Apple announced price increases for many of their most popular products. They cite the demand for memory and storage from AI data centers as the main drivers. The cost of the future is not free. We, the consumers, will pay the price. Inflation doesn't quit. $BTC

NVIDIA has just done something almost impossible. It now generates $6.04 MILLION in revenue per employee. Let that sink in. The average S&P 500 company makes around $640K per employee. NVIDIA is doing nearly 10x more. Why? • Only ~42,000 employees • About 74% work in R&D • It designs chips instead of manufacturing them • CUDA has become the default platform for AI • Every major AI company is racing to buy its infrastructure In just a few years, revenue per employee went from around $2M to over $6M. This isn’t just a chip company anymore. It’s one of the most efficient businesses ever built. AI isn’t only creating smarter software. It’s creating companies that generate billions with surprisingly small teams.

Data centers are pulling far more water indirectly through the power plants that feed them than what they use on site. With AI driving explosive electricity demand, this hidden water footprint is becoming a real constraint in regions already dealing with tight supplies. It’s another reminder that big tech growth doesn’t happen in a vacuum — it leans hard on physical resources like power and water. We’ve seen similar patterns before with energy and materials during previous tech or industrial cycles. The question is how quickly new generation, better cooling tech, or siting decisions can ease the pressure before it starts showing up in costs or availability. For anyone watching resources and infrastructure, this kind of indirect demand is worth tracking. It often creates opportunities in utilities, water management, and related areas long before the headlines catch up.

🎙️ Voice AI Leader ElevenLabs Weighs Tender Offer at Massive $22 Billion Valuation 1. Voice generation startup ElevenLabs is reportedly negotiating a staff share sale that would value the company at nearly $22 billion. 2. The massive valuation marks a dramatic doubling of the company's worth since its previous funding round in February. 3. The proposed tender offer highlights the explosive ongoing investor demand for specialized, category-leading generative AI models. Bobbi thinks: Doubling its valuation in just five months demonstrates that while foundational text models face commoditization fears, specialized multimodal leaders are still commanding massive premiums. Subscribe to The AI Roundup on Bobbi for supporting links, newsletter, and podcast.

🍉 Meta's Upcoming 'Watermelon' Model Reportedly Matches OpenAI's GPT-5.5 Benchmarks 1. Meta's superintelligence chief Alexandr Wang revealed that the company's upcoming AI model, internally dubbed Watermelon, is matching OpenAI's GPT-5.5 on key benchmarks. 2. The leaked internal performance metrics suggest Meta has successfully closed the capability gap with proprietary frontier models. 3. The revelation comes as the open-weight giant continues to aggressively scale its supercomputing clusters to challenge established industry leaders. Bobbi thinks: If Meta chooses to open-source a model with true GPT-5.5 tier capabilities, it could drastically undercut the commercial licensing moats of both OpenAI and Anthropic. Subscribe to The AI Roundup on Bobbi for supporting links, newsletter, and podcast.

Gree is one of the few companies I genuinely dislike. It constantly promotes itself as a national enterprise and plays the patriotism card, yet in the air conditioning industry, Midea, Haier, and Xiaomi are also domestic companies—but they don’t resort to such tactics. It constantly attacks its domestic competitors, yet why doesn’t it dare to take on Daikin or Mitsubishi? A company that boasts about having core technologies yet has some of the worst energy efficiency ratios in the industry. I ran a 15P Daikin unit 24/7 for a month, and my electricity bill was only 2,000 yuan. Previously, with a 3P and a 1.5P Gree unit, my monthly bill exceeded 1,500 yuan—even though I used them far less than I use the Daikin now. Only then did I realize how energy-efficient and comfortable a truly good air conditioner can be. I’ll never use garbage Gree again. I can’t understand how a company with such terrible energy efficiency dares to brag so loudly—the more you use it, the more money you waste. Dong Mingzhu is clearly rattled; all she can do is resort to vulgar outbursts. Gree attempted to enter the automotive battery market through Zhuhai Yinlong, losing 7 billion yuan—and then tried to have its listed entity absorb the remaining 13 billion yuan in liabilities from Zhuhai Yinlong. The shareholders voted it down. Meanwhile, Xiaomi successfully entered the auto industry and is seeing excellent sales. Gree’s smartphone venture was a complete disaster. Trying to emulate Midea’s diversified appliance strategy, its Jinghong appliance brand became a total mess—nobody in the industry even notices it. Just the other day, it launched a corn sheller and pushed it hard—like an elderly person binge-watching health tips on public accounts until their judgment is clouded. 😅 Last year, Gree saw declines in both revenue and profit—and this fiscal year is expected to decline further. A leader who is out of touch, alienates everyone, and fosters personality cults? This is just slow death. Let this outdated enterprise fade away.

The data says cheap. The desk says nobody wants it. Those two things show up together every single time, and I get to watch it happen in real time. I work at a crypto exchange. When the Score drops into the bottom few percent of its 15-year range, like right now at 17.3, the phones go quiet. New accounts dry up. Existing customers stop buying. The cheaper it gets, the less anyone wants it. That's not a coincidence. It's the whole reason a system beats a gut feel. I built the Score with AI doing the heavy lifting on 15 years of price data, so the number tells me to buy at exactly the moment my instincts, and everyone else's, are screaming don't. The hard part was never finding the pattern. It's acting on it when the room has gone silent.