Kids have security blankets. Tech CEOs have security compute clusters.
Apple did something similar with NAND storage for the iPad mini. They took a bet that could have been wrong. It was not wrong. Competitors had a hard time because of it.
It's not binary where you either have compute or not. You definitely do need GPUs, but there's already masses of compute, I believe it doubles every ten months or so just from Nvidia's chips. Many factors make it a very irrational decision
1) Companies were spending hundreds of billions collectively on AI capex. Meta alone was 75 billion projected this year. This is an extraordinary bet, given that the most revenue any AI company makes is a few billion by OpenAI.
2) When DS came out, it was a huge validation of the moatless idea. These SOTA companies have no moat, at best they are spending tens of billions to maintain a few months edge.
3) DS was also a huge validation of the compute saturation idea - that SOTA models were always massively efficient. At best it was traded for iteration speed.
4) Many other more technical arguments - Jevons paradox, data exhaustion (synthetic data can only be generated for a fixed set of things), apparent diminishing returns (performance relative to compute, the denominator has been exponential but the numerator logarithmic)
So on one hand you have these SOTA models which are becoming free. On the other hand you have this terrible business model. I strongly suspect that AI will go the way of Meta's Metaverse - a staggering cash burn with no realistic path to profitability.
It's one thing to invest in a new technology with tangible benefits to your product. It's another to spend vastly, vastly more into vague promises of AGI. To put it into perspective, Meta will spend on AI capex in a few months of 2025 as much as Apple spent on NAND in total. What advantage is there to be had with SOTA models? You do 20% better on some AIME/IQ/competitive coding benchmark, which still translates atrociously to real world issues.
But Nvidia will be very successful because these companies frankly have lost a lot of the plot and are FOMOing like mad. I still have memories of the 2013 AI gold rush where every tech company was grabbing anything with AI in them, which is how Google got DeepMind. They are being enormously rewarded by it by the stock market with Meta's price 6x since it's lows.
I think your whole argument is based on this being true, but you didn't give much argument about why there is no ROI. 400M USD isn't hard to generate...even a moderate ad engagement lift on X would generate ROI and that's just 1 customer.
Imagine going back in time and showing every VC how great the search business will be in 20-30 years. The only rational response would be to make giant bets on 20 different Googles...and I think that's what's happening. These all seem like rational investments to me.
I think a similar thing is playing out with AI. In 5-10 more years these LLMs will replace a google search today (and maybe be even better).
[1]: https://sherwood.news/tech/the-trillion-dollar-mystery-surro...
xAI also announced a few days ago they are starting an internal video game studio. How long before AI companies take over Hollywood and Disney? The value available to be captured is massive.
The cluster they’ve built is impressive compared to the competition, and grok 3 barely scratches what it’s capable of.
Microsoft is in process of building optical links between existing datacenters to create meta-clusters, and I'd expect that others like Amazon and Meta may be doing the same.
Of course for Musk this is an irrational ego-driven pursuit, so he can throw as much money at it as he has available, but trying to sell AI when you're paying 10x the competition for FLOPs seems problematic, even you you are capable of building a competitive product.
https://centreforaileadership.org/resources/deepseeks_narrat...
If you’re using your compute capacity at 1.25% efficiency, you are not going to win because your iteration time is just going to be too long to stay competitive.
xAI bought hardware off the open market. Their compute edge could dissappear in a month if Google or Amazon wanted to raise their compute by a whole xAI
Also, what isn't clear is how RL-based reasoning model training compute requirements compares to earlier models. OpenAI have announced that GPT 4.5 will be their last non-reasoning model, so it seems we're definitely at a transition point now.
Ha ha. I'm sure their play to claim airdrop idle game will be groundbreaking.
What you're seeing right now is pure flex and a signal for the future and competition. A much maligned AI team that hasn't even been around for very long at all just matched or topped the competition without making use of the latest training techniques yet. The message this is intended to send is that xAI is a serious player in the space.
This is a great example of how a misleading narrative can take hold and dominate discussion even when it's fundamentally incorrect.
SemiAnalysis documents that DeepSeek has spent well over $500M on GPUs alone, with total infrastructure costs around $2.5B when including operating costs[0].
The more-interesting question is probably why do people keep repeating this? Why do they want it to be true so badly?
[0]: https://semianalysis.com/2025/01/31/deepseek-debates/#:~:tex...
Deep Seek R1 is literally an open weight model. It has <40bln active parameters. We know that for a fact. That size of model is definitely roughly optimally trained over the time period and server times claimed. In fact, the 70bln parameter Llama 3 model used almost exactly the same compute as the DeepSeek V3/R1 claims (which makes sense, as you would expect a bit less efficiency for the H800 and for the complex DeepSeek MoE architecture).
It appears that LLM chat interfaces will replace Google SERPs as the arbiters of truth. Getting people to use your LLM allows you to push your world view. Pushing his "unique" world view appears to be the most important thing to modern Musk.
In that light, paying 40B for Twitter, and billions for Grok training makes perfect sense.
The beauty of a failed investment is that it never goes below zero. So upside is the only thing they care about. Why invest in a near-zero chance for a random SAAS to take off, when you can invest in a near-zero chance of creating superhuman artificial life?
Yes but why? This is what I really don't understand.
Say AGI is achieved within a reasonable timeframe. Odds are that no single company will achieve that, there will be no monopoly. If that's the case, where is the trillion dollars value for investors? From every claim we hear about it, AGI will lead to hundreds of millions of jobs disappearing (all white-collar jobs), and tens of millions of companies disappearing (all the companies that provide human-produced services). Who is going to buy your AGI-made products or services when nobody is paid anymore, when other companies, big and small, has ceased to exist? Sure, you can make extraordinary accomplishments and advance humanity far, far ahead, but who is going to pay for that? Even states won't be able to pay if their taxable population (individuals and corporations) disappear.
So where will the money come from? How does it work?
> due to the additional reasoning latency.
They're also less creative for non-STEM topics
In any case, Elon won't win this race cause the best talent will not work for him. He used to have good reputation and a lot of money, which is a deadly combination. Now he only has the latter -- not enough when leading AI people can make 7 figures in other companies.
To be clear 1: I'm not saying that people who currently work on Grok are not great. It's not about hiring some great people. It's about competing in the long run - people with other options (e.g. offers from leading AI labs) are more likely to accept those offers than joining his research lab.
To be clear 2: I'm not talking about Elon's reputation due to his politics. I'm only talking about his reputation as an employer.
He has the vision and marketing skills but it's not going to be enough for leading the AI race.
I think the situations are a bit comparable given timelines however.
A perfect analogy for AI … your ability to replace talent with money. And if you don’t have the talent, it’s gonna cost you 100x more.
That sure seems to be the message given in Apple AI commercials. From those commercials the tag line for AI should be "enabling idiots everywhere".
Any source? I’m a heavy user of Claude and pay for the Teams plan just for myself so I won’t get throttled. Love it. But I’ve been impressed with O1 Pro lately. That said, I don’t like paying both €166 for Claude Teams and €238 for OpenAI Pro. :)
Per court filings by the administration, Musk is not in charge of DOGE, nor does he have any role in DOGE, nor any decision-making function in government at all, he is a White House advisor unconnected to DOGE.
DOGE uses only X links, and I am sure Grok will be the next gov contract. After all he has all the data on everybody down to your IRS tax returns.
