Nvidia is lagging its competitors with respect to stock price appreciation. However, it's still outperforming its competitors in revenue, margin, and future demand. While Nvidia's growth has indeed slowed down, it still is growing faster than Intel or AMD, either looking at percentage or dollar growth. Intel's sales are flat. While AMD's sales are growing by about 40% over the last year, Nvidia's sales have grown by 65%, despite being much larger to start with. AMD's PE ratio is now an astounding 151! Intel's PE is undefined since its profit is negative.

How about for questionable slop? Most of the AI stuff is perhaps detectable by highly trained individuals, but most readers are not highly trained. Also, many readers eagerly slurp up human-created slop of obvious low quality and accuracy, so why would there be a rejection of AI-created slop of obvious low quality and accuracy?

AI slop already works for revenue generation. That's why it's gaining steam. I'm not advocating for AI slop, but it's obvious that it's working, maybe not for the readership public but definitely for the generators.

This announcement of an "agreement" sounds like PR. Is there a contract for N wafers? Is Apple paying for poor yield, or is Intel absorbing low-yield cost? Is this a real pre-contract agreement or just some raw meat to throw at Trump?

Apparently this huge 245TB SSD will cost around $80k, which will be around 10x the cost compared to the same capacity HDD set. The HDD power will be 2-3x more, but that power is noise compared to the rack power needed by everything else. There is an advantage in form factor density compared to a set of smaller SSDs, but that density comes at the cost of performance due to a single interface, single queue, and more latency due to TRIM. Performance is a big deal because that's why the much higher cost of SSDs is justifiable. The only real reason for this huge SSD is to save rack space, and that is something, but is it worth the decreased performance?

Seems like a dangerous attitude. This barging in on the Scientologists is very much like the teenagers knocking on doors and then running away. Would it be acceptable to distinguish between the doors of neighbors that we like and don't like? It's an outrage when the teenagers bother the nice old widow but something we condone when it's the unfriendly curmudgeon. That doesn't sound right.

Doing work that doesn't result in additional comprehension or knowledge is not only non-beneficial, but because it's emotionally defeating, it can be worse than not doing any homework because all the student learns is that they don't get it and likely never will.

The big problem with education is that many/most teachers aren't good at tutoring/teaching. That is, they say stuff in class that doesn't help the students learn new concepts. Or with many teachers, they don't even both with the lecturing. They simply assign stuff and hand out tests. At home, the same pattern repeats. What would be great would be either work that by design is self-tutoring. Or maybe some sort of AI-based instruction that forces the student to ask questions about what they don't understand.

It's the exact opposite. Sure, the MAGA base can ignore gas prices, but the non-MAGA republicans and the right-leaning independents are much more easily swayed by gas prices, and it's this third of voters that produces political tsunamis. Carville continues to be right - it's the economy, stupid. That continues to be true even if a third of voters are intransigent MAGA.

It seems like bookstore.org is doing more beneficial things for small bookstores than might be apparent. They allow small bookstores to benefit from online orders without incurring the significant costs of buying or setting up the ordering, inventory, or fulfillment systems. The small bookstores do send some traffic to bookstore.org via both direct orders and referrals from their own websites. The bookstore gets 30% of the cover price for orders and referrals and 10% from the general fund otherwise.

Although the small bookstores are also taking advantage of other trends (like selling coffee and stationary, hosting events, curating booklists, etc.), bookstore.org plays a significant role in the revival of the small bookstores. The 30% profit from referring to bookstore.org is far greater than the small 1-3% profit from in-store sales. The average small bookstore gets around $5-10k per year, which is significant considering the small budgets that most small bookstores operate on.

NAEP scores have been dropping since 1970 and SAT from before 1970, i.e., before Carter.

What is more important than cost is power usage or total hardware needed or memory needed or something other than a price. Cost is just a number selected by salespeople. The Chinese have been accused of price dumping in many markets, so a lower price is not surprising or necessarily indicative of a technological advantage. If the power usage is indeed lower, then that is significant. However, if power were indeed significantly lower, then that would have been the headline.

The problem with cameras and screens is that they take time to get used to. Focal lengths and clarity are different. My car has a switchable camera/LCD and mirror rear-view mirror. I always use the non-camera mirror because it's easier to recognize objects in both daylight and nighttime due to object clarity and the focal length issue. Distinguishing object distances in the mirror at a quick glance is far easier than with the LCD screen.

As far as side mirrors go, the huge problem with side mirrors is that they don't save power. The slight decrease in drag is completely balanced by the LCD power. This was the conclusion from one of the university long-range research EVs from a few years ago.

That depends on one's viewpoint. For executives with large stock bonuses, growth is the only thing that matters. For employees, cash flow leading to steady employment should be the most important, but since some executives make layoff decisions based on their stock bonuses, employees also need growth for those companies. For consumers, steady production with reasonable prices. For investors, it depends on whether one targets growth or income/dividends.

This is the real problem. Data center use electricity and water, create minimal jobs for the area and resources used, and generate a small amount of taxes. The property taxes for commercial (especially in California) are minimal, there are no sales taxes, and income-related taxes are also low due to the relatively low number of workers.

You'd think that the extreme electricity and water usage would incur a correspondingly extreme cost, but in a mind-baffling way, the extreme resource usage costs far less per unit than households pay. Instead, households bear the brunt of the required electricity and water resource infrastructure and procurement.

The way to even out the pros and cons is to mandate a economic return to the area. Impose a tax based on the output of the data center. Call it either a sales tax or a value added tax. Of course, this would scare off the data centers, which is just fine because there is no lost economic value to the residents of that area.

To put numbers to this issue but not necessarily settling the question: "And the authority has ample capacity. At our eventual highest operating level, we could potentially draw 500,000 gallons of water per day. The water utility currently has an excess supply of 8 million to 10 million gallons per day. So the utility actually has way more capacity than it can actually sell to their customers right now.”

... and "about 7% of U.S. employees are eligible" which means that 7% of Microsoft employees have age + Microsoft years that add up to 70+. That's something I would have expected at the old IBM.