Three revolutionary products — you know the line. A wide-screen iPod with touch controls, a revolutionary mobile phone, and a breakthrough Internet communications device. That was how Steve Jobs introduced the iPhone.

I’ve linked to this snippet before, usually to note how the audience didn’t really understand what an “Internet communications device” was, even though that was the iPhone’s most revolutionary capability.

The phrase I’m thinking of right now, however, is “Are you getting it?” Are you getting that messaging — chatbots now, natural interfaces later — is how we will communicate with AI? Are you getting that not only will we not program computers, we won’t use them — AI will? Are you getting that pre-built UI — write once, run everywhere, for everyone — is dead? These are not three separate predictions: this is reality, right now, in 2026. The future is here, even if it’s not widely distributed. Or is it? — Read More

AI has gotten cheaper more quickly than any other transformative technology in history. The cost of achieving a given level of AI performance has fallen about 47% per quarter since 2023, or 13× per year. That price drop is four times faster than DNA sequencing, six times faster than compute, 18 times faster than lithium batteries, and (in the century up to 1973) 54 times faster than electricity.

… It is now widely understood that the AI boom is a macroeconomic force powerful enough to raise prices for the inputs it demands: chips, power, even the labor of electricians. Less well recognized is a paradoxical flip side: the price of the output from all those data centers is falling extraordinarily rapidly. — Read More

In one of my classes I asked the question I was afraid to ask but I just needed the answer to: “Who is afraid of not getting a job after graduating?” About eighty percent of the 150 people in the room raised their hands. … The other story arrives by email. PhD students who cannot wait to graduate, because they want to join a frontier lab and they have concluded that research in academia is meaningless. They are counting the years until they can leave.

I believe both stories are wrong, and wrong for the same reason. They assume the future of research belongs to whoever has the most GPUs. I think the opposite is true. Academia is probably about to have a renaissance, and the most exciting work of the next decade will happen in university labs — not in spite of their limited resources, but because of them. — Read More

All of Silicon Valley seems to be coming together to launch a new alternative to college for teens dreaming of becoming the next big tech founder.

Venture capital firm Andreessen Horowitz, also known as a16z, announced Tuesday that applications are now open for the first class of the creatively named The Horowitz Andreessen Academy. — Read More

A software factory harnesses loops at scale. You can run the loop with humans in it (light factory), trading judgment and concentration against speed and breakage. Or you can ignore the humans (dark factory) and let those agents scope, build, and ship code without anyone reading the details. If people stop reading, though, they’ll stop understanding your software. Your hardest job now is knowing which checks to build and how much autonomy to delegate.

This idea of the software factory is a term that dates back to Bob Bemer’s paper, “The economics of program production,” given in 1968. For half a century, many have dreamed of a world in which software is a repeatable and instrumentable production process (analogous to stamping out car parts in a factory) rather than the isolated craft of individuals. Historically, this dream has generally (although not universally) fallen flat, in part because of the difficulty of stamping out ideas.

But in the last two years, things have changed dramatically enough that now it makes sense to take a fresh look at the old dream. And since some subtleties can easily be glossed over, it’s worthwhile to be somewhat precise about exactly what’s really new and different, and what may be recurring traps, dressed up as new opportunities. — Read More

S

Salesforce is now making its own AI model. So is Crowdstrike. So is Thomson Reuters.

On Tuesday at its Dreamforce 2026 event in San Francisco, Salesforce announced Koa, its own domain-specific reasoning model that’s purpose-built to enable agents to handle business tasks more effectively while keeping your data private.

… The company doesn’t see this as a vehicle for job or SaaS replacement, but as an enterprise empowerment tool that is more precise, more secure, and more tailored to the AI needs of companies that use Salesforce. — Read More

A new AI model called Koa is one of the biggest announcements from Salesforce this week at its giant Dreamforce tech conference. Koa is the company’s first reasoning model, built on Nvidia’s open-weight Nemotron model. The two companies worked together to post-train Koa to excel at sales, marketing, and customer-support-related tasks.

Koa is a shining example of how the enterprise world’s needs for AI are diverging from what the frontier labs are offering. Proprietary AI labs would rather have enterprises uploading files, code, prompts, and feedback directly into their models and agents, and spending millions to do so. — Read More

… AI has a way of amplifying the culture it arrives in, including the good parts. A team that experiments and learns gets more room to do both. An engineering-driven organization that has neglected product work can now produce the consequences of that neglect much faster. And spend an extraordinary amount of time reviewing them.

That matters when we talk about a software factory, because a product team gets to shape much more than the rate at which code comes out. — Read More

The debate on US datacenters has never been so politically charged. Four states have acted in under two months. New York has stopped issuing environmental permits for datacenters, Texas has paused the next step in its massive ERCOT interconnection queue, Pennsylvania has pulled datacenters out of fast-track permitting and made state permits conditional on new guardrails, and Oregon has frozen datacenter deals on state-owned land.

Beyond the state level, more than 300 towns, cities and counties have voted to halt datacenters over the past year and a half. If you read the news, the US datacenter buildout looks like it is being legislated to a halt, one government board at a time.  — Read More

Over the past several weeks, the AI sector has become a magnificent anthropological study in collective hysteria—a caricature of a serious industry completely unmoored from reality. As self-important executives across the globe rush to microphones to sound the alarm, the line between saving humanity and saving a stock price has entirely evaporated. If you’re confused, you’re not alone.

To understand the current state of artificial intelligence, you must look past the dense fog of corporate PR and institutionalized fear. Strip away the math and the jargon, and you find a time-honored Silicon Valley tradition: a high-stakes financial theater where truth is the damsel in distress, tied to the train tracks, while a band of billionaire eccentrics plays the roles of both the villains and the caped crusaders chasing the elusive stone.The “stone” at the heart of this multi-billion-dollar romance is Artificial General Intelligence (AGI), a theoretical silicon messiah that tech titans dangle before the public to keep stock prices high and regulators at bay.

Consider the stage-managed grandiosity of recent weeks. On one side of the valley, Greg Brockman and Nvidia’s Jensen Huang pace the floors of heavily air-conditioned auditoriums, hands waving like television evangelists, heralding the imminent arrival of Artificial General Intelligence (AGI). It is a beautiful, cinematic pitch: the silicon messiah is coming, and it will cure disease, solve climate change, and incidentally justify Nvidia’s dizzying, gravity-defying market capitalization.

But then, the script takes a sharp, theatrical turn. Suddenly, the very priests who built the temple are rushing out into the street, shouting that the altar is on fire.

Within days of each other, young prodigies and weathered executives began sounding the alarm. Jacob Coxon, a brilliant, terminally earnest safety researcher, dramatically walked away from his keyboard, leaving behind a viral trail of warnings that the industry is casually gambling with human survival. A quick confirmation by Evan Hubinger alongside OpenAI’s Jakub Pachocki’s essay, An Alien Mind, the message was clear and terrifying: the machine is becoming too powerful, self-improving beyond our control, and it might just destroy humanity by the end of the decade.

The chorus rose to the executive suites. Dario Amodei of Anthropic published a sweeping essay pleading for a synchronized pause. Sam Altman of OpenAI nodded in solemn, public agreement, while Elon Musk of xAI chimed in from the digital ether, all calling for a coordinated industry slowdown to save humanity from itself.

It was a masterclass in what a cynic might call “existential marketing.”

To the casual observer, it looks like a boardroom full of Oppenheimers weeping over their creation. But if you follow the money, an entirely different, far uglier picture emerges. Over at David Sacks’ firm, the venture capitalist didn’t buy the tears for a second. Sacks took to the microphone to slam the top executives, bluntly accusing them of trying to bypass antitrust laws to form a classic market cartel.

And why would they want a cartel? Because behind the closed doors of San Francisco boardrooms, the sweat is beginning to bleed through the custom-tailored hoodies.

This isn’t the first time the Valley has tried to pull off this kind of regulatory theater. We’ve seen this script before. Back in 1995, Bill Gates looked out from his fortress in Redmond and realized Netscape’s web browser was about to turn Windows into a glorified, irrelevant set of plumbing. Gates didn’t panic by engineering a better product; he panicked by trying to rewrite the rules of the road. Microsoft executives famously marched into Netscape’s Mountain View headquarters and tried to slice up the internet market like an afternoon cake, politely suggesting that Netscape stick to Mac and Windows 3.1 while Microsoft took everything else. When Netscape refused to play ball, Microsoft simply suffocated them by tying Internet Explorer to the operating system and cutting off their oxygen. Altman, Amodei, and Musk are playing the exact same game, just with a modern, apocalyptic twist. When you can’t guarantee you’ll win a fair fight on product merit, you try to partition the market. Except this time, instead of using operating system monopolies to choke out the competition, they are trying to use the threat of a science-fiction apocalypse to trick Uncle Sam into building the moat for them.

The cold, hard truth of the AI business is that it depends on a terrifyingly small handful of customers. These are massive corporations buying bulk compute to figure out what, exactly, they are supposed to do with it. For the past two years, the spending only went up. But recently, the data scientists at corporate spend-management platform Ramp noticed a chilling anomaly in the ledger. For the first time since the boom began, spending among the top 1% of AI buyers has begun to contract. The enterprise customers, having spent millions on automated chatbots that occasionally lie about the company’s return policy, are quietly cutting back.

When you see AI labs suddenly increase their frantic pressure for government regulation, it isn’t because they are frightened of a science-fiction Terminator waking up in a server rack. It is because the financial picture is getting worse by the day. They are begging for government-mandated “safety” fences because fences keep out competitors. They want compliance costs so astronomicallyhigh that no scrappy startup can ever challenge their territory. They are trying to lock in their market share before the venture funding dries up entirely.

But the ultimate joke—the real house of cards—is that even if the U.S. government grants them their cartel, the frontier cannot be paced.

Enter the eccentric misfits from across the Pacific. While Silicon Valley was busy calculating how many billions it takes to train a model to write corporate emails, a Chinese startup called DeepSeek quietly changed the laws of economic gravity. In September, DeepSeek rolled out models priced so aggressively low that it made the American labs look like they were selling landlines in the age of the iPhone. DeepSeek’s models are priced cheaper without prompt caching than Anthropic’s or OpenAI’s are with it. On a blended price index, DeepSeek’s inference costs arrived on the order of 200 times cheaper than Anthropic’s average pricing.

The brilliant, eccentric hackers proved that you don’t need a hundred-billion-dollar sovereign wealth fund to build frontier intelligence; you just need better engineering and a complete disregard for Silicon Valley’s margin expectations.  Adding to the geopolitical friction, the Chinese government asserted that “Fear mongering, confrontation, competition will just disrupt [the] process of global AI governance.”

This leaves the Western tech titans in a catastrophic squeeze. Their domestic customers are tightening their purses, a foreign competitor has commoditized their only product by a factor of two hundred, and their massive infrastructure debts are coming due.

So where does the truth lie?

It lies exactly where President Donald Trump suggested when he breezily dismissed the existential warnings as “things that won’t happen,” declaring that the United States must simply push ahead to win the geopolitical race. China’s Minister of State Security, Chen Yixin, echoed the sentiment, saying AI has become “the main battleground for global technological competition and a new arena for strategic rivalry among major powers.” In a world governed by raw hubris and nationalist competition, a polite corporate agreement to slow down is an illusion. Besides, the federal government isn’t about to suspend the Sherman Antitrust Act just so a handful of venture-backed tech barons can legally collude on market supply.

The existential threat isn’t that a super-intelligence is going to turn humanity into batteries. The threat is that the entire, multi-billion-dollar apparatus is built on bad math, artificial scarcity, and a desperate desire to build a regulatory moat before the world realizes that the emperor has no clothes.

The tech barons aren’t trying to save us from the stone; they are just desperately trying to figure out how to keep bromancing it until the check clears.