There’s always an opportunity for latecomers—you can catch up! $100M in open interest is just the beginning. The next goal: $1B in open interest. Currently, entropy’s main trading pairs are ANT and OAI, and the emergence of pairs like PEARL also shows the team’s sharp market intuition. Aside from VAR, entropy is arguably one of the most successful projects to date.
New energy vehicle insurance is really expensive—I found it’s roughly twice the cost of insurance for gasoline cars. Previously, for my Porsche Cayenne in its third year, with a vehicle damage coverage of over 700,000 RMB, the premium was only 5,800 RMB, and the agent even gave me a rebate of over 1,000 RMB, bringing the annual cost down to just over 4,000 RMB. For my NIO car in the first year, with damage coverage around 600,000 RMB, I remember the premium was over 11,000 RMB. For my Xiaomi car in the second year, with damage coverage around 300,000 RMB, the quote was 8,000 RMB—even after removing miscellaneous fees, it still came to over 7,000 RMB. I can’t help but wonder why new energy vehicle insurance is generally so expensive overall. However, except for the first year, I don’t recommend buying new energy vehicle insurance through apps or dealerships. Go through an insurance agent instead—you’ll save hundreds, and with rebates, you could cut 1,200 to 1,500 RMB off the price. Save where you can, spend where you must—and never forget what you do for a living. The market’s a bit bearish—those who understand, understand.
OpenAI's response to Claude's launch. Reset twice weekly for free. However, the inference speed is still terrible. 🤣
Previously, I thought Musk was foolish for spending $60 billion on Cursor. Now, I think Musk is pretty sharp. But Grok BOT’s moat isn’t that strong—Codex has DOT, Claude has Duet. However, for finance professionals, Grok BOT is a game-changer due to its deep integration with X. Beyond traditional news media, the most relevant financial and AI-related content is on X. But here’s an important point: If Grok BOT doesn’t compete in programming, it’s easy to push its capabilities as a personal AI assistant close to the ceiling. That’s because the highest ceilings—programming, video generation, 3D rendering, game development—are the most difficult engineering tasks. But how much money and effort does it take to compete in those areas? Like Muse, it’s enough to focus on users deeply embedded in its own ecosystem.
AI certainly isn’t in a bubble. The bubble is in beer. I’m only willing to call it an overheated investment in AI infrastructure. After all, no one wants to fall behind in the Fourth Industrial Revolution. After all, global internet companies, governments, and capital are all heavily investing in and building out AI. Do you really think everyone is foolishly investing in a bubble? It’s just that the overheated investment in AI infrastructure has caused hardware prices to skyrocket, violating Moore’s Law—the very principle that hardware should become more powerful while becoming cheaper. Performance gains are limited, yet prices have doubled or tripled. In terms of labor efficiency gains, global programmer salaries total roughly $2 trillion. A 10% reduction in workforce would save $200 billion. Combined, OpenAI and Anthropic’s annual recurring revenue doesn’t even reach $200 billion. This doesn’t even account for efficiency gains in pharmaceuticals, administrative work, film production, and other sectors. In 1860, the U.S. fought the Civil War over the control of enslaved labor, when its population was 31 million and farmers exceeded 5 million. With the widespread adoption of agricultural mechanization and large-scale farming, the U.S. now has a population of 340 million—but employs just over 3 million farmers and remains the world’s largest agricultural exporter. AI is not a bubble—just as the tractor appeared, it only makes the world more efficient.
Microsoft has approximately 220,000 employees and has issued separation agreements to 60,000, allowing voluntary resignation with severance. Of course, not everyone will leave—perhaps 20,000 to 30,000 will depart. Either way, Microsoft already laid off 6,000 employees in July. The median compensation for Microsoft employees—including salary and stock—is around $200,000. When factoring in associated costs like office space, each employee costs roughly $300,000 annually. Laying off 10,000 employees saves $3 billion per year; laying off 30,000 could save $10 billion annually. Microsoft’s capital expenditures this year are approximately $170 billion, all invested in Azure—these computing resources can be converted into revenue. Currently, GPU depreciation is amortized over six years, but in practice, their useful life far exceeds that. In truth, layoffs have minimal impact on cash flow—but finding a justification for them is inevitable. This is likely also why MSFT’s stock has remained so resilient.
The data Whoop can monitor isn’t any different from other products, since there are only a few sensors available. As for the algorithms— those are the least valuable part; every major company can easily fork them. Seriously, how can this thing be valued so highly? Google and Fitbit are already moving in—if Apple enters, isn’t that the knockout punch? A $359 annual subscription fee? Isn’t this just SaaS? Put a health-monitoring hardware shell on it, add a bunch of celebrity investors, and you get a sky-high valuation? Western companies are falling further behind in the hardware industry, relying mostly on marketing. Whoop’s contract manufacturer previously revealed that the actual hardware cost is around $15—extremely cheap. Meanwhile, selling anxiety and lifestyle trends seems to be the dominant startup formula in the West—think Lululemon, Alo, Whoop, Oura—and even gluten-free food. 😅
