of breaking today's encryption — are getting closer, and the principle of "harvest now, decrypt later" is already in play. State actors are collecting encrypted data today on the assumption they'll be able to crack it in the future. Google moving to quantum-proof the web's basic security layer is one of those infrastructure stories that gets almost no attention but affects essentially everyone. Good. They should do more of this and less of everything else they do.
NASA has pushed back its Moon landing plans again. Artemis III, which was supposed to attempt an actual landing in 2027, will now be a test flight only. The Moon landing is being pushed to Artemis IV in 2028. NASA said they're "increasing cadence," which is the kind of phrase that organisations use when they're announcing a slowdown. The Air Force's new ICBM situation is weirdly complementary to this — the Sentinel missile is apparently nearly ready to fly but there's nowhere to put it, because the assumptions built into the original strategy, and I'm quoting here, "obviously didn't come to fruition." There's a theme emerging in American defence and space programmes right now, which is that ambition keeps outrunning execution, and the gap between what gets announced and what gets built is becoming too large to paper over.
In music — Neil Sedaka has died at 86. Breaking Up Is Hard to Do, Oh! Carol, Calendar Girl — songs that shaped what pop music sounded like before anyone had really defined what pop music was. He wrote for himself and for dozens of other artists across six decades. Genuine craftsman. Worth a moment.
And Rolling Loud Australia has been cancelled one week before it was due to happen in Sydney and Melbourne. One week. If you'd bought flights and booked accommodation, well — this is the festival industry's version of a supply chain risk designation. No official reason given beyond the usual vague statement. The festival economy is fragile in ways that keep surprising promoters who should probably know better by now.
That's February done. Don't let the door hit it on the way out.
It's the last Friday of February 2026, and the news is moving fast enough to give you whiplash.
Let's start with the story that is genuinely historic. Paramount — David Ellison's Paramount — is acquiring Warner Bros. Discovery. That means HBO, CNN, Warner's entire film catalogue, all of it, folding into one massive entertainment conglomerate. Netflix had been circling Warner Bros. for months, and then walked away. Now, whether that was strategic discipline or a genuine miscalculation, we'll probably be debating for years. What you're left with is an industry that looked fragmented and chaotic eighteen months ago and is now consolidating at extraordinary speed. Two centibillion-dollar entities becoming one. The streaming wars are over. This is the post-war map being drawn.
And speaking of consolidation — Jack Dorsey just cut Block's workforce in half. Half. Not a round of layoffs, not a "restructuring," not a "right-sizing." He halved it. Dorsey's been pretty upfront about his admiration for Elon Musk's management philosophy, and this feels like a direct application of it — the bet that most companies are carrying twice the headcount they actually need and that the pain of cutting is worth the clarity it creates. Maybe he's right. Maybe Block runs leaner and faster now. But the framing is what's striking: Dorsey didn't just announce the cuts internally. He basically said your company should do this too. That's either visionary or deeply uncomfortable, depending on whether you still have a job.
The story I suspect will matter most in the long run, though, is what Anthropic just did. The Pentagon issued what amounts to an ultimatum — give us unrestricted access to your AI systems — and Anthropic said no. Publicly, clearly, with CEO Dario Amodei saying he "cannot in good conscience" accede to those demands. The specific sticking points are lethal autonomous weapons and mass surveillance. Anthropic's position is that there are things their technology should not be used for, full stop, regardless of who's asking. Now, plenty of AI companies have defence contracts and are happy to have them. Anthropic is drawing a line that most of its competitors won't draw, and doing it at a moment when the pressure to fall in line with federal priorities has never been higher. Whether you think that's principled or naive probably depends on your priors. But it's a real thing that a real company just did, and the consequences — commercial and political — are going to be significant.
Something slightly more mundane but worth flagging: Plaid just got valued at eight billion dollars in an employee share sale. That's up thirty-one percent from its April valuation. Plaid, for context, is the infrastructure that quietly powers most of your fintech apps — the thing connecting your bank account to whatever app you're using to move money around. The fact that it's climbing back toward its 2021 peak valuations tells you something about where investor sentiment on fintech is sitting right now. The froth is coming back, selectively.
On the wearables front, Ultrahuman is launching a new smart ring — the Ring Pro — with a fifteen-day battery life and a four hundred and seventy-nine dollar price tag. Ultrahuman's been in a messy public dispute with Oura, the market leader, and this feels like their serious attempt to pull U.S. market share. Whether a smart ring at nearly five hundred dollars finds an audience when Oura already owns the mindshare is the real question. The spec sheet is impressive. The brand recognition gap is not small.
And before you go — briefly, because it deserves a mention — Suno, the AI music generation company, just hit two million paid subscribers and is reportedly generating three hundred million dollars in annual recurring revenue. That number is staggering for a product that essentially lets you type a prompt and get a song back. There are multiple copyright lawsuits piling up alongside those revenue figures, which is probably the more important part of the story. The music industry's legal reckoning with AI-generated content is happening right now, and Suno is at the centre of it.
That's your Friday. Anthropic said no to the Pentagon. Warner Bros. said yes to Paramount. And Dorsey thinks your company has too many people.
Have a good weekend.
It's fifty years since a dingy bar in downtown Manhattan accidentally invented the future of rock music, and somehow that's not even the most consequential thing happening in culture today.
Let's start with the date — Thursday, February 26th, 2026 — and with something that's been quietly sitting in the background of every conversation about social media and teenagers for years. Meta is finally doing something concrete about it. Starting next week, Instagram will send alerts to parents when a teen on a supervised account repeatedly searches for terms related to self-harm or suicide. Not just a nudge to the kid, not just a resources pop-up — an actual notification to the parent saying, go check on your child. Now, you can have the whole privacy debate, and it's a legitimate one. But if we're being honest, the reason this matters is that we've spent the better part of a decade watching platforms optimise for engagement while teenagers struggled, and this is at least a step in the opposite direction. It's not a fix. It's a signal. Whether Meta follows through with more than optics is the question worth watching.
From one kind of platform anxiety to another. Salesforce had its year-end earnings this week and CEO Marc Benioff came out swinging against what he called — I love this — the "SaaSpocalypse." The idea that AI agents are going to hollow out the entire software-as-a-service industry by just... doing the jobs that enterprise software used to manage. Benioff's basically saying: we've been through panics before, we're fine, trust us. And the numbers weren't bad. But here's the thing — the concern isn't that Salesforce disappears overnight. It's that AI agents start completing tasks that used to require a CRM login, and the stickiness that made SaaS so lucrative starts to erode. Benioff knows this. That's why he's been repositioning Salesforce around its own agentic products. The bravado is partly performance. Watch what they build, not what they say.
Speaking of agents — Anthropic just acquired a Seattle startup called Vercept, which built AI that can operate a computer the way a person would. Click, type, navigate, complete tasks inside actual applications. This is the computer-use category, and it's moving fast. What makes this acquisition particularly interesting is the context: Meta apparently poached one of Vercept's founders before the deal closed. So Anthropic moved quickly. And now you've got the two biggest names in frontier AI — Meta and Anthropic — both scrambling for the same capability. The race to build AI that doesn't just answer questions but actually does things is the real competition right now, and Vercept just became a small but telling piece of that puzzle.
Google, meanwhile, is pulling its robotics moonshot — a unit called Intrinsic — back under the main corporate roof after five years as an independent Alphabet project. Which sounds like administrative housekeeping until you realise what it signals. Google is done letting physical AI sit off to the side as an experiment. They're folding it into the core business, which means it gets resources, it gets integration, and it gets treated as a real product priority. The phrase being used is "physical AI" — the idea that intelligence needs to exist in the world, not just on a screen. Between this and the computer-use agent arms race, the through-line is clear: the next phase of AI isn't about better chatbots. It's about AI that acts.
Now. CBGB. Fifty years ago this year, a former marine and folk singer named Hilly Kristal opened a grimy bar in Manhattan's East Village and somehow stumbled into hosting the Ramones, Blondie, Television, Talking Heads — the entire foundation of American punk and new wave. There's a 101-track box set out celebrating it. And yes, by all accounts the bathrooms were absolutely rank. But what CBGB represents is something the music industry keeps forgetting and keeps having to rediscover: scenes happen in uncomfortable rooms with terrible margins and no algorithm pointing anyone there. The streaming era has given artists global reach and given them almost nothing in return per play. Spotify just announced it paid out over £860 million for UK artists' music last year, which sounds enormous until you do the per-stream math. And separately, a court just overturned a 138% rate hike that BMI tried to impose on concert promoters. Everyone in the music business is fighting over the same shrinking pie. CBGB thrived because nobody was trying to monetise it properly. There's a lesson in there somewhere.
That's your Thursday. Don't let anyone tell you the bathrooms have to be clean for something important to happen.