The X-CEO

Linda Yaccarino is out as CEO of X.

She took over when the place was basically a radioactive wasteland for advertisers. Two years later? Still messy, but no longer a full-blown disaster.

Yaccarino helped scrape off (some of) the stank, even if Musk kept smearing it back on, and now ad revenue is finally expected to grow this quarter. First time in four years.
Still nowhere near pre-Musk levels, but the wheels are not falling off anymore.

Whoever steps in next better have thick skin, and probably a helmet. Working with Elon is hard enough. Working with Elon plus his new political party is a whole new circus.

X, by the numbers:

Since Musk bought the platform, it has been on a rollercoaster. Revenue has been cut in half, advertisers bolted, and monthly users dropped 15 percent, with even steeper losses in the U.S. The platform's value fell to 9 billion dollars at one point, far from the 44 billion Musk paid. A recent deal claimed it is worth 44 billion again, but that feels more like tech-world fan fiction than fact. And only about 1 in 400 users pays for X, so the subscription model is not carrying much weight.

Here is the plot twist. X is technically more profitable than before. Musk cut the staff from 8,000 to just 1,500, slashed costs, and ran the place lean. It is like the house is half-burned down, but the heating bill is lower. Still, X owes 1.2 billion dollars a year in loan payments, and unless something changes fast, that bill is going to hit hard.

Rapid Fire

🪖 The AI talent war is turning into a full-on soap opera. ChatGPT just poached four top engineers from Tesla, Meta, and xAI in one week. That comes right after Meta raided OpenAI not long ago. This is not just about paychecks, even though some companies are throwing out ten million dollar offers. It is about bragging rights, influence, and who gets to build the smartest product in the room.

At this point, AI engineers are the hot girls in school. Everyone wants them. No one can keep them. And it matters. These talent swings shape who builds the next big thing, who is falling behind, and who is about to lose their seat at the table.

🇮🇳 India just gave Starlink the go ahead. After years of red tape and waiting, Elon’s satellite internet play can finally go live in the world’s second most populated country. There is still a lot of work ahead. They need to grab spectrum, build ground stations, and pass security checks. But the biggest barrier is now out of the way.

Starlink is already signing deals and plans to set up three gateways across India. The focus for now is enterprise and government clients, but consumer access is not far behind. With Amazon’s Kuiper still sitting on the sidelines, the satcom race in India is heating up. And it matters, because whoever wins this market is not just selling internet. They are shaping who connects the next billion people online.

🩺 A new MIT study found that medical AI chatbots can give worse advice if you make a typo. Something as small as an extra space, casual language, or saying “maybe” made the bots more likely to tell patients not to seek medical help. The researchers tested GPT 4, Meta’s LLaMA, and a medical-specific model across thousands of patient messages. The actual symptoms stayed the same, but once the language got a little less polished, the advice got noticeably worse.

It gets worse. The study found women were disproportionately told to self manage their symptoms, echoing the gender bias already baked into real life healthcare. And yet, hospitals are already using these tools to triage real people. This is not just a weird tech glitch. It is a reminder that AI systems can quietly reflect, and even amplify, the same old problems we have offline.

Polling Zone

Meme Check

Recently Deployed

For everyone who has opened Blender, cried a little, and closed it five minutes later, this one is for you. Tencent just released Hunyuan3D PolyGen, a name that feels like it was made by committee, but the tool itself is impressive. It promises art grade 3D models, built specifically for the people creating the digital worlds we play in. And unlike most AI demos that look good on paper and fall apart in practice, this one is already being used in real game studios.

So what does it actually do? It takes rough point cloud data and turns it into clean, high quality 3D models that are ready to drop into a commercial game. It handles both triangle and quad faces, which is huge for anyone who has struggled moving assets between tools. Under the hood, it uses an AI system that predicts and builds geometry piece by piece, like writing a sentence in 3D. It even improves over time by learning what good work looks like. Tencent’s own teams say it has cut their asset creation time by more than 70 percent. That means less time wrestling with wireframes and more time focusing on the creative work that actually matters.

Tech Radar

Nvidia just became the first company ever to hit a $4 trillion market cap. Six years ago it was worth $144 billion. Now it is more valuable than Germany. That kind of money could fund NASA for 160 years or give every person on Earth a $500 check. Or if you are feeling bold, just buy Nvidia.

What is driving the rocket ship? AI. Nvidia makes the chips behind everything from ChatGPT to self-driving cars, and right now no one else comes close. As AI takes over more of the economy, Nvidia is not just selling tools — they are the toolbox.

That’s a wrap for today.

That’s all for today.

X is still standing, somehow. Nvidia just became the most valuable toolbox on Earth. And AI chatbots might be gaslighting you over a typo. The future is weird, fast, and full of plot twists — but we’ll keep breaking it down for you, no fluff, no BS.

See you Thursday. Bring coffee.