Meanwhile, in Beijing

Sometimes, as a self-absorbed American, it’s easy to forget that tech innovation doesn’t only happen in Silicon Valley.

Remember a few months ago when DeepSeek dropped, sent markets into a panic, and reminded everyone that other countries can actually build stuff too? Well, here’s another wake-up call. 

Moonshot AI, a China based company, just released a new open source model to compete in the Chinese AI market. If China isn’t leading in AI, they are a very close second, so any serious update from across the Pacific is worth paying attention to. Especially on a slow Sunday like this one.

What is Moonshot AI and what did they release

Moonshot AI is a Beijing-based startup that popped up in 2023 with one big goal—build smarter AI that could eventually rival human intelligence. Their main product, the Kimi chatbot, made waves for handling absurdly long conversations—up to 2 million Chinese characters in a single go. Backed by heavy hitters like Alibaba and Tencent, they shot up to a $3.3 billion valuation and earned a spot as one of China’s top contenders in the global AI arms race.

So what’s new? They just dropped Kimi K2, an open source model built to claw back relevance in China’s cutthroat AI scene. It’s better at writing code, juggling complex tasks, and playing nice with tools—basically, it’s a smarter, more useful sidekick. Moonshot says it even outperforms DeepSeek’s latest release and goes toe-to-toe with U.S. models from companies like Anthropic.

Why does this matter?

Because it is not just about performance benchmarks or fancy upgrades. It is about power and reach. While U.S. companies like OpenAI and Google are locking their best models behind paywalls, invite-only access, and endless red tape, Chinese companies are going in the opposite direction. They are open sourcing everything, letting developers, startups, and researchers around the world build on their models without begging for permission.

This is not just a nice gesture—it is a smart strategy. Open source spreads fast. It gets embedded into products, workflows, and developer communities. And the more people who use your tech, the more influence you have over the direction of the industry. Moonshot knows this. 

By releasing Kimi K2 to the public, they are trying to win back relevance, attract talent, and remind everyone that they are still a serious player. If Chinese models start showing up in tools, apps, and classrooms everywhere, it shifts the balance. It becomes less about who has the most advanced model in theory and more about who is actually being used in practice. Moonshot is betting big that Kimi K2 can help them take back some of that ground.

Rapid Fire

💰 xAI is reportedly looking to raise another $10 billion, aiming for a $200 billion valuation. That would mark a massive leap from the $18 billion valuation it had just over a year ago. Musk denies the reports, saying the company has “plenty of capital,” but filings and leaks tell a different story. In just the past two months, xAI raised $10 billion through a mix of debt and equity, and sold another $300 million in shares. Sources say the new round could kick off as early as next month, and it would be the third big share sale in less than 60 days. The Saudi sovereign wealth fund is also rumored to be involved, adding some global firepower behind the scenes.

So what’s fueling the push? Part of it is timing—xAI just rolled out Grok 4, Musk’s latest shot at the chatbot market, and saw the CEO of X step down in the same week. But reports also say the company is burning through $1 billion a month, which puts pressure on the runway. If the funding goes through, xAI would become one of the most valuable private companies on the planet. It is a clear signal that Musk wants to keep pace in the AI arms race, whether to compete with OpenAI or bake smarter tools into Tesla and X. Either way, this is not just about valuation. It is about staying relevant—and staying alive.

🔢 In May, thirty of the world’s top mathematicians gathered in secret in Berkeley, California, to take on an AI chatbot powered by OpenAI’s new reasoning model, o4-mini. The goal was simple: stump the bot with problems even seasoned researchers would struggle with. But over two days of testing, the AI blew them away. It solved complex PhD-level questions in minutes, showed its work, and even got a little cocky about it. One mathematician watched in stunned silence as the bot cracked a number theory problem live, casually announcing at the end that no citation was needed because it computed the answer.

The gathering, organized by Epoch AI, wasn’t just for kicks—it was part of a larger effort to benchmark how far reasoning models have come. The verdict? A lot farther than most experts expected. The bot performed like a top-tier grad student, just faster and with more confidence. And while the researchers did manage to craft ten questions the AI couldn’t solve, the event left many wondering what the future of math looks like when machines can already outpace some of the best minds. As one participant put it, the real danger might not be in what AI knows—but in how confidently it acts like it knows everything.

🎮 A lot of people think there’s a secret shadow government pulling the strings behind the scenes. If that were true, Nintendo would absolutely have a seat at the table. Over the weekend, a popular piracy site was taken down by the FBI—one that let gamers download titles and mods for the brand-new Switch 2.

Let’s be real: I know plenty of people who’ve pirated stuff before, mostly just to watch sports (please don’t come for me, FBI). And they’ve gotten away with it for years. But the Switch 2 has been out for less than two months, and Nintendo already has entire sites getting nuked—by a U.S. federal agency. This is a Japanese company we’re talking about. If that doesn’t scream power, I don’t know what does.

Tech Radar

Intel just spun out its robotics unit RealSense and gave it a $50 million check to go do its own thing. The company makes tech that helps robots see, move, and avoid smashing into stuff, which more companies seem to want right now. RealSense is now led by Nadav Orbach, a longtime Intel exec, and the plan is straightforward: build new products, meet demand, and cash in on the rising interest in physical AI. The money came from MediaTek and Intel’s own venture arm, which is also being spun out. Intel is still holding onto a small stake, just in case this becomes something big.

So why does this matter? Because robotics is becoming the next AI land grab. Tesla, Amazon, Nvidia—they are all pouring money into it. Analysts think humanoid robots could become a five trillion dollar market by 2050. Intel has been cutting costs after a rough stretch, and spinning off RealSense is their way of saying they are happy to play a supporting role if the upside looks good. It is another sign that AI is not just about chatbots anymore. It is getting physical.

Recently Deployed

Walmart is giving its 1.5 million U.S. associates an AI upgrade. From shift planning and task management to real-time translation and a beefed-up chatbot, the company is rolling out tools designed to make day-to-day work a lot smoother. One new system cuts shift scheduling time from 90 minutes to 30. Another lets workers talk to customers in 44 languages without missing a beat. And the chatbot? It now turns corporate how-to guides into plain instructions, like how to handle a return without a receipt. Think less clunky process, more "hey, here’s how to do this" on demand.

This is not just tech for tech’s sake. Walmart knows that keeping workers happy, efficient, and not buried in outdated systems is good business. It also knows it is sitting on one of the largest frontline workforces in the country. The message is clear: invest in your people, give them smarter tools, and you might just move faster than the competition. It is also a reminder that the future of work is not just about robots replacing jobs. Sometimes, it is about AI making a hard job suck a little less.

That’s a wrap.

That’s all for today. AI is rewriting the rules across the board—whether it’s powering Walmart’s workforce, spinning off billion-dollar robotics arms, or casually solving PhD-level math problems for fun. The players are global, the stakes are rising, and the weird just keeps getting weirder.

Catch you next time. Same feed, slightly smarter machines.