I think it's important to understand the distribution of revenue and profit for the competing companies.

Some are like OpenAI with some revenue but huge losses, mostly from capital expenditures for computer equipment rather than operating expenses. These companies project dramatically increasing revenue but still significant losses for several years. These companies will complete for at least several more years.

Some are like the smaller companies that are struggling. They spend lots of money with minimal revenue. These companies will go away soon.

Then there are the hyperscalars. These companies spend huge for capital expenditures but they also bring in the revenue. They may not break even on AI, but their overall businesses still enjoy huge profits. These companies have no danger of going away anytime soon.

This is usually what happens in almost every emerging field and market. After many years, the competitors will be winnowed down to a handful or less, as lesser companies go bankrupt, quit, or are bought out. This is normal and to be expected.

As Carville said, "It's the economy, stupid!" It's as true as ever now. Yes, other factors are at play, including AI, but it's mainly the economy.

The big question in our faces is why the US economy is uniquely suffering. We need to turn our eyes toward Washington to answer that question. The tax increases via tariffs combined with the negative stimulus via direct government job cuts and government funding cuts. Plus negativity towards tourism in the US and buying American products.

It's one thing for Washington to deny science and medicine, but our leaders are going for the trifecta in denying economics. None of this is a surprise, and none of it goes away by cancelling critics and pounding the propaganda pulpit.

My original point was that some Indian IT folks will join the Indian contracting companies in order to have a chance to immigrate to the US. If that chance falls from 2% per year to 0%, then some of those workers won't bother to join the contracting companies, and as a result, the overall quality of IT workers at those companies will decrease. They will still join other Indian companies but less so the companies that were sending workers to the US via the H1B visa.

The underlying story is about Chinese technological superiority in some fields. However, the direct takeaway is not that this superiority cannot be overcome but rather that venture capitalists are having a hard time finding the profits to even fund such ventures. Of course, if this is really true, then the US could attack the problem in the very same way that China attacked it in the first place. Don't try to use capitalism but rather insert government funding/subsidies, forced production and purchases, and other mandates, laws, and leverage to artificially push technological progress and shield those efforts from the ravages of capitalism.

Quite a bit of the technological progress from the US has come from government research and projects, so the US could do something if it choose to do so. Unfortunately, the current administration not only lacks the vision and political will but is undermining technological progress in the name of partisan expediency.

Back in the late 90's, Lucent was "loaning" X dollars to big telecoms to buy X dollars worth of equipment. This giving away of equipment had the intended effect of propping up the Lucent stock price. Well, until the ruse was discovered, and the Lucent stock almost became worthless.

What's the difference between the Lucent scandal and what Nvidia is doing? Well, it depends on the accounting. Is Nvidia getting ownership shares of OpenAI? Does Nvidia get to book sales from all these GPUs going to OpenAI?

Uber guy saying 10-15 years has the same confidence interval as Elon saying imminent. Of course, they don't need to state their confidence interval because they're not engineers or technologists so much as salespeople.

More specifically, people in power want to minimize their costs. This can be done via several means, such as automation, laws, or slavery.

That's the wrong denominator. There are 3-4 million IT workers in the Indian contracting companies, and not all of them are the technical workers that would be considered for an H1B. 60,000 out of 3-4 million is a pretty good chance. That's like a 1 out 50 chance of getting to immigrate to the US with your family, and you get that chance every year. After 10 years, that's around a 20% chance of being selected, and if you're politically savvy in the company, the chances are much higher.

The key in the survey question is how much money one is wiling to give up. 1 or 2 percent, sure we'd all be willing to give up an insignificant portion of our compensation. 50%? Well, I imagine very few would be willing to give up that much.

These quixotic thoughts are just that. Most of those who are willing to say that they would accept a pay cut or quit or move out of an area not only wouldn't but wouldn't actually seriously consider it. Just like those surveys a few years ago saying that the majority of Californians were thinking of moving out of state, but in reality only a very small percentage actually did. They are willing to voice their desires, but in the end, money is a far greater motivation.

If this Trump directive goes through (and it's not clear that this isn't yet another TACO move), there are several significant impacts due India.

First, Indians in the US send around $35 billion in remittances to India. That's enough to be a few percentage points of the entire Indian GDP.

Second, the possibility of being sent to the US on an H1B is an enticement for Indian workers in the Indian contracting companies. If work gets send to India as a replacement for importing H1B workers, does that impact the quality of workers that the Indian contracting companies can attract? We're talking about 60,000 Indians sent to the US every year. That's not a small chance of being able to immigrate to the US.

It may depend on the type of position. Some Indians graduated from an IIT and then from a top US school with a PhD. They get paid way more than $100k/year. They do possess harder to find skills, and those are the types of immigrants that the H1B should be targeting. A $100k increase in costs for these workers is substantial but far less than 50%.

However, many Indians and non-Indians did not graduate from top schools in their country or the US. Many of these people are hired for their low cost. A $100k increase in their cost would likely be more than 50% of the current costs.

The key to setting an effective fee is to price it to distinguish between these two types.

Both of these ideas are problematic.

First, technological innovations have already come to some crops like most of the grains grown in the Midwest. The reason most fruit and vegetable harvesting is not automated is that the technology doesn't yet exist. Maybe in a few years or decades, we'll have machines or robots that can pick fruits and vegetables, but likely not anytime soon.

Second, it's not just a matter of money. The work is hard, literally back breaking, and often unpleasant. A wage that is just a living wage may not be sufficient to attract enough workers. Remember that many young people don't want the factory jobs that Trump is trying to bring back to the US, and farm work is even harder than factory work.

Furthermore, the price inflation from paying much higher wages would make that produce pretty much impossible to sell.

In our world, there are currently only two practical solutions. Either use undocumented workers or import produce. Trump is attacking both solutions at the same time.

This is partially true and false. Yes, some/many of the so-called ghost cities have filled up and become vibrant cities. However, there are others that have some people but have nowhere near the planned number of residents.

China's real estate situation is problematic. It is not good. Several large construction companies have gone bankrupt. The problem is that Chinese and Asians in general prioritize real estate as an investment vehicle. Just like the stock market, when the market values go up, prices have upward momentum and the government, companies, and people are happy. However, when the values go down, then the bubble pops, and downward momentum is hard to reverse.

One extremely challenging problem is the story of unfinished homes. In China, people buy homes and start paying mortgages before the unit is finished. Since several of the largest construction companies in China went bankrupt, it's likely that many of these homes will never be finished. Some estimates put the number such "ghost" homes at 20 million. Some of these homes were investment properties that require mortgage payments but will never yield any return. This is a problem with no easy solution.

Ignoring for the moment that this research came from a Taiwanese university and not from China, it doesn't make sense that Americans think that the Chinese are stupid. The recent extreme US measures regarding tariffs, product bans, visa and immigrant scrutiny, and concerns over international military and geopolitical relations only happen because the US considers China to be formidable. Doing all this for a stupid adversary would be stupid.

I'm not sure what this means. NVLink is baked into each GPU, and most customers buy racks rather than single GPUs. These racks have NVLink switches included. So, NVLink has essentially sold just as well as GPUs.