Now, I expect that these companies will get better at recovering from these unexpected increases in load as they gain experience with the problem. Because of capacity constraints with those pricey GPUs, they can’t always scale their way out of these problem, but they can redistribute resources, and they can get better at load shedding and other sorts of graceful degradation to limit the damage of overload. And I bet that’s where they’re both investing in reliability today. At least, I hope so. Because this problem isn’t going to go away. If anything, I suspect their loads will become even more unpredictable as people continue to innovate with LLMs. Because AIs don’t seem to do any better at predicting the future than humans.
The first: These are the ultimate in capital-intensive enterprises. Musk announced SpaceX’s intention to build 10,000 fully reusable rockets, each over 400 feet tall. At a cost that Payload Research estimates at $35 million each, that’s $350 billion in cash for the likes of krypton-gas burners, solar arrays, and stainless steel alloy. xAI is a major builder of high-cost data centers than run such products as its Grok chatbot. In 2025, it reportedly burned through $8 billion in cash, primarily to fund such behemoths as its $20 billion “MACROHARDRR” facility in Mississippi. The upshot: These aren’t low investment software plays that if successful, could easily post 35% net GAAP margins. In general, it’s extremely rare for “manufacturers,” whether it’s in aircraft or data generators, to achieve such lofty levels of profitability.
。新收录的资料对此有专业解读
/tmp → /sysroot/tmp
Small inputs seem to work fine,
assert widen([0, 1, 2], Var(2)) == Var(2)