Jensen Huang to college grads: "Run. Don't walk" toward AI

https://www.axios.com/2026/05/...

Nvidia founder and CEO Jensen Huang told graduates at Carnegie Mellon University in Pittsburgh yesterday that demand for AI infrastructure is creating a "once-in-a-generation opportunity to reindustrialize America and restore the nation's capacity to build."

Why it matters: With many college grads fearing AI could obliterate their career dreams, Huang pointed to boundless opportunity as a "new industry is being born. A new era of science and discovery is beginning ... I cannot imagine a more exciting time to begin your life's work."

Nvidia, which makes AI chips, is the world's most valuable company. Huang told 5,800 recipients of undergraduate and graduate degrees that the AI buildout will require plumbers, electricians, ironworkers, and builders for chip factories, data centers and advanced manufacturing facilities.

"No generation has entered the world with more powerful tools â" or greater opportunities â" than you," he said. "We are all standing at the same starting line. This is your moment to help shape what comes next. So run. Don't walk."

"Every major technological revolution in history created fear alongside opportunity," Huang added. "When society engages technology openly, responsibly, and optimistically, we expand human potential far more than we diminish it."

Full speech: https://www.youtube.com/watch?...

...like "Tell me about Tiananmen Square" or "Tell me about Xinjiang".

Is this what you want for the future?

My thoughts back when R1 came out:

https://www.afcea.org/signal-m...

...an article worth considering from Princeton University's Zeynep Tufekci:

We Were Badly Misled About the Event That Changed Our Lives

Since scientists began playing around with dangerous pathogens in laboratories, the world has experienced four or five pandemics, depending on how you count. One of them, the 1977 Russian flu, was almost certainly sparked by a research mishap. Some Western scientists quickly suspected the odd virus had resided in a lab freezer for a couple of decades, but they kept mostly quiet for fear of ruffling feathers.

Yet in 2020, when people started speculating that a laboratory accident might have been the spark that started the Covid-19 pandemic, they were treated like kooks and cranks. Many public health officials and prominent scientists dismissed the idea as a conspiracy theory, insisting that the virus had emerged from animals in a seafood market in Wuhan, China. And when a nonprofit called EcoHealth Alliance lost a grant because it was planning to conduct risky research into bat viruses with the Wuhan Institute of Virology â" research that, if conducted with lax safety standards, could have resulted in a dangerous pathogen leaking out into the world â" no fewer than 77 Nobel laureates and 31 scientific societies lined up to defend the organization.

So the Wuhan research was totally safe, and the pandemic was definitely caused by natural transmission â" it certainly seemed like consensus.

We have since learned, however, that to promote the appearance of consensus, some officials and scientists hid or understated crucial facts, misled at least one reporter, orchestrated campaigns of supposedly independent voices and even compared notes about how to hide their communications in order to keep the public from hearing the whole story. And as for that Wuhan laboratoryâ(TM)s research, the details that have since emerged show that safety precautions might have been terrifyingly lax.

Full article

And the likeliest explanation is things connected with the GDPR "right to be forgotten":

https://mashable.com/article/c...

The cooling inlet for a nuclear plant is barely big enough for a scuba diver to enter. It could probably be smaller, but they make it that big so a scuba diver can regularly go in there to clear it of fouling by biological organisms (mussels, seaweed, etc). What they're proposing here is on an entirely different scale. The marine industry spends billions on anti-fouling paint, and it still requires regular scraping (every few weeks to months) and repainting (every few years). It's a massive endeavor. I suspect the best solution is not going to be propeller-like turbines, but something more like a flapping tail. It will be less efficient but will continue to function even if heavily encrusted, allowing you to stretch out maintenance intervals to where they're economically feasible (it costs probably one or two orders of magnitude more to send a diver down to clean things than to send a maintenance worker to a wind turbine).

Diesels are already more efficient than EVs in most of the world (where electricity is predominantly generated from fossil fuels). Diesel cars can already hit 40% efficiency, trucks 50%, and ships 60% efficiency. Electricity generated from fossil fuels is about 40% efficient (coal) to 60% (gas). Call it 50%. Power lines have about 5% transmission loss, battery charging losses are about 15%, battery discharging losses about 15%, and electric motor efficiency about 90%. For an overall EV efficiency of (50%)*(95%)*(85%)*(85%)*(90%) = 31%.

That's actually pretty close to the efficiency of the gas engines used in cars. Years ago I backed out the efficiency of a Nissan Versa using the Nissan Leaf as an energy consumption equivalent (since they share the same chassis and thus aerodynamics), and it came out right around 30%. Those new 6-speed, 8-speed, 10-speed transmissions and CVTs help a lot. It's why I've been saying for over a decade now that the priority needs to be converting our power generation to renewables and nuclear, and shutting down fossil fuel power plants. If you don't do that first, switching from a gas vehicle to an EV may make you feel better, but doesn't really reduce carbon emissions. (The energy cost advantage of an EV isn't because of efficiency; it's because coal and natural gas are an order of magnitude cheaper per Joule than gasoline and diesel. MPGe does not take generation, transmission, nor charging efficiency into account. It can't because those will all be different depending on your local power plant's efficiency, your distance from the power plant, and type of charger you use. So you can't really compare MPGe to MPG.)

The USB-C spec includes something called alternate mode. That's where the four wires carrying high-speed data (mirrored so 8 wires total) plus 4 lower speed data wires in a USB-C cable can be repurposed to carry signals other than USB. Common alternate modes include HDMI, Displayport, and Thunderbolt (PCIe signals).

Lightning only uses 4 wires for data (not mirrored). It would be completely trivial for Apple to create a Lightning alternate mode, and have all iOS devices switch to USB-C connectors. They could've done it a decade ago when the USB-C spec was first being laid out (Apple is a member of the USB-IF, and in fact it holds one of the seven board member seats). The only thing preventing Lighting over USB-C is their own obstinance and greed (selling overpriced Lightning cables).

Our brains seem to function as a neural network. Their drawback is that the longer they spend repeatedly exposed to the same patterns, the harder time they have adjusting to changes or new scenarios. Stereotypes, discrimination, the crotchety old man/woman who won't change their ways. Death is nature's way of resetting those neural networks, so they're flexible and adaptable again.

In a world without aging, I suspect there will be a lot more wars (and deaths due to war, counteracting most of the increase in lifespan) due to fewer people changing their minds or compromising. If you thought today's politically polarized society was bad, this is going to make it a lot worse.

Heh. Back around 1992 I helped run a mail server that hosted a mailing list. One day I got a frantic call from a co-sysadmin that the system was down and he wasn't able to login remotely. Since I had local access, I began poking around the system to figure out what the problem was. Our server had a 512 MB HDD. Someone had scanned a photo he'd taken, converted it to a BMP instead JPEG so it was about 12 MB, and emailed it to everyone on our mailing list (about 150 people). The first few copies got sent OK. But as our network connection to back up (took about a minute to send 12 MB over DSL back then), the mail server began queuing copies of the mail for later delivery to other recipients - writing the 12 MB BMP to disk over and over. Until our 512 MB HDD was full and the server ground to a halt.

Anyway, if you have some large file(s) which you wish to email, the proper way to do it is to host it on a website or Dropbox or Google Photos or YouTube (for videos) or something similar, then email people the URL. That way they can grab the file(s) by opening the URL in their borwser, instead of it clogging up the Internet, mail servers, and people's 15 GB free Gmail or Hotmail storage limit (Yahoo's is 1 TB so less of an issue there).

Every nuclear plant operator pays a portion of their receipts into a decommissioning fund. The fund is enough that although San Onofre Nuclear Generating Station was closed early (28 years instead of the expected 50), its decommissioning fund ($4.1 billion) is sufficient to handle the estimated decommissioning costs ($4.4 billion). (The fund amount is a current snapshot, while the decommissioning costs are spread over decades. So the current fund will accumulate interest over the decades, and will be more than sufficient to pay for costs).

If you add up the highest estimates I could find for the costs of the Fukushima and Chernobyl cleanups ($660 billion and $700 billion respectively), and divide it by the total sum of nuclear power generated (about 2300 TWh * 30 years + 2300 TWh/2 * 20 years), their cleanup only end up costing $1.46 trillion / 92000 TWh = about 1.5 cents per kWh. If you use average cleanup cost estimates, it's closer to 0.5 cents per kWh. If you ask me, I'm more than willing to pay that for nearly unlimited, on-demand, greenhouse gas-free electricity. Since the average U.S. household uses 10700 kWh/yr, it amounts to just $160 per household per year worst-case ($53.5 if you go with the average). By contrast, the $25 billion in forgiven student loan debt thus far has cost us ($25 billion / 144 million taxpayers) = $174 per taxpayer. The numbers are even more favorable for nuclear in other countries, where household electricity use is lower, and price is higher.

The only problem nuclear power has is that it's an incredible concentrated power source, so you don't need a lot of nuclear plants to supply a lot of power. This means when there's a single accident or disaster, its scope is correspondingly much bigger. But to properly assess the drawbacks, you need to compare the costs against the amount of power generated. Otherwise you get into the same situation as with car and plane crashes. Because the fatality figures are much higher when a plane crashes, people who don't think twice about getting into a car have an irrational fear of flying. Even though planes are much safer than cars.

This also affects the viability of insuring nuclear plants. Insurance requires large numbers to average out probabilities - as you increase the size of the insured population, the width of the bell curve shrinks. Insuring 3000+ gas and coal plants is easy - because there are so many of them their premium ends up close to the average cost of an accident. By contrast, insuring 55 nuclear plants is a lot harder, and their premium ends up being closer to the worst-case cost of an accident rather than the average cost. Normally in situations like this the government steps in an act as insurer, because it's for the long-term greater good. But unfortunately our policies regarding this are largely driven by people who evaluate it emotionally, rather than rationally and mathematically. This is largely what's driving the push to small modular reactors: not waste reduction, but as a workaround for the math which makes it more expensive to insure a few really large plants, compared to breaking up those plants into lots of smaller ones with the exact same risk.

Same problem as the U.S. The ISP owns the lines and provides the service, and local government regulations prohibit anyone else from laying down lines. Meaning you effectively have only one choice of ISP.

We need to change this to follow the gas company model. You probably only want one set of lines, so one company can own the lines. But it's prohibited from providing service. Instead, it leases the lines (under government oversight since it's a utility) to lots of companies which provide service. That breaks the vertical integration between line-owner and service-provider. The part of the service which should only have one company (laying down and maintaining the lines) has just one company. The part of the service which should have multiple companies providing service (ISPs) have multiple companies competing.

Vertical integration is desirable early on. When cable TV companies first began, nobody knew the best way to lay out wires to each home. Some sort of hub and spoke model was probably better than one central station with individual lines to each home. But how many hubs should there be? What's the optimal average length of the spokes? What sort of technology do you use to split and combine the signals at the hubs? These sorts of questions are better answered by competition with vertical integration. But today, all this has been sorted out. Competition has arrived at the best near-optimal solution. And in fact all the cable companies use the same hardware (following the DOCSIS standard) for cable modems. Once things reach that point, there's no longer any need for such vertical integration, and the economy is better served by eliminating vertical integration and splitting up ownership of the wires from providing the service.

Contrast this with cellular service space in the U.S. Unlike cable Internet, cellular technology is still developing, and tighter vertical integration is desirable when the technology is in its development stage. e.g. When 3G service rolled out, two of the carriers tried GSM, two tried CDMA. And in the end CDMA turned out to be superior (for those of you who still think GSM won, the GSM standard was updated to incorporate CDMA). Likewise when 4G service rolled out, Sprint tried WiMax (which used SOFDMA), while the others used LTE (which used fixed OFDMA). And in the end LTE turned out to be better. You want vertical integration when these sorts of things are being decided, so a mis-step at one end (GSM phone companies eschewing CDMA) does not necessarily sink the company since they have other sales providing them time and money to redirect (GSM phone companies survived from revenue for providing phone service, until they could implement UMTS based on wideband CDMA a year later). Likewise, although the WiMax debacle hurt Sprint (they basically had to pay to build a 4G network twice) to the point where it eventually had to be bought out by T-Mobile, it did not immediately kill the company since they still had significant revenue from service.

So there are development stages where vertical integration is desirable, and stages where it's no longer desirable. It's this transition we're getting caught on. Uncertainty over when we should transition, and people not understanding that this transition exists so the old regulatory ways may no longer be optimal.

Not that I condone the behavior. But Amazon took so long to begin cracking down on it that a lot of companies which wanted to be honest, may have begun "brushing" just to stay competitive. So except for a handful of big companies who already had brand recognition, all the honest Amazon sellers have either been driven out of business, or have begun "brushing" themselves just to stay in business. You need to nip this sort of stuff in the bud.

The spam calls in the U.S. (at least the ones I get) mostly aren't targeted. The spammers just randomly or systematically dial numbers, so something like the GDPR (preventing the sale of your phone number) wouldn't help.

The problem is the phone companies refuse to implement a system where the call recipient can verify the caller's phone number or ID. The caller ID system is laughably easy to spoof. Blocking the number doesn't help since it's not the number the spammer is actually calling from; and it may even wind up hurting you as you may end up blocking a legitimate number that you wish to receive calls from. The rationale for callers being able to change their caller ID on their own is that each phone line here has its own phone number. If a company has one main number but uses multiple lines for outgoing calls, it will probably want to assign the caller ID on all of them to show their main number. I wouldn't have a problem if that's all they were able to do - change the caller ID value to one of their other numbers. But there is nothing to prevent them from setting the caller ID number to anything they wish. It's even possible to get a spam call here, and the caller ID shows your own phone number.

This seems like it'd be an easy problem to solve with public/private keys. Assign each phone number a private key known only to the line owner, and add the corresponding public key to a global database. When someone calls you, they encode some changing value (e.g. the time of the call) with your public key and their private key, then send you their public key. Your phone sees their public key, decodes the message using that and your private key, to confirm the time of the call, thus confirming the caller is the person who owns the private key corresponding to that public key. That makes it impossible to spoof someone else's number (you'd need their private key to do that).

But the phone companies have zero interest in actually fixing this problem because they get so much money from the spammers. In fact several of them play both sides for profit. They sell service to the spammers, then sell spam-blocking upgrades to everyone else. I don't think it'll be fixed until we do the equivalent of the IPv4 => IPv6 transition but for phone numbers, and we all know how well the phase-out of IPv4 is going.

Apple CEO Tim Cook had his home blurred from mapping apps after issues with a stalker. [...] The case for blurring? "Having strangers from all over the world stare at your home isn't necessarily something you want to happen -- but it can be done in seconds on the mapping apps we all carry around on our phones." ("Stop people from peering at your place," suggests the article's subtitle.)"

Hate to break it to Mr. Cook. But if his home is visible to the public and a stalker knows his address, said stalker can just hire someone to take pictures of the house and email them back. And the stalker can then stare at those pictures as much as he wants, no matter how much Google et al blur the online pics.

They're trying to spin this as a right to privacy issue when it's not. If you don't want your house to be visible to the public, build a bigger wall in front of your house. Don't mess with new technologies which have revolutionized travel. If I'm asking a bunch of people to get together at a location many of them have never been to, I can send them a street view link so they'll know what it looks like when they get there. If I'm driving a long truck or a trailer and want to make sure I can get into and out of a location, I can check online without having to waste time and fuel going there to find out. Your right to privacy does not override the public's right to see things that are visible from public locations.

Only it's bullshit. What it is is tax evasion

That was my conclusion too. A Ponzi scheme is zero-sum - the money gained by the fraudster equals the money lost by the victims. Currencies are only zero-sum if they do not enable transactions which previously were not able to happen. That's why money is so great. Before money, if you needed milk and had eggs to trade, you could only get milk if you found someone who needed eggs and had milk to trade, or were able to put together a multi-person trade (e.g. your milk for someone's chickens for someone's flour for someone's eggs). If you couldn't find an appropriate trade, you couldn't get your milk or trade your eggs, even though lots of other people might be trading milk or seeking eggs. Money enabled lots of transactions to occur which previously wouldn't happen in a barter economy.

Cryptocurrencies (in their current form) do enable new transactions, so they're not zero-sum and thus not a Ponzi scheme. Unfortunately, the new transactions they enable are overwhelmingly illegal ones. Things that society deemed as harmful overall (narcotics, tax evasion, blackmail, human trafficking, etc), so have banned, and enforce that ban by making it harder to conduct financial transactions around them. Cryptocurrencies bypass all those financial bans. So yeah money is flooding into them because they enable all these illegal activities. But their net effect on the economy is negative. The value lost by the victims of these activities exceeds the value gained by the perpetrator. Like how a burglar who steals a $500 TV costs you more than $500 because of incidental damage and you having to invest more time into shopping for and setting up a new TV.

In that respect, they're worse than a Ponzi scheme.