Thanks for the info. I was aware of Google's TPUs but I wasn't aware they were so widespread. As for the others, well apparently my understanding is out-of-date.

In any case, the actual layout and timing-studies for a new chip strike me as being far too involved for an AI company that wants to concentrate on other things, like curating data and training models with it. Surely they must have outside help -- lots of it.

I suspect you underestimate what it takes to design a chip. AI companies can't afford to focus any part of their business on that effort, because it's way too involved.

If AI companies use bespoke designs for chips, they're not designed and taped-out in-house. They would hire another company to do it for them. At which point it's pretty much the same as buying off-the-shelf, because many chip-makers are catering to AI clients already.

Nvidia has risen dramatically over the past decade or so because it found a crossover market for its video chips, specifically in high-performanced computing, cryptocurrency, and more recently AI-training. I think its stock-price will keep rising as long as it keeps innovating or it finds new markets, and its customers don't find alternatives. However, it has suffered a few pullbacks over the years, so it's not a buy-and-forget stock.

I think Tony Isaac's point is a company that is already very big cannot get a lot bigger without getting smaller first.

Intel is a good example of that. Its stock-price had a big rise up to mid-2000 until the dot-com bubble burst. Then it fell by a factor of two. For about a decade and a half it did pretty much nothing, then rose to challenge its 2000 high starting in early 2020. Then it fell again, reaching early-2010s levels last year. And then suddenly it had a spike upwards in the past year or so, fed by new management, the AI surge, and partnerships.

In all of these cases, Intel's stock price rose after lows, some of them significant. It didn't just keep rising, because it couldn't. There's only so much capital in their sector of the marketplace. An already-large company would need a whole lot more of that capital to grow, percentage-wise, than a smaller company.

You are right that individual investors are not limited by market-cap. However, individual investors need to consider how much more a company's stock-price can grow if it has already grown a great deal. It may need to sell off before it can rise again.

There are plenty of robots that have significant manual dexterity, and can even do sculpture. See here, for example.

Zen Buddhism believes that all things have a Buddhist-nature, i.e., they are "alive" at least in some sense. So, Buddhism is quite compatible with Animism, though not the same. Buddhism is a spiritual and mental discipline that encourages adherents to strive for personal liberation, aka "Nirvana." Animism just holds that all objects contain "spirits" that can be exploited in one way or another.

As for people who treat LLMs as though they're "conscious" -- well, I'll give them a pass. I just asked Gemini what it would tell someone who thought it was conscious. Among many other things, it came up with this gem:

Zen Buddhism claims that all things, including inanimate objects, possess Buddah-nature, meaning they are fundamentally "alive" in some sense. So, at least it's consistent for a Buddhist temple to initiate a robot.

Now, the fact that the robot was fed the lines -- even the voice -- and that it was on loan, does tend to detract from any message the temple was trying to convey.

Here's a direct link to the NYT article.

Yes, there are lots of ways to eavesdrop on a conversation, if you're in the spy business. But you can't rely on any single method. You need to use a spectrum of them, with the hope that a few, or even one, will yield intelligence.

And let's not forget that a great deal of the spy business entails making human connections, not employing James Bond gadgetry.

Finally, you mention govenment employees using unsecured channels. Yes, we know that happens (not naming names) but it's yet another method you can't rely on, but can exploit if it occurs.

For the record, I'm not in the spy business, never have been, and do not want to be.

Hardly. From TFS:

As I said, they're peers with ETrade and Schwab.

And as for them deserving a story, they're abandoning an "agile" approach in favor of larger teams focused on specific projects. It will be interesting to see whether they succeed with the new development model. So, the story fits the Slashdot ethos of "News for Nerds. Stuff That Matters."

Fidelity Investments. They're a financial-services company that offers trading, banking, financial planning, education, and other stuff. Similar to ETrade, Schwab, etc.

I saw no such admission in the GP's post. You, on the other hand, are confused. When Trump tore up the JCPOA, he did the "best thing" from Iran's perspective to liberate them to work towards uranium enrichment. And here we are now, where Trump wants a deal again, much like the one he tore up.

We are all "coping" -- you included -- with the consequences of the current administration's strategy, in the form of surging energy prices, a lack of goods that run through the Strait of Hormuz, a cost of $1B per day for US military activities, inflationary pressures, and global tensions.

It is Trump who has weakened the US position, by tearing up the JCPOA. You telling yourself that "Trump was right all along" sounds like your cope.

The JCPOA was created in 2015 and took effect at the beginning of 2016. It limited Iran's enrichment of uranium to a maximum of 3.67%.

Trump tore up the JCPOA in 2018, during his first term.

Consequently, Iran was no longer constrained by the JCPOA in 2021.

Too bad Trump tore it up then. He could have just let it run for its 10-year lifespan and renegotiated a renewal now. Instead, we have the current shitshow.

Weapons-grade uranium is 90%. The JCPOA limited Iran's enrichment to 3.67%.

Where are you getting this 60% figure?

This. The parent AC is peddling false equivalence.

Why do you keep posing questions with unsubstantiated premises?

I daresay nobody here wants to see a nuclear-armed Iran. We differ on how to prevent that from happening.