> Those who realize how much you can accelerate your research with the help of AI will succeed.
Yes, but, in the last week we saw an AI lab front-run[1] the research of mathematicians doing what you suggest. The lab threw something like $15M of compute at a problem and the researchers were able to spend nowhere near that. I think the authors are more concerned about that kind of asymmetry and race to publish the results.
[1] - I am not going to debate whether that was deliberate on the part of the lab or if it crept into training data, etc. I don't know and don't think it matters towards the point of the authors here.
This has always happened, way before AI. You'd spend months or years building and growing your business, and then Google would release a feature or product that would kill your business overnight because they can throw way more money at the problem, plus their branding. That's life.
But you're talking about business, where competition has always been expected. OP is talking about mathematical research, which has stood on hundreds of years of cultural tradition driven by human individuals sharing ideas, collaborating, building on each others' work, all for the benefit of humanity.
Okay... Pop Tarts are also Kellogg's, not a tobacco company. But let me pick some other examples.
Eggo waffles predate RJ Reynolds getting into this by 10 years.
Twinkies were introduced in the 1930s.
I'm not defending ultra-processed foods nor tobacco companies. I'm just saying the authors have a major blind spot if they think the concept or popularity of ultra-processed foods started in the 1960s.
The Twinkie of 1940 is very different from the Twinkie of 2000. Lots of old and sort of OK things got corrupted when pulled into the corporate-Zerg-slime
I don't think companies expanding their business should be the only reason to sell. Apple was a desktop and laptop computer company when they released the iPod. Nokia was a company making rubber products when they started making cellular equipment in the 1970s. Amazon was a ecomm company when they released AWS. Etc.
Sure, I'm cherrypicking the success stories but I don't think it's the only signal people should use.
$8.7B in losses over the last 5 quarters sounds bad, but that is also with $20.4B capital expenditure in the period and double digit percent increases per quarter in user metrics.
By traditional business metrics it looks toxic but compared to a typical tech unicorn it's not really outstanding.
Considering it's a tech unicorn with strategic and financial leverage like no other (their own global Internet infrastructure, access to arbitrary billions in capital, ability to build the biggest chip fab in the world), it's certainly fair to call it a gamble, but to call it failing is a stretch imo.
The project was first teased by Musk in early 2026 and officially announced on March 21, 2026 during a special event at the defunct Seaholm Power Plant in Austin, Texas [1][2]
I wouldn't really say SpaceX expanded its business. It was merged with a social network and AI firm to juice its IPO valuation. The SpaceX part of the business is interesting, but the value is generously only about 20% of the value of the company. Anyone who wants to invest in AI isn't going to want a near last place model company. And who wants to put money into Twitter?
Sure, but that's more of an indictment of SpaceXAI. All those resources, and they get upstaged by a bunch of much more compute constrained Chinese startups.
I disagree. They're renting out GPUs to other providers because they realize it's a long game that they're in a better position to win because they control infrastructure as well as hardware.
The iPod was a peripheral to use alongside your MacBook, it was meant to drive Apple computer sales initially and it's not like it was Apple's first attempt at a portable device (Newton). Nokia was building telecom equipment (stemming from their telephone cable experience, the insulation being a product of their original rubber business) and military equipment along with a bunch of other consumer electronics before they got into mobile phones. Amazon released AWS because that was their infrastructure they had built for themselves to handle their massive e-commerce business.
Companies generally don't just jump into a totally new market unless they've already got some experience.
The SpaceX acquisitions of xAI and X have absolutely nothing to do with space. While I'm sure SpaceX software engineers benefit greatly from cheap access to Grok, it and X have no relevance to the core business of SpaceX. It's the same level as "huh?" if Boeing decided to buy up an AI company and social media platform. The only reason xAI and X are even part of SpaceX is because Musk fucked himself with being forced to buy Twitter and he decided to bundle them with an actually profitable company.
Spacex has no existing customer to satisfy their future launch ambitions with Starship, hence data centers in space, just like they did with starlink to provide the necessary demand for Falcon 9. The launch business is not profitable without a regular cadence, too many fixed costs - ground infrastructure and personnel.
X can supply training data to to Xai and Grok is heavily integrated into X so there is some synergy there.
Starlink would be a major customer. NASA is already eyeing Starship to get things to the moon. It's just such an enormous expense to develop that no one but SpaceX is willing to fund it directly.
SpaceX had adult leadership for a long time. I suspect there was a tacit agreement with Musk about a bit of separation.
He broke through that professional firewall when he merged xAI with SpaceX and then pushed it to go public as he did. I also posit that Gwynne Shotwell is weirdly undervalued for what she's accomplished and what she's put up with, and I think that will show up some time in the future.
There was enough technical analysis to show that even Musk's cult feed can't overcome the significant financial hardships he laid on SpaceX. I personally think it was his jump the shark move. He can't really go up again after this. He no longer has any untainted assets for mega-exploits.
>'Gwynne Shotwell is weirdly undervalued for what she's accomplished and what she's put up with, and I think that will show up some time in the future.'
Gwynne is fiercely loyal to Elon. I don't expect to see significant strife between them.
There are a small single digit number of 'new space' company president opportunities in the world and one of them requires dealing with Musk. Makes sense to be loyal when there's so few other equal opportunities. No doubt many companies would love to have her at the helm but very few offer the ability to build reusable rockets.
The thing is, everyone's interpretation of "environmental impact" is different. To one person it can mean, "don't put the construction garbage in the river." To another it can mean "not one single Delta Smelt can be scared because of this construction."
And because it's so flexible, in states like California where we have aggressive environmental laws, it's leveraged as the NIMBY trump card. When it can't block a project, the process is used to make it inordinately expensive and take decades. One example would be the environmental studies for the CA High Speed Rail.[1]
The deregulation stuff isn’t about nimby. It’s making nimby 10x worse by making it hyper local. That means poor people who are poorly organized get boned. State regulations tended to help with that.
I live in upstate NY, the rebuild of the GOP here is around hyper local issues, mostly apartments and solar. MAGA changed the discourse and allows the rabble rousers to say the quiet part out loud. (Ie bike infrastructure and apartments will bring poor black people to rape and pillage)
Do you have any specific examples of how new state regulations actually eased the regulatory burden for building something? Adding new regulations at the state level almost never removes the hyper local restrictions, it just adds a new layer of compliance on top.
How can the solution to burdensome regulations be MORE regulation?
A law banning certain types of regulations at a lower level I would not call “state regulation” I would call that deregulation.
I suppose I agree with you that the solution to burdensome regulation is to write new laws which forcibly deregulate lower level things. But calling laws like that “state regulation” is true only in the most technical sense. Your example is a textbook example of deregulation.
The problem most ignore is that you will either have public regulation from the government or private regulation from companies.
Would you rather have a published regulation that is subject to democratic and judicial review and applies equally to every one. Or a company terms of service that forces binding arbitration and your only recourse is to not use the service?
I lived in a neighborhood with an hoa and it was the worst form of regulation I've ever dealt with. I own a health care provider and the regulation when it was direct medicaid was infinitely better than after it was privatizatized with managed care.
To be fair, part of the inordinate expense is just because it takes longer for the environmental reviews (costs are in expected year of construction, so pushing a project a decade into the future can increase costs by 30-40+% (inflation + interest) depending on the specifics, even if everything else costs the same).
That's why the cost estimates for CA HSR jump a bunch every time an administration hostile to it enters the white house.
> project a decade into the future can increase costs
A very good point.
I don't agree we can blame Trump for HSR though. 2/3 of the time that has passed have had Democrats in the white house. HSR is nearly all pure-California-style self-inflicted wound. And honestly it's just the most visible project California has failed with, there are many others. The one I'm personally angry about is Prop 1. We're now 12 years after, and have no additional water resources even broken ground. It's shameful.
It’s an fully intra-state train line in a state that has an economy bigger than France. California should be able to build an entire EU-style HSR network with zero regard to what’s happening federally.
CA can and did recently, but that's money that will need to come from and/or could have gone to other programs/etc.
I think the most recent cut was $4+ billion in federal funds for HSR. It's going to come from cap and trade instead, but the state could have had both.
Agriculture uses 90% of the water in California growing cotton, alfalfa and almonds, which are all very water intensive, Humans watering their lawn, drinking water and bathing, use a fraction.
Note: The way they divide up water usage. They have a third category, listed As environment, but that’s misleading because the people who have the water rights can always use a lot more water at their discretion at any time. where as the common citizen cannot.
We can in part. Like you said, it took a ton of time because of environmental review, and that's on the state. With that said, construction didn't begin until 2015, and Trump pulled billions in funding in his first term in 2016 and did the same in 2024.
infrastructure building is a bad word in the United States of America currently, Unfortunately, our major competition in the world is building infrastructure like crazy, but there is still hope, Because once that high-speed rail system starts building west to east or even east to west, all those areas in the middle of the country will change their tune. They will all be fighting on getting a piece of the rock, or I should say a piece of the rail.
The only mistake California made was not including one of the adjacent states as a destination, be it to Las Vegas, Phoenix, or Portland, all of a sudden business interests and those crazy Republicans would be on board because they’re concerned about missing out on making money, which is basically God in America.
I may not live to see it. but once they start going interstate with high-speed rail anywhere within the United States, the tune will change. It’s just amazing that so many people are short-sighted about it. More of that short-term thinking humans are famous for.
Note: Those who think ahead long-term obviously have already bought land on both sides of the route of that high-speed rail line in California, and probably along the proposed route leading to Las Vegas in the future, and the same applies to any possible line to Phoenix or to the Arizona high plateau.
No cope to it the Republicans in California are powerless, but when there was an opportunity at the last minute to lobby on putting the line on the eastern side of the Central Valley (which probably added three or four years to the process), guess what they cried like babies, to put it on the eastern side of the valley even if it meant more delays.
Originally, the line was scheduled to run on the western side because that side was less populated. In other words, you had pretty much a clear lane as far as the right-aways were concerned, and this scenario will play out once high-speed rail runs out of California to the east.
All those conservative red areas will all of sudden change their tune, and there is no cope again money is God in those red areas that is the one thing those red areas understand particularly those at the top that and taking away the vote, firing a gun or blaming a worker that is less fortunate.
This video about Shoreham was my entry into the Proper People's channel. If you like this kind of abandoned structure exploration, I recommend checking out their other videos.
> Hard disagree. When you've had to chase through a change in untold and actually unknown numbers of duplications of code in different permutations and fix them because they are all on fire simultaneously, you'd disagree too.
The other end of this spectrum is dealing with the architecture astronaut's up-front abstraction. Totally overengineered for solving the initial requirements, but then constantly needing new hacks to make it cope with new requirements as they come up in the normal course of work.
That's why there's a balance in there, it's somewhere between "always duplicate code even when you know a lot about the problem" and "always write abstractions even when you know very little about the problem."
> So where is all the productivity going? Where is the value?
Infrastructure doesn't produce value overnight. How long did it take the Interstate System to provide measurable value? I asked Gemini. Supposedly increased national productivity by 25% over 39 years[1]. But if you drove on a newly finished interstate in 1959, you saw the same cars just moving a lot faster.
That's what we're seeing right now. People can produce an incredible amount of stuff really quickly with AI. Is it directly connected to measurable productivity across the entire economy? No, because, realizing a mass productivity increase from infrastructure takes time.
Yes, but, in the last week we saw an AI lab front-run[1] the research of mathematicians doing what you suggest. The lab threw something like $15M of compute at a problem and the researchers were able to spend nowhere near that. I think the authors are more concerned about that kind of asymmetry and race to publish the results.
[1] - I am not going to debate whether that was deliberate on the part of the lab or if it crept into training data, etc. I don't know and don't think it matters towards the point of the authors here.
reply