Will Meta's $14 Billion Bet on AI Ever Pay Off? (cnbc.com) 65
"A year after spending over $14 billion to bring in Alexandr Wang and a group of his top Scale AI engineers to revamp its artificial intelligence efforts, Meta is at least back on the map in AI," reports CNBC, "though it's still far behind OpenAI, Anthropic and Google in the market."
Wang's big accomplishment was the delivery of the Muse Spark AI model in April, marking Meta's first jump into proprietary foundation models and away from a strict adherence to open source, or open weight as it's more commonly called in AI... "Meta needs to provide more proof points of both adoption and commercialization," said Ralph Schackart, an analyst at William Blair who recommends buying the stock. "Investors are looking for Meta to monetize a new AI-first product, beyond the substantial positive impact AI is having on enhancing the advertising models." Wall Street, at least so far, is unimpressed. Meta's stock is down 18% over the past 12 months, the worst performer in the megacap group, along with Microsoft, which has its own challenges in AI. That's even after Meta reported 33% revenue growth in the first quarter, the fastest rate of expansion for any period since 2021.
For Meta, the problem started with what some industry experts called, in hindsight at least, a strategic blunder. The company jumped into AI with its Llama family of models, offering an open-source approach that allowed developers to freely tinker, while the other big model makers charged for access. In April of last year, Meta's release of Llama 4 fell flat, failing to captivate developers and leading Zuckerberg to reconsider his company's approach to AI development... Since the release of Muse Spark, Meta has unveiled new AI and business-related subscription plans as part of an effort to expand its business beyond online ads. Historically, it hasn't worked. Meta still counts on ads for 98% of revenue. Schackart said he wants to see "tangible evidence of a growing list of new, AI-first products created by Muse Spark, even if monetization lags." He said that's "what investors are looking for."
No matter how good Wang's model may be, Zuckerberg has a high hill to climb with developers coming off the Llama debacle. "I think the AI community largely ignores Meta at this point," said Rob May, CEO of the startup Neurometric, which works in the realm of token engineering.... Krish Subramanian, the CEO of consulting firm KOI AI and former product head at IBM Consulting, said developers are more excited about Google's AI models than what Meta is offering. The appeal of Llama was that it specifically targeted developers wanting open-weight alternative models, while with Muse Spark, Meta has made little effort in that direction, he said. "The lack of developer trust will come back to hit them if they don't focus on third-party developers," Subramanian said, noting that it took years for Microsoft to regain trust from open-source coders during the early days of Azure. "To just focus on a walled-garden kind of an ecosystem and ad revenue as the main source of income, they probably will never become the big player," he said.
A Meta spokesperson pointed to Wang's recent comments about the company's continued support for the open-source ecosystem, and said Meta still plans to offer outside developers access to Muse Spark's underlying technology via an API, as it previously announced. "We're already testing with some early partners, and look forward to releasing it this month," the spokesperson said.
"That Zuckerberg's metaverse and virtual reality ambitions have generated over $80 billion in total losses since late 2020 makes the AI pitch a tougher sell," the article points out, citing this observation from Howard Yu, business professor at Switzerland's International Institute for Management Development.
"He's running out of the space for his credibility to last," Yu said. "I think the virtual reality foray may have burned up a lot of his goodwill in front of investors."
For Meta, the problem started with what some industry experts called, in hindsight at least, a strategic blunder. The company jumped into AI with its Llama family of models, offering an open-source approach that allowed developers to freely tinker, while the other big model makers charged for access. In April of last year, Meta's release of Llama 4 fell flat, failing to captivate developers and leading Zuckerberg to reconsider his company's approach to AI development... Since the release of Muse Spark, Meta has unveiled new AI and business-related subscription plans as part of an effort to expand its business beyond online ads. Historically, it hasn't worked. Meta still counts on ads for 98% of revenue. Schackart said he wants to see "tangible evidence of a growing list of new, AI-first products created by Muse Spark, even if monetization lags." He said that's "what investors are looking for."
No matter how good Wang's model may be, Zuckerberg has a high hill to climb with developers coming off the Llama debacle. "I think the AI community largely ignores Meta at this point," said Rob May, CEO of the startup Neurometric, which works in the realm of token engineering.... Krish Subramanian, the CEO of consulting firm KOI AI and former product head at IBM Consulting, said developers are more excited about Google's AI models than what Meta is offering. The appeal of Llama was that it specifically targeted developers wanting open-weight alternative models, while with Muse Spark, Meta has made little effort in that direction, he said. "The lack of developer trust will come back to hit them if they don't focus on third-party developers," Subramanian said, noting that it took years for Microsoft to regain trust from open-source coders during the early days of Azure. "To just focus on a walled-garden kind of an ecosystem and ad revenue as the main source of income, they probably will never become the big player," he said.
A Meta spokesperson pointed to Wang's recent comments about the company's continued support for the open-source ecosystem, and said Meta still plans to offer outside developers access to Muse Spark's underlying technology via an API, as it previously announced. "We're already testing with some early partners, and look forward to releasing it this month," the spokesperson said.
"That Zuckerberg's metaverse and virtual reality ambitions have generated over $80 billion in total losses since late 2020 makes the AI pitch a tougher sell," the article points out, citing this observation from Howard Yu, business professor at Switzerland's International Institute for Management Development.
"He's running out of the space for his credibility to last," Yu said. "I think the virtual reality foray may have burned up a lot of his goodwill in front of investors."
Re: (Score:2)
No (Score:4, Insightful)
None of the large LLM "investments" will pay off. They tried token-based pricing, at significantly below what they have to charge to only break even, and it was far too expensive for the value it provides. And there is absolutely no reason to expect cost to come down anytime soon. Optimizing AI implementations has been going on for 70 years. They would need a fundamentally new idea, but these cannot be forced.
Re:No (Score:4, Interesting)
Time is on their side, if they survive for long enough.
Hardware gets better. Models still get smaller. Inference gets optimized. Overall the cost per token (measure it how you want, dollar, electricity, compute, they are all proportional) is going down. The cost the usual customer who wants to ask the AI for a cake recipe is willing to pay stays the same and also the cost the high-paying customers (science, etc.) are willing to pay stays the same when the model quality keeps increasing at the current pace.
So currently they are not competing for dollars per token, but for building a customer base, which will pay off the investment later when it becomes easier to amortize model training. That's a bet that can fail for some companies, but can be subsided for others (like GAFAM). The big question is what happens with OpenAI and Anthropic. OpenAI will be bought by Microsoft before they fail (Microsoft already announced such things). For Anthropic it isn't that clear, but on the other hand they are currently in a good position not to fail. I would also not bet against OpenAI, if they don't do something stupid.
The third thing is API integration. You already names the per-token pricing schemes. While the "Once per week a cake recipe" customer paying $20 is fine, the website paying $20,000 per week in API tokens is more important to keep as customer. The website owner on the other hand is best kept by providing them a free or cheap plan for asking for cake recipes so they don't test the competition's model.
Re: (Score:3)
Time is on their side, if they survive for long enough.
"Long enough" will be a lot longer than they have left. And I do not even think time will help. The severe problems with LLMs become harder to ignore every day. The stunts intended to signal might and power become more and more feeble while the language gets more bombastic. And the burning of mountains of cash continues.
Re: (Score:2)
Unprofitable and no real prospects to get profitable. "Move on" indeed.
The best reason for this demented hype I have heard so far is that apparently a lot of CEOs are deep in "AI psychosis" now. They try some simplistic thing (because CEOs cannot do actual work in most cases), find it works, find they "coded" something, fail to understand the severe limitations of what they made and then think they can fire actual engineers.
No means yes... (Score:2)
they don't have to make a profit on the investment. The investment is defensive. The goal is not to be outside the moat.
Re: (Score:2)
The problem, is that they need to keep digging the moat, and it will never end.
The billions of dollars being spent on GPUs today, is billions thrown down the hole because they'll have to replace all of them in the next 3 years with new GPUs at the same pricing or higher. And electricity sure as shit isn't getting cheaper, and new equipment will be using more electricity, not less.
There is no economy of scale where this begins to look profitable in the current model. Economies of scale only work when input
Re:No [Or I hope not?] (Score:2)
I think you're answer is too broad and fuzzy to be "insightful", even though I hope it is accurate for the monstrosity that Meta nee Facebook has become.
However from the broader perspective, I am less sanguine. I think the economics are different if you have enough renewable energy. As in China. Then there may be almost no cost for the tokens but basically just the sunk cost of the hardware. The key question for profitability will become the value of "intelligence" and there is a key advantage in the abilit
Re: (Score:2)
You need to take into account that the hardware does not keep forever. For this type of hardware, the errors probably start to become a problem after 5 years or so, because it is being run fast and hot. Hence hardware is not a sunk cost, but an ongoing cost.
Re: (Score:2)
Yes, but... The hardware costs could be regarded as marginal if the value of the "intelligence" "produced" is high enough. There's also the fact that the reprogramming for improvements can include increasing the robustness against deteriorating hardware.
If I was a betting man, I'd still be more inclined to bet on the Chinese (or Indian) approach over the YOB's. Which makes me wonder if there are any other players to consider. Any chance the Germans might blindside everyone (again)? And is it too early to co
Re: (Score:2)
Could, may, etc. does not make for useful predictions. And while resilience to hardware failure sounds nice, the things degraded by ageing are the higher-power elements, like com line drivers. These give you full-chip loss and they are relatively homogenous in effect, i.e. they will take out almost all your hardware in a relatively short time.
On the minus-side, LLM results look less and less valuable every day as actual research (not just gut-feelings and hype) starts to roll in.
Re: (Score:2)
Not very persuasive but you have me speculating if you were one of those overclockers. I suspect the main philosophical difference is that you think hardware is more important than software and I'm in the other camp. We seem to be in an abnormal condition because the computer hardware has been changing so rapidly for a few decades now...
Anyway, at the rate things are changing these months, I am actually worried that I may live long enough to see how it all comes out in the wash.
Never held accountable (Score:5, Insightful)
The fact Zuckerberg is still at the head of this company is an indictment against our economy, our culture and the entire problem people have with the wealthy: once you are wealthy you can fuckup seemingly an infinite number of times and not suffer any consequences.
First the general social issues with Facebook with it's corrosive algorithms. That should have sunk him. Then sinking how many billions into the metaverse, so much so that you renamed the company? Could any of us make that scale of mistake and not only keep our job but even make more money?
And now this, another failed billion dollar loss and even if he is removed from the CEO role he'll get a multi million exit package, keep all the stock, remain on the board and probably be able to swing more venture capital to new companies.
I don't support what Luigi Mangione did by any stretch but I also find it incredulous when people act perplexed about people who do support his action and like, when you see shit like this...
Re:Never held accountable (Score:5, Interesting)
When someone has created a successful product, they usually think it's because they're smarter than other people, and they're usually wrong.
Facebook is the only really successful product Zuckerberg has ever created. It succeeded because he was in the right place at the right time, and that doesn't happen very often. All of Meta's other major products are things they bought instead of building themselves. His attempts to build other things from scratch have mostly failed.
He decided "the metaverse" was the future of computing, just as everyone else was embracing AI as the future of computing. If he weren't the founder and single largest shareholder, that would have gotten him fired.
Ironically, Quest is actually the most popular platform for VR gaming. If he'd been content just to create a gaming platform, it would be considered a success. Instead he blew $80 billion trying to turn it into the future of computing, making it a massive failure.
Re: (Score:2)
+5 Insightful
Everyone's gobsmacked when they first try VR and many swept away with ideas about a future where everyone's there. But fuckerberg had the losing combo of self loathing for missing the mobile boat, a Ceasar complex and gobs of ill-gotten wealth so of course he believed himself the visionary who would bring it to the world.
lol no
Re: (Score:2)
He decided "the metaverse" was the future of computing, just as everyone else was embracing AI as the future of computing.
He's trying to do both things. AFAICT the AI stuff is not much worse than anyone else's. As far as the metaverse, it's clear that he didn't read anything anyone ever wrote about it. It might be the future of computing, someday, but it's the far far future now as long as we have crap efforts like his to go by.
It's obvious that few corporations are really going to want to operate something like the metaverse with enough freedom for it to work as well as it's going to. Second Life proves that it's possible, bu
Re: (Score:2)
Re: Never held accountable (Score:2)
No one can fire him. He's the dictator of Facebook. He has controlling interest.
Re: (Score:2)
Yes, that's the reality right now, I wasn't arguing that, I was arguing the perception and perverse incentives it creates.
Also maybe we should make it so that when a company reaches a certain size there can't be a single controlling interest. Meta employs 70k people, a large amount of stock value, they control a very influential media platform. Their actions have effects on way more people than Zuckerberg.
Re: (Score:1)
Re: (Score:1)
You're a knuckle dragging troglodyte with nothing to offer and deserve nothing but contempt.
The fact the only argument you can make is envy is so empty headed I'm surprised you're conscious enough to operate a keyboard. Take your nonsense elsewhere.
Re: (Score:2)
Nice random overt misogyny that you volunteered without anyone asking for it.
Gives us good insight into who YOU are.
Re: Never held accountable (Score:2)
So, who should control it if not the people who own it?
Re: (Score:2)
Oh, there can still be moves to make him change his mind, even if he holds 50% +1 share:
The banks can tell him to get fucked because they're not pissing good money away on a shitty idea that's going nowhere after $80B of investment.
Institutional investors can tell him to stop dumping cash on this bonfire, or they liquidate their holdings and shitcan his personal wealth in a day.
There's still plenty of leverage on this guy, if the people with their hands on the biggest lever of them all (money) decide to use
Re: (Score:2)
The fact Zuckerberg is still at the head of this company is an indictment against our economy, our culture and the entire problem people have with the wealthy: once you are wealthy you can fuckup seemingly an infinite number of times and not suffer any consequences.
On what basis do you suggest shareholders remove him as head of the company? It's easy to point to fuckups, but the hard pill to swallow is they just reported another record net profit of $25bn for the quarter keeping their upwards trend. Their share price is high. Their revenue is down slightly from their record Q4 2025, but they made a bigger profit on that slightly reduced revenue.
So if you were the board, or a shareholder, why would you shoot the golden goose who has clearly run a company that has healt
Re: (Score:2)
I mean on that standpoint if an employee at Meta fucks something up on the scale of $80B they won't get fired if the stock and revenues are doing well? I'm not even necessarily justifying his firing, he probably can't even be fired with how much he controls. I'm just speaking to the perceptions it creates. This guy can fuck up and no reasonable change to his life will happen. When people act incredulous because "why hate billionaire" that's a big part of it I think.
Like you said, even just losing his job
Re: (Score:2)
I mean on that standpoint if an employee at Meta fucks something up on the scale of $80B they won't get fired if the stock and revenues are doing well?
An employee of Meta is that, an employee of Meta. Their responsibilities extend to the requirements of their direct boss and are not subject purely stock and revenues. A CEO is an employee of the board and shareholders. Their responsibilities extend to creating returns and thus their performance is wholly defined by stock and revenues.
You're right though he can't be fired in general, but the point is: why would anyone fire him? He is demonstrating great performance.
To revert back to your example with a pers
Re: (Score:2)
The line goes up even farther if they don't spend $80B on Second Life v2 with the exact same results as Second Life v1.
That's the point.
AI investments may not be meant to "pay off" (Score:5, Insightful)
We have already seen how normal "consumers" have become irrelevant as customers, we have seen how "retail brokerage" customers have become irrelevant as "investors", and the next stage has already begun, where the world economy is shaped to address the needs of AI/robots, not the needs of puny humans.
Re: (Score:2)
Helping you write emails and collect cooking recipes with AI is the excuse, not the objective.
Everyone knows Meta = Facebook (Score:2)
Meta doesn't really know how to do anything else with any skill. At best, they're an afterthought in all these other areas... and mostly they don't even reach that status.
Even when they buy out other companies, the only acquisitions that "work" are Facebook-adjacent ones like Instagram.
Re: (Score:2)
> Meta doesn't really know how to do anything else with any skill.
They don't know how to do Facebook very well either: it's been pretty much stagnant and enshittified to death for the past 22 years, and it feels like a forum for greying people whose greying friends haven't bothered to move on either, or to get the date of the next annual meeting of the bridge club.
Re: (Score:2)
... and it feels like a forum for greying people whose greying friends haven't bothered to move on either, or to get the date of the next annual meeting of the bridge club.
That's not entirely fair. I know quite a few young people who... maintain their Facebook accounts so they can talk to their grandparents.
Betteridge's law of headlines (Score:2)
No.
https://en.wikipedia.org/wiki/... [wikipedia.org]
Yeah it will (Score:2)
And it's something more than regular Capital like we are used to seeing. It's Capital that is explicitly designed to eliminate jobs. It's an entirely new automation technique. Something that we haven't seen since the assembly line was invented.
And yeah if you go back to the industrial revolution it's going to look pretty fucking primitive compared to a modern factory. But t
Re: (Score:2)
AI is only indirectly the capital. The customers are the real capital. AI is the service you provide to them.
Re: (Score:2)
But the difference between the ruling elite that are pushing for this massive automation and you is that they think in terms of decades and generationally and you think about whether or not you can pay rent next week.
Citation needed. They don't seem to be able to see the past or the present, let alone the future.
There's no reason why we won't ignore the next big push
Are you in marketing? You're not actually familiar with how LLMs have already reached a dead-end that can't be fixed by adding more of the same.
Re: (Score:2)
I'm not so sure. All AI and robots will do for billionaires is make labour cheaper. But they already have basically unlimited money so I don't feel that is the root motivation.
I think when you get to that level you either think of yourself as better than everyone, or secretly don't and try to compensate. They want to be the one who brought phones or the metaverse or AI or robots to the world because of course (in their mind) only a great person could do that. And of course control, being able to change the
Zuck on Meta's AI Mess and its New 'North Star' (Score:2)
Zuckerberg on Meta's North Star: "The Most Talented People in the World" Matter [slashdot.org]
Has Meta ever done anything that paid off? (Score:5, Insightful)
I don't mean this as an easy dunk, I really want to know if anyone can think of any counterpoints. I mean they have to exist right?
It really seems to me that Meta, and by extension Zuck, are the worst tech company and CEO of all time, propped up by a money machine they arguably fell into by accident.
But think about it, their only successes since Facebook itself, are just having cash first and buying their competitors.
Otherwise:
- their mobile efforts were basically a failure, but they bought Instagram
- their "pivot to video" was a money losing joke
- within that, remember when they wanted to be Twitch? That was a thing for a while.
- their messaging never really succeeded despite basically starting with the market cornered, but they bought WhatsApp
- where's Farmville today? they were supposed to become an app platform. That was a thing.
- they were going to be the world's ISP for a while, seriously, does anyone even remember that? Starlink broke them.
- and of course, The Metaverse is arguably the single biggest failure any big tech company has ever done
So seriously, am I missing any? Like Google is similar, but they birthed Chrome, gmail, Android, there are big wins among their failures. Meta . . . only misses.
Even if there are AI winners, it's just really difficult to imagine Zuck ultimately doing anything other than taking billions from his ad business and setting them on fire. But maybe the next $100 billion they burn will pay off, it kind of feels like they have to succeed eventually.
Re: (Score:2)
Facebook ads pay well.
Re: (Score:2)
You're right. Fuckerberg is a loser who was in the right place at the right time and was willing to bend over for advertisers. The rest is just trying to buy success.
Re: (Score:2)
Yep, they're basically SoftBank 2.0. They got incredibly lucky tripping into a market at the right time with an acceptable product and made a fuckton of money that they continue to lose on shitty investments to this day.
Well duh! (Score:2)
The company jumped into AI with its Llama family of models
Of course they'd have issues. Everyone knows llamas can be pains in the asses. Should have gone with alpacas.
No, nothing ever works out (Score:2)
Facebook Home / HTC First "Facebook Phone" — 2013 — discontinued within weeks;
HTC First price cut to $0.99.
Internet.org / Free Basics — 2014 — banned in India 2016; largely abandoned.
Oculus / Reality Labs (VR/AR/metaverse) — Oculus acquired 2014 ($2B); rebrand to Meta 2021 — cumulative operating losses $83.5 billion over 21 quarters as of Q1 2026; budget cut ~30% Dec 2025; 1,000+ jobs cut Jan 2026.
Horizon Worlds — 2021 — missed internal user targets; part of R
Re: (Score:1)
My last 99 investments were flops ... so this one has *got* the be the big one.
Given their record... (Score:3)
No (Score:1)
Easy come, easy go! (Score:2)
They'll make more money, but they're never going to win the AI race and make that back.
It Already Did (Score:2)
The part people miss is the assumption that "AI" = "LLM"
The reality is that "AI" = "Machine Learning"
What Meta (Facebook and their other properties) are using machine learning for? Ad delivery. They're working on more powerful and faster models for inference, not for text, but to service more targeted advertising based on more data on the individual.
This isn't just a theory or hyperbole, previous generations of Facebook "AI" accelerators have found their way onto the used market, ones used for this exact pu
No it won't (Score:2)
But Meta will waste money on AI, and not provide any useful products and then buy out any companies that do provide AI and threaten Zucks moat, and the bought out government regulators will let them.
More likely than it's bet on Metaverse (Score:2)
AI is something that people actually use, as opposed to a small subset of people who are so hooked on facebook games that they want to *live* inside them.
So charging for AI is the only $$ it makes..... (Score:2)
Who They Talking To? (Score:2)
I wonder if all these tech-bros (tech-brahs?) spend a large amount of time getting their egos stroked by AI chatbots.
Don't want to hear any gainsays to your wisdom and business savvy, just chat with your AI pal.
Let us hope not (Score:2)
No. (Score:2)
None of the "AI" companies have any path to profitability, other than a stock market full of suckers.
token engineering? (Score:2)
"said Rob May, CEO of the startup Neurometric, which works in the realm of token engineering"
Goddamn I gotta get better at bullshitting my way into billions of dollars. "Backupify" was a company that guy worked on and exited, probably with all the money. seriously.
We're on the map (Score:2)
"Meta is at least back on the map in AI"
In this map, Anthropic is the Pacific Ocean, and Meta is Lake Eerie.