Speaker 0:  meta is spending a 145,000,000,000 this year to turn their office into an ai laboratory, and the test subjects are their own employees. if you thought your boss was hovering before, wait until you hear about the software recording every single one of your keystrokes to build your digital replacement. meta's high stakes ai surveillance. um, they've pivoted, you know, from, uh, the metaverse and whatnot. they're kinda moving away from it, but not moving away from it depending on, uh, who you talk to. but they're definitely reducing their spending in that area. but meta has deployed the model capability initiative, mci, a mandatory software suite that tracks every mouse movement and keystroke of its us staff. the data is being used to train ai agents to replicate human work coinciding with 14,000 role eliminations. this is the ultimate quiet part out loud moment for big tech proving that the path to ai roi in this involves literal human de stealing and surveillance backed automation. what do you think about this, john?

Speaker 1:  yeah. this is not so before we dive into it, i think it was, um, jensen wong, the ceo of nvidia. he was saying, um, he expects his employees to be i can't remember how much, uh, spend on ai that he expects them to be. so, like, although this isn't surprising, this is scary that, like, obviously, this is the first one that's kind of come to the forefront and being covered. i think this was a new york times article. say i'm surprised, but this is, uh, it's controversial, i would say. so let's let's kinda get into the the nitty gritty. so according to the reporting, what is doing is they're they're requiring workers to adopt ai tools across coding, writing, product planning, communication, and internal operations. so one kind of revealing moment involved employee backlash against mandatory ai integrations on company laptops. so when workers asked whether they could opt out, meta ceo andrew bosworth, uh, he reportedly responded saying there's no option to opt out on your corporate laptop. and so ai adoption is essentially no longer voluntary. so what employees are worried about? they're worried about surveillance. they're worried about loss of autonomy. they're worried about declining craftsmanship, productivity tracking, and ai replacing parts of their own jobs. so what meta is saying is that the reason why they're doing this is to help train models, their own models, which to me on a high level makes sense. um, but i think this is coming from more of a place of fear, um, where meta, obviously, they believe the future of social media search, advertising, entertainment, and computing, it's gonna be fundamentally shaped by ai agents and generative systems, which is what was that one purchase they made? uh, i think we covered manus. what's that? manus. oh, yeah. yeah. but it's now getting reversed. um, so, essentially and feel free to to to push back on this. but if meta loses the ai race, its entire business model could weaken. um, but that's where we're at. what are your thoughts on this?

Speaker 0:  well, i think we're at a stage where every big move that a company does that seems to lead to reduction of headcount is always blamed on ai. yeah. doesn't matter. we're blaming it on ai. whether it's true or not, i think we are in a situation where the costs are up, uh, profits are down, so people are reducing, um, the cost internally and looking at who's actually productive and things are shrinking. and they're using ai as an excuse for that because they're making these massive dumps in ai, and then we hear the regrets, and then we see the articles that ai is not really replacing. it's 95% failure. so it's hard to see who's doing what. but let's take a step back and think about the reality that meta is in right now and, um, see if there's some truth behind this that we can kinda maybe shed some light on. so, uh, metal meta's capital expenditure has been revised upward for a hundred and twenty five to a 145,000,000,000. this is nearly double what the company spent in 2025. a significant chunk is for their nvidia blackwell ultra, uh, chip orders, which is 13,000,000,000. so they've got a huge capex related to ai, and they've got a positioning problem. this chart didn't turn out too good, but anthropics got 40% of enterprise ai market share. openai has got 27. google's got 21%. all others, that's everyone else, including meta, is combined, is under 12%. so they have very limited visibility, uh, by, uh, the customer focus. and so as a result, if people aren't using your ai, you're not training your ai, your ai isn't getting better faster. people realized recently that anthropic got it right by saying, let's focus on development because with development, we can do everything else. we can trade our models to train themselves and develop better. everyone else is just starting to catch up to that meme. the more people you have using your ai, the better your ai becomes, the better you can train your stuff. so what do you do? then you turn around and say, well, people aren't using our ai products. even internally in our company, we have all these things. we need to start forcing adoption. and not only that, it's the whole it's the start up mindset of eating your own dog food. right? so you gotta use your own ai. we wanna track how you're using your own ai. we wanna track all the re stuff around using ai because we need to train our ai to do better, to be able to do for everyone. and we just got, you know, 50,000, a 100,000. like, i don't know the number. we'll get to it in a bit. probably, uh, the employees are now being able to use, uh, forced to use this product every day, eight hours a day. that's a lot of data. is it really being trained to completely replace these employees, or is it being trained to make their ai better? what's the outcome here? i don't know. but everybody's trying to get as many users of their own ai to increase their training, um, and visibility and whatnot. and if you don't force people, like, to use your ai or your company's ai or your enterprise private ai, they'll use whatever. they'll use chatdtp. they'll use anthropic. they'll use that. and those things aren't private, so your information gets out there. so corporations have to do things. so that's kind of my mindset around this at least, um, the situation meta finds itself in. now i think of this update like a master chef being forced to wear a body cam and a heart rate monitor while they cook. management isn't just watching to see if the chef is working. they're using the footage to program a robot that can flip the same pans and then season the same sauces. once the robot gets this right, the chef is handed a pink slip. so i think that might be an overreaction because everyone using ai, uh, every time they type in the ai is training the ai no matter what ai you're using. but is it training you to replace your job, or is it training the ai to understand you better, to be able to respond better, to give you better results, not specifically to replace you. and i think there's a lot of fear around ai, um, that people think of ai as the terminator and a lot of business thing, whether it's coming from my job and it's gonna take my income and whatnot. and people aren't really fostering a, uh, more of an ironman situation. what are the tools we can give you that will enable you to do your job at the next level? and how do we make that happen and enable you? so we sit back and say in a situation of fear, the employees are saying they're gonna replace us. we have a huge morale issue. nobody wants to do this, um, and what meta is doing is wrong. so i'm kinda i don't know the motivation. i don't know, um, what happens in the management boardrooms, but i can see the business state, and i can see a business reason behind it that isn't we need to spy on our employees.

Speaker 1:  yeah. well, it's not just meta, though. silicon valley, they're they're kind of quietly transitioning from what we would call a perks culture to more of a performance culture, in my opinion. so, um, for years, most of these major tech firms, they emphasize flexibility, campus culture, experimentation, employee comfort, all that sort of stuff. but i feel that now the industry is shifting more towards leaner organizations, stack ranking, performance pressure, and ai enabled productivity expectations. and i think meta right now may be one of the clearest examples of this transition. so, like, the old move fast and break things culture is involving into move fast with ai or become obsolete. do you think what are your thoughts on that?

Speaker 0:  i absolutely think that's the case. um, we need to move fast with all this. but i also as we started talking briefly offline before we got started, some of this is also like you said, everyone's doing it. microsoft recall. yeah. um, they're tracking every key keystroke, taking a screenshot every eight seconds or whatever it is. it might be three seconds, and there are you know, you can go back in time and that stuff is being hacked all the time we see. so they're all doing it. yeah. um, so that aside, are they using this to figure out which employees are actually productive, which ones are actually doing work, which ones are, you know, working from home are, you know, only on their computer an hour a day. um, some of that is bound to be happening as well because, you know, with the work from home culture that we have now, like we're i was saying, we used to do management by walking around and then meetings and productivity reports and stuff. but a lot of times, you just can't tell until you see stuff. and when the age of ai, you're not sure how much is a person, how much is ai. and so you want peep cost companies wanna have a better control of their, uh, information, their data. they don't want public ai using it, and they also want their employees, you know, working a regular work hour. so we're in a situation here where people are training ai to replace their jobs all across the world. it wouldn't be any different in meta. um, i'm just not sure that what percentage of this is driven by let's replace jobs. 10%, 20%, and how much is driven by we are way behind in ai and need to train our models in order to get, uh, uh, the meta, um, the lama models, uh, being used by more companies. so the mci captures periodic screenshots, which are likely being processed, uh, by multimodal lms to bridge the gap between text code and visual ui execution. meta is paying a player ai research packages exceeding a 100,000,000 in one year, creating a massive internal wealth gap where the builders earn 500 times more than the users being monitored. um, despite the massive spend, meta's reality labs, uh, metaverse division is still hemorrhaging cash, losing over 4,000,000,000 in, uh, q one twenty twenty six alone. and i just wanted to, uh, mention ucedd, jensen huang. if an engineer's particular, those earning around 500,000 in base salary do not consume at least $250,000 worth of ai tokens per year, he would be deeply alarmed.

Speaker 1:  yeah.

Speaker 0:  so what do you think about this? um, is meta is cannibalizing its own culture to build an ai workforce or making its ai superior, uh, to compete? or is it all in the same thing betting that silicon efficiency will eventually outrun human resentment? would you stay at a job that recorded every keystroke to train your replacement, or is the meta paycheck no longer worth the privacy price tag? let us know what you think in the chat. we'd love to hear from you.