Big Shots Roundup #20: Pichai's Half-Trillion Revenue & Altman's Superintelligence Safety Unit

The big shots were sure busy this week. Pichai showed off earnings, Altman fired off a bunch of posts, and a European big shot is making some noise. Li Bo, want to break it down one by one? Sure, Xiaoyu. But let me start with a judgment call. Put all the voices from this week together, and the direction is actually really clear. In one sentence: the giants are flexing their muscles while fighting for control of the safety narrative. Let me explain. Go ahead, start with Pichai. On Wednesday he tweeted that Q3 wrapped up with an annualized revenue run rate of five hundred billion dollars, up twenty-five percent year over year. Yeah, for Alphabet to grow twenty-five percent at that size is a scary number. And the key is the notes he added after. You mean the search query volume one? He said year-over-year growth was the highest in five years, and he specifically called out commercial queries. Exactly. A lot of people thought AI would eat search, but Pichai just laid out the data showing AI is actually driving query volume up. That's a direct response to the naysayers. And cloud. He wrote that Google Cloud revenue grew one hundred three percent, and first-party model APIs are processing close to thirty-two billion tokens per minute, up forty percent quarter over quarter. That one-oh-three growth rate shows enterprises are actually spending money. Token processing up forty percent quarter over quarter is real demand running. He also mentioned Argon and Carbon leading the frontier, and Genie 4 as the new generation world model. Are you familiar with these two names? That's Google's model codename approach — they don't give specific version numbers, just talk about tiers. My read is, Google wants people to remember one thing: they're consistently leading. So the takeaway from the earnings is, Google is proving that AI investment has already turned into revenue? Right. The conclusion is, AI isn't a cost item for Google — it's a growth engine. And that'll influence how other companies tell their story to the board. Okay, second one: Altman. He posted '6.1 Terra is here,' and said he's really looking forward to a new era of computer use. The point isn't the number 6.1 — it's computer use. OpenAI is shifting the narrative from chatting to operating computers. Wait, isn't that the agent-does-the-work路线? Everyone's crowding onto that. Exactly. And he specifically emphasized the word 'new era' — clearly trying to position Terra as the step where agents take over the desktop. But Altman's heavier post this week was actually about safety. He said he's grateful Aleksander accepted the request to take on this task. I take that one pretty seriously. His exact words were that when it comes to confronting superintelligence, the safety and security challenges are enormous. That's interesting. On one hand, releasing a new model pushing computer use, on the other, specifically bringing someone in to handle safety. Not contradictory at all — they go together. The more you let models operate computers, the more you need to have the safety team in place first. Do you think this is proactive, or driven by external pressure? I lean toward both. Regulators and public opinion are watching, and internally they do need it. But publicly saying 'superintelligence' is about grabbing the power to define it. Third one: Europe. Arthur Mensch posted asking whether Europe should concede frontier AI to the US and China. He flat out said that would be a fundamental mistake. His exact words were 'build or surrender.' That's pretty heavy. He also replied to another post saying, if using nuclear energy is an obstacle, then we deeply apologize. That was responding to people questioning his compute energy consumption. The subtext is, we're even willing to go nuclear, and you're still bringing up energy? So Mensch isn't on the same channel as Pichai and Altman — he's fighting for Europe's position? Right, he's the contrarian type. The giants talk growth and safety; he talks sovereignty — that Europe can't be absent. One more, an interesting one. Alexandr Wang posted 'The Scale Mafia,' listing a long string of people who came out of Scale. That one resonated with me. He said Scale was the first batch of companies to max on IOI and IMO, and those people are now scattered everywhere. On his list there's the founder of Cognition, the founder of Modal, and the product lead for Claude Code. See, a company that started in data labeling became a talent incubator. The signal is, what's really scarce in this industry is people who can actually make things work. So putting all these voices together, what direction do you see? Three lines. One, AI revenue is starting to materialize. Two, computer use is becoming the main battlefield. Three, both safety and sovereignty are fighting for narrative control. What does that mean for regular people or for those of us building products? It means in the next six months, desktop agents will pop up密集ly, and compliance and safety capabilities will become a hard threshold for whether a product can get into big enterprise clients. So the conclusion is, on the surface the big shots are each saying their own thing, but underneath it's the same chessboard speeding up. Right, looking at this week it comes down to one sentence: whoever gets capability, revenue, and safety all running at the same time first gets the ticket to the next round.