The statement here that Nvidia invests in OpenAI is a bit misleading. Nvidia would pay out nothing to OpenAI if OpenAI turns out to be too poor to pay for capacity. So they are not that exposed to the death of OpenAI specifically. They would be more at risk of making too many GPUs to prepare for the deals.
Oracle takes a lot more risk, but in case OpenAI fails to grow quickly, it can still probably find buyers for its capacity in the next 5 years. There are many rich firms that will continue to invest in AI whether or not AI makes money.
> Nvidia would pay out nothing to OpenAI if OpenAI turns out to be too poor to pay for capacity. So they are not that exposed to the death of OpenAI specifically.
Nvidia has invested billions in previous rounds of OpenAI raises also. Pretty sure it is not nothing.
Also OpenAI rents from CoreWeave that Nvidia has invested in.
> "circular" risk more refers to the recent 100B "investment", and that is quite misleading
No it's not. It refers to a web of companies sending money back and forth. Nvidia investing in OpenAI, OpenAI investing Coreweave and that goes back to Nvidia except recently 10x the scale. Amazon, Broadcom, Intel and many more are now all in on it.
The point is that risks of the circularity cannot be established well if one leg isn't quite dangerous. Nvidia's direct exposure is much more limited than it seems. Additionally in your example, Coreweave is tiny. It's mainly companies like OpenAI that are highly leveraged.
This is a poor analogy. Magnus Carlsen stays because chess consumers decide to pay for humans even though they are inferior to Stockfish. BigCorp will always pick machine over you if they can.
I agree, that was a weak analogy. Magnus stays employed because chess fans value watching humans compete, not because engines didn't replace his capabilities.
I've updated the post title to "Train with coding assistants like Magnus Carlsen trains with chess engines" to focus on the main point: the methodology. Magnus uses chess engines as sparring-partners to improve his game after matches. Same can be done by developers who will use coding assistants to level up their skills.
Yes, but it’s also a poor analogy in the sense that AI hasn’t replaced one of the best of the best but has easily surpassed the playing capabilities of every fair, average, and above average player out there.
I am not worried about AI replacing the best of the best any time soon, I’m worried about it replacing the fair to middling…relatively soon.
Companies automate the parts that are commodity. On messy product work (drifting specs, integration, liability), human + AI + good process > AI alone. The machine proposes; the human sets goals, constrains risk, writes/reads tests. That combo ships faster and with fewer costly mistakes than letting an ai free-run.
Productivity replaces people, if you get more done from a team of 5 than your old team of 10 you generally fire 5 people.
Programmers have significantly higher unemployment than the general workforce today, but it’s hitting a wide swath of white collar jobs and that’s not going away. The industrial revolution replaced manual labor, so people moved to more mentally challenging jobs but AI can eventually replace anybody from CEO’s on down.
Demand isn’t doubling economy wide any time soon. So you might keep a team of 10 if you’re outcompeting a different team of 10 and getting all of them fired.
Even if you’re keeping your job expect huge downward pressure on wages.
Computer ratings are kind of random because there's no meaningful sample of them playing top players. They're obviously much stronger, but specific numbers are kind of meaningless beyond comparing them to other software which they have competed against.
But Stockfish is not going to beat Magnus with queen odds. Nakamura has beaten Lc0 with knight odds in blitz (though he got crushed overall), and queen odds in bullet, all while chatting on stream. And fast time controls favor computers simply because they're practically impossible to flag and will almost never miss a tactical idea.
So you would enjoy playing Magnus more than playing Stockfish. Because it would actually give you a chance. And that's the whole point of playing games: to enjoy them, and to both have a chance at winning, even if that chances is remote. Stockfish takes that joy away. For me, that makes Magnus far superior to Stockfish. If I want to lose a game playing with myself I'll play solitaire. Or better yet, I'll go read a book. But the chance to interact with a human at the peak of their skill would be unbeatable for me in terms of enjoyment, far to be preferred over playing a game with a computer program. Though I can see the satisfaction in programming a computer to be that good, I would not enjoy the game because the other side would not enjoy it either.
>So you would enjoy playing Magnus more than playing Stockfish.
Oh, no, I wouldn't enjoy playing him more. I'd enjoy playing Stockfish infinitely more because I could learn new strategies from it. Knowing you're going to lose allows you to learn from all moves and all mistakes.
Against Magnus, I'd have no chance. I'm barely over a 2000 rating. Magnus could play with queen odds and drunk off his ass while blindfolded and would wipe the floor with me.
German was the lingua franca of physics at the time, so to speak.
Starting in the 1930s, though, that tradition began to change... for reasons that I'm sure won't ever apply to American English. Nosirree, Bob, we're special. Great, even.
It's a bit of an aside but I believe this is one reason Zuckerberg's vision for establishing the superintelligence lab is misguided. Including VCs, too many people get distracted by rock stars in this gold rush.
Just last week I said something inline with that[0]. Many people conflated my claim that Meta has a lot of good people with "Meta /is/ winning the AI race". I just claimed they had some of who I think are some of the best researchers in the field, but do not give them nearly the same resources or capacity to further their research that they give to these "rock stars". Tbh, the same is true for any top lab, I just think this happens more at Meta because Meta is so metric and rock star focused.
So I agree. The vision is misguided. I think they'd have done better had they taken that same money and just thrown it at the people they already have but who are working in different research areas. Everyone is trying to win my doing the same things. That's not a smart strategy. You got all that money, you gotta take risks. It's all the money dumped into research that got us to this point in the first place.
It's good to shift funds around and focus on what is working now, but you also have to have a pipeline of people working on what will work tomorrow, next year, 5 years, and 10 years. The people are there that can do that work. The people are there that want to do the work. The only thing is there's little to no people that want to fund that work. Unfortunately it takes time to bake a cake.
Quite frankly, these companies also have more than enough money to do both. They have enough money to throw cash hand over fist at every wild and crazy idea. But they get caught in the hype, which is no different than an over focus on the attribution rather than the process or pipeline that got us the science in the first place.
Maybe not that surprising, in large part the Germans built up the Chinese auto industry, and Tesla went to China to tap into that manufacturing knowledge/talent and backported it to their US factory.
Could be part of it, but the US just doesn't have cheap cars anymore. The days of the Geo Metro and the Dodge Neon with a 5 speed and crank windows is over. Car companies have decided to relegate people (in the USA) with either low income, or who cant stomach the type of depreciation every car suffers from, to the used market.
The reality is that there's no margin in cheap cars. You need to look at numbers instead of vibes.
The difference between 4 crank-up window regulators and 4 power window regulators is less than $100. 4 power lock actuators cost less than $20. Switches for all the above are, what, maybe $10?
The same math applies to power mirrors, auto climate control, heated seats, cruise control, and all the rest.
The production cost of a car with "power everything" vs. "manual everything" is a few hundred dollars at most. But consumers expect a much bigger discount for the inconvenience of missing those features (or, conversely, are willing to pay a much larger premium to add those features to a baseline car).
That the US doesn't have cheap cars is simply the reality of what the market demands. Cheap (new) cars don't sell (for more than what it costs to produce them).
I wasn't implying that throwing manual window regulators on a 2025 model car would be a significant cost reduction. It was just 2 examples that I could think of to back up my point that there are fewer affordable cars than there used to be.
My (admittedly very, very limited) personal experience owning cars actually suggests cars are getting cheaper over the past couple decades. Specifically, my data looks like this:
- A new Honda Accord LX in 2003 was ~$19k
- A new Honda Accord LX in 2020 was ~$23k
In today's dollars, that's roughly $33k and $29k, respectively. These numbers are very approximate, but it means the same car model in 2020 was about 12% less expensive than the one in 2003. And the new version has a whole lot of improvements and features the old one didn't. (They cheaped out and removed the lock from the glove compartment though!)
Stepping back and thinking about the complexities that go into manufacturing a modern automobile, it's wild to me that they can cost so little compared to what you get. It's a machine that can travel 200+ thousand miles and last for decades with barely any maintenance.
Commercial-scale vehicles (semi trucks, busses) cost an order of magnitude more than personal vehicles, yet share many of the same complexities. Like, how are cars so cheap for what they are? Manufacturing volume, I guess.
A friend of mine bought a used car in 2007 for ~$4500 in 2025 dollars.
In my mid-20s I bought a 3-year-old Accord for $16k (using a $12k loan) and that was a big stretch for my finances at the time, despite having a good early-career tech job.
Your $17k figure is a lot of money for most folks in the US.
Then the question becomes why would they help a company that competes with their portfolio companies. And even stronger case that they got some shares in return.
One reason is that helping contributors that you know will survive (as proton was already well known) grows the market. Basically give the competitor of piece of the cake because they will enlarge the cake for everyone.
1% of the world is over 800m people. You don't know if the net impact will be an improvement.
reply