Lex

What’s a technology that you think is overhyped? I’m going to give a sideways answer to this, which is that the venture capital business model needs to be understood as requiring hype. You can go back to the Netscape IPO, and that was the proof point that made venture capital the financial lifeblood of the tech industry. Venture capital looks at valuations and growth, not necessarily at profit or revenue. So you don’t actually have to invest in technology that works, or that even makes a profit, you simply have to have a narrative that is compelling enough to float those valuations. So you see this repetitive and exhausting hype cycle as a feature in this industry. A couple of years ago, you would have been asking me about the metaverse, then last year, you would have asked me about Web3 and crypto, and for each of these inflection points there’s an Andreessen Horowitz manifesto. It’s not simply that one piece of technology is overhyped, it’s that hype is a necessary ingredient of the current business ecosystem of the tech industry. We should examine how often the financial incentive for hype is rewarded without any real social returns, without any meaningful progress in technology, without these tools and services and worlds ever actually manifesting. That’s key to understanding the growing chasm between the narrative of techno-optimists and the reality of our tech-encumbered world.

Stand by this: www.politico.com/newsletters/...

Exactly, not much is done for the IRL solution that people have been begging for support on.

🔐
www.eff.org/deeplinks/20...

I Don't Like What's Happening I love ending pieces with personal thoughts about stuff because I am an emotional and overly honest person, and I enjoy writing a lot. I do not, however, enjoy telling you at length how brittle everything is. An ideal tech industry would be one built on innovation, revenue, real growth based on actual business returns that helped humans be better, not outright lie about replacing them. All that generative AI has done is show how much lust there is in both the markets and the media for replacing human labor — and yes, it is in the media too. I truly believe there are multiple reporters who feel genuine excitement when they write scary stories about how Dario Amodei says white collar workers will be fired in the next few years in favour of "agents" that will never exist. Everything I’m discussing is the result of the Rot Economy thesis I wrote back in 2023 — the growth-at-all-costs mindset that has driven every tech company to focus on increasing quarterly revenue numbers, even if the products suck, or are deeply unprofitable, or, in the case of generative AI, both. Nowhere has there been a more pungent version of the Rot Economy than in Large Language Models, or more specifically GPUs. By making everything about growth, you inevitably reach a point where the only thing you know how to do is spend money, and both LLMs and GPUs allowed big tech to do the thing that worked before — building a bunch of data centers and buying a bunch of chips — without making sure they’d done the crucial work of “making sure this would create products people like.” As a result, we’re now sitting on top of one of the most brittle situations in economic history — our markets held up by whether four or five companies will continue to buy chips that start losing them money the second they’re installed. I am disgusted by how many people are unwilling or unable to engage with the truth, favouring instead a scornful, contemptuous tone toward anybody who doesn'…

There is an overwhelming condescension that comes from fans of generative AI — the sense that they know something you don't, something they double down on. We are being forced to use it by bosses, or services we like that now insist it's part of our documents or our search engines, not because it does something, but because those pushing it need us to use it to prove that they know what's going on. To quote my editor Matt Hughes: "...generative AI...is an expression of contempt towards people, one that considers them to be a commodity at best, and a rapidly-depreciating asset at worst." I haven't quite cracked why, but generative AI also brings out the worst in some people. By giving the illusion of labor, it excites those who are desperate to replace or commoditize it. By giving the illusion of education, it excites those who are too idle to actually learn things by convincing them that in a few minutes they can learn quantum physics. By giving the illusion of activity, it allows the gluttony of Business Idiots that control everything to pretend that they do something. By giving the illusion of futurity, it gives reporters that have long-since disconnected from actual software and hardware the ability to pretend that they know what's happening in the tech industry. And, fundamentally, its biggest illusion is economic activity, because despite being questionably-useful and burning billions of dollars, its need to do so creates a justification for spending billions of dollars on GPUs and data center sprawl, which allows big tech to sink money into something and give the illusion of growth. I love writing, but I don't love writing this. I think I'm right, and it’s not something I’m necessarily happy about. If I'm wrong, I'll explain how I'm wrong in great detail, and not shy away from taking accountability, but I really do not think I am, and that's why I'm so alarmed. What I am describing is a bubble, and one with an obvious weakness: one company's ability to s…

I hate that so many people will see their retirements wrecked, and that so many people intentionally or accidentally helped steer the economy in this reckless, needless and wasteful direction, all because big tech didn’t have a new way to show quarterly growth. I hate that so many people have lost their jobs because companies are spending the equivalent of the entire GDP of some European countries on data centers and GPUs that won’t actually deliver any value. But my purpose here is to explain to you, no matter your background or interests or creed or whatever way you found my work, why it happened. As you watch this collapse, I want you to tell your friends about why — the people responsible and the decisions they made — and make sure it’s clear that there are people responsible. Sam Altman, Dario Amodei, Satya Nadella, Sundar Pichai, Tim Cook, Elon Musk, Mark Zuckerberg and Andy Jassy have overseen a needless, wasteful and destructive economic force that will harm our economy and the tech industry writ large, and when this is over, they must be held accountable. And remember that you, as a regular person, can understand all of this. These people want you to believe this is black magic, that you are wrong to worry about the billions wasted or question the usefulness of these tools. You are smarter than they reckon and stronger than they know, and a better future is one where you recognize this, and realize that power and money doesn’t make a man righteous, right, or smart. I started writing this newsletter with 300 subscribers, and I now have 67,000 and a growing premium subscriber base. I am grateful for the time you’ve given me, and really hope that I continue to help you see the tech industry for what it currently is — captured almost entirely by people that have no interest in building the future.

The AI bubble is the ultimate growth-at-all-costs Rot Economy campaign - a fake industry built on the back of selling ultra-expensive and unprofitable software that consumers and businesses don't need, with only NVIDIA and consultants making money in the end.
www.wheresyoured.at/the-haters-gui/

I'm Alarmed! I realize I've thrown a lot at you, and, for the second time this year, written the longest thing I've ever written. But I needed to write this, because I'm really worried. We're in a bubble. If you do not think we're in a bubble, you are not looking outside. Apollo Global Chief Economist Torsten Slok said it last week. Well, okay, what he said was much worse: “The difference between the IT bubble in the 1990s and the AI bubble today is that the top 10 companies in the S&P 500 today are more overvalued than they were in the 1990s,” Slok wrote in a recent research note that was widely shared across social media and financial circles. We are in a bubble. Generative AI does not do the things that it's being sold as doing, and the things it can actually do aren't the kind of things that create business returns, automate labor, or really do much more than one extension of a cloud software platform. The money isn't there, the users aren't there, every company seems to lose money and some companies lose so much money that it's impossible to tell how they'll survive. Worse still, this bubble is entirely symbolic. The bailouts of the Great Financial Crisis were focused on banks and funds that had failed because they ran out of money, and the TARP initiative existed to plug the holes with low-interest loans. There are few holes to plug here, because even if OpenAI and Anthropic somehow became eternal money-burners, the AI trade exists based on the continued and continually-increasing sale and use of GPUs. There are limited amounts of capital, but also limited amounts of data centers to actually put GPUs, and on top of that, at some point growth will slow at one of the Magnificent 7, at which point costs will have to come down from things that lose them tons of money, such as generative AI.

Before you ask- But, Isn’t The Cost Of Inference Going Down? You do not have proof for this statement! The cost of tokens going down is not the same thing as the cost of inference goes down! Everyone saying this is saying it because a guy once said it to them! You don't have proof! I have more proof for what I am saying! While it theoretically might be, all evidence points to larger models costing more money, especially reasoning-heavy ones like Claude Opus 4. Inference is not the only thing happening, and if this is your one response, you are a big bozo and doofus and should go back to making squeaky noises when you see tech executives or hear my name. But Ed, What About ASICs? Okay, so one argument is that these companies will use ASICs — customized chips for specific operations — to reduce the amount they're spending. A few thoughts: When? Say OpenAI and Broadcom actually build their ASIC in 2026 (they won't) — how many of them will they build? Do they have contracts with companies that can actually produce high-performance silicon, of which there are only three (Samsung, TSMC, and arguably SMIC, which is currently sanctioned), and these companies typically have their capacity booked well in advance. Even starting a production run of a semiconductor product can take weeks. Do they have the server architecture prepared? Have they tested it? Does it work? Is the performance actually good? Microsoft has failed to create a workable, reliable ASIC. What makes OpenAI special? It takes a lot of money to build these chips and they are yet to prove they're better than NVIDIA GPUs for AI compute, and even if they do, are they going to retrofit every data center? Can they build enough? If this actually happens, it still fucks up the AI trade. NVIDIA STILL NEEDS TO SELL GPUs!

We are in a bubble. We must admit we are in a bubble and prepare investors and consumers as such. And no, the cost of inference isn't coming down, nor are specialised "ASIC" chips going to save the day. Outside of magic and pixie dust, this bubble *will* burst.
www.wheresyoured.at/the-haters-gui/

The AI Trade Is Entirely About GPUs, And Is Incredibly Brittle As A Result As I've explained at length, the AI trade is not one based on revenue, user growth, the efficacy of tools or significance of any technological breakthrough. Stocks are not moving based on whether they are making money on AI, because if they were, they'd be moving downward. However, due to the vibes-based nature of the AI trade, companies are benefiting from the press inexplicably crediting growth to AI with no proof that that's the case. OpenAI is a terrible business, and the only businesses worse than OpenAI are the companies built on top of it. Large Language Models are too expensive to run, and have limited abilities beyond the ones I've named previously, and because everybody is running models that all, on some level, do the same thing, it's very hard for people to build really innovative products on top of them. And, ultimately, this entire trade hinges on GPUs. CoreWeave was initially funded by NVIDIA, its IPO funded partially by NVIDIA, NVIDIA is one of its customers, and CoreWeave raises debt on the GPUs it buys from NVIDIA to build more data centers, while also using the money to buy GPUs from NVIDIA. This isn’t me being polemic or hysterical — this is quite literally what is happening, and how CoreWeave operates. If you aren’t alarmed by that, I’m not sure what to tell you. Elsewhere, Oracle is buying $40 billion in GPUs for the still-unformed Stargate data center project, and Meta is building a Manhattan-sized data center to fill with NVIDIA GPUs. OpenAI is Microsoft's largest Azure client — an insanely risky proposition on multiple levels, not simply in the fact that it’s serving the revenue at-cost but that Microsoft executives believed OpenAI would fail in the long term when they invested in 2023 — and Microsoft is NVIDIA's largest client for GPUs, meaning that any changes to Microsoft's future interest in OpenAI, such as reducing its data center expansion, would eventuall…

Microsoft, Meta, Google, Apple, Amazon and Tesla aren't making much money from AI — in fact, they're losing billions of dollars on whatever revenues they do make from it. Their stock growth is not coming from actual revenue, but the vibes around "being an AI company," which means absolutely jack shit when you don't have the users, finances, or products to back them up. So, really, everything comes down to NVIDIA's ability to sell GPUs, and this industry, if we're really honest, at this point only exists to do so. Generative AI products do not provide significant revenue growth, its products are not useful in the way that unlocks significant business value, and the products that have some adoption run at such a grotesque loss.

The AI trade isn't based on company revenues, profits, the usefulness of LLM products, or anything other than the perpetual growth of GPU sales. This industry is brittle, and lacks any compelling investment story beyond vibes. And the vibes are sickly.
www.wheresyoured.at/the-haters-gui/

Companies Are Using The Term "Agent" To Deceive Customers and Investors "But Ed!" you cry, "What about AGENTS?" Let me tell you about agents. The term "agent" is one of the most egregious acts of fraud I've seen in my entire career writing about this crap, and that includes the metaverse. When you hear the word "agent," you are meant to think of an autonomous AI that can go and do stuff without oversight, replacing somebody's job in the process, and companies have been pushing the boundaries of good taste and financial crimes in pursuit of them. Most egregious of them is Salesforce's "Agentforce," which lets you "deploy AI agents at scale" and "brings digital labor to every employee, department and business process." This is a blatant fucking lie. Agentforce is a god damn chatbot platform, it's for launching chatbots, they can sometimes plug into APIs that allow them to access other information, but they are neither autonomous nor "agents" by any reasonable definition. Not only does Salesforce not actually sell "agents," its own research shows that agents only achieve around a 58% success rate on single-step tasks, meaning, to quote The Register, "tasks that can be completed in a single step without needing follow-up actions or more information." On multi-step tasks — so, you know, most tasks — they succeed a depressing 35% of the time. Last week, OpenAI announced its own "ChatGPT agent" that can allegedly go "do tasks" on a "virtual computer." In its own demo, the agent took 21 or so minutes to spit out a plan for a wedding with destinations, a vague calendar and some suit options, and then showed a pre-prepared demo of the "agent" preparing an itinerary of how to visit every major league ballpark. In this example's case, "agent" took 23 minutes, and produced arguably the most confusing-looking map I've seen in my life.

It also missed out every single major league ballpark on the East Coast — including Yankee Stadium and Fenway Park — and added a random stadium in the middle of the Gulf of Mexico. What team is that, eh Sam? The Deepwater Horizon Devils? Is there a baseball team in North Dakota? I should also be clear this was the pre-prepared example. As with every Large Language Model-based product — and yes, that's what this is, even if OpenAI won't talk about what model — results are extremely variable. Agents are difficult, because tasks are difficult, even if they can be completed by a human being that a CEO thinks is stupid. What OpenAI appears to be doing is using a virtual machine to run scripts that its models trigger. Regardless of how well it works (it works very very poorly and inconsistently), it's also likely very expensive. In any case, every single company you see using the word agent is trying to mislead you. Glean's "AI agents" are chatbots with if-this-then-that functions that trigger events using APIs (the connectors between different software services), not taking actual actions, because that is not what LLMs can do. ServiceNow's AI agents that allegedly "act autonomously and proactively on your behalf" are, despite claiming they "go beyond ‘better chatbots,’" still ultimately chatbots that use APIs to trigger different events using if-this-then-that functions. Sometimes these chatbots can also answer questions that people might have, or trigger an event somewhere. Oh, right, that's the same thing. The closest we have to an "agent" of any kind is a coding agent, which can make a list of things that you might do on a software project and then go and generate the code and push stuff to Github when you ask them to, and they can do so "autonomously," in the sense that you can let them just run whatever task seems right. When I say "ask them to" or "go and" I mean that these agents are not remotely intelligent, and when let run rampant fuck up everything and c…

Companies Are Using The Term "Agent" To Deceive Customers and Investors: When you hear the word "agent," you're meant to think of an autonomous. This is a lie. None of these products are autonomous agents, and anybody using the term agent likely means "chatbot."
www.wheresyoured.at/the-haters-gui/

The Magnificent 7's Weakpoint: NVIDIA As I write this, NVIDIA is currently sitting at $170 a share — a dramatic reversal of fate after the pummelling it took from the DeepSeek situation in January, which sent it tumbling to a brief late-April trip below $100 before things turned around. The Magnificent 7 stocks — NVIDIA, Microsoft, Alphabet (Google), Apple, Meta, Tesla and Amazon — make up around 35% of the value of the US stock market, and of that, NVIDIA's market value makes up about 19% of the Magnificent 7. This dominance is also why ordinary people ought to be deeply concerned about the AI bubble. The Magnificent 7 is almost certainly a big part of their retirement plans, even if they’re not directly invested. Back in May, Yahoo Finance's Laura Bratton reported that Microsoft (18.9%), Amazon (7.5%), Meta (9.3%), Alphabet (5.6%), and Tesla (0.9%) alone make up 42.4% of NVIDIA's revenue. The breakdown makes things worse. Meta spends 25% — and Microsoft an alarming 47% — of its capital expenditures on NVIDIA chips, and as Bratton notes, Microsoft also spends money renting servers from CoreWeave, which analyst Gil Luria of D.A.Davidson estimates accounted for $8 billion (more than 6%) of NVIDIA's revenue in 2024. Luria also estimates that neocloud companies like CoreWeave and Crusoe — that exist only to prove AI compute services — account for as much as 10% of NVIDIA's revenue. NVIDIA's climbing stock value comes from its continued revenue growth. In the last four quarters, NVIDIA has seen year-over-year growth of 101%, 94%, 78% and 69%, and, in the last quarter, a little statistic was carefully brushed under the rug: that NVIDIA missed, though narrowly, on data center revenue. This is exactly what it sounds like — GPUs that are used in servers, rather than gaming consoles and PCs (. Analysts estimated it would make $39.4 billion from this category, and NVIDIA only (lol) brought in $39.1 billion. Then again, it could be attributed to its problems in China, esp…

NVIDIA's quarter-over-quarter growth has also become aggressively normal — from 69%, to 59%, to 12%, to 12% again each quarter, which, again, isn't bad (it's pretty great!), but when 88% of your revenue is based on one particular line in your earnings, it's a pretty big concern, at least for me. Look, I'm not a stock analyst, nor am I pretending to be one, so I am keeping this simple: NVIDIA relies not only on selling lots of GPUs each quarter, but it must always, always sell more GPUs the next quarter. 42% of NVIDIA's revenue comes from Microsoft, Amazon, Meta, Alphabet and Tesla continuing to buy more GPUs. NVIDIA's value and continued growth is heavily reliant on hyperscaler purchases and continued interest in generative AI. The US stock market's continued health relies, on some level, on five or six companies (it's unclear how much Apple buys GPU-wise) spending billions of dollars on GPUs from NVIDIA. An analysis from portfolio manager Danke Wang from January found that the Magnificent 7 stocks accounted for 47.87% of the Russell 1000 Index's returns in 2024 (an index fund of the 1000 highest-ranked stocks on FTSE Russell’s index). In simpler terms, 35% of the US stock market is held up by five or six companies buying GPUs. If NVIDIA's growth story stumbles, it will reverberate through the rest of the Magnificent 7, making them rely on their own AI trade stories. And, as you will shortly find out, there is no AI trade, because generative AI is not making anybody any money.

The entire AI bubble is a mirage - nobody is making a profit and everybody is losing money other than NVIDIA. As a result, NVIDIA is the weakpoint - the AI bubble is based on them selling MORE GPUS every single quarter FOR FOREVER.
www.wheresyoured.at/the-haters-gui/

Safer Drivers, Stronger Devices | Microsoft Community Hub Surface is advancing Windows driver development by adopting Rust, a memory-safe programming language, to improve device security and reliability. Through...

Safer drivers, stronger devices (with #rustlang)

techcommunity.microsoft.com/blog/surface...

Maybe that was about Elizabeth Eckford and the white woman that expressed remorse. Famous photo of her yelling at Elizabeth

And I hate that she wasn't given any grace or moment to be a child during these times. She had to be a little soldier. With no one stopping the hate that was being spewed at her.

Great write up by @inanna.recursion.wtf of learning from @baileytownsend.dev’s Rust Statusphere example and deploying via Wasm in Cloudflare Workers.

Mastercard: Sex Work is Work. End Your Unjust Policy. Mastercard's new policy unfairly targets the adult content industry, making sex workers more vulnerable, especially Black trans women. It must be stopped.

This itch.io NSFW policy change is unfortunately nothing new, the problem isn't the platforms themselves but rather Visa/MC strong-arming these services into revising their NSFW policies.
So if you're eligible I invite you to sign this ACLU petition: action.aclu.org/petition/mas...

EU cloud gang challenges Broadcom's $61B VMWare buy in court CISPE cites recent channel changes, but the deal was decided on different matters +COMMENT  Trade group Cloud Infrastructure Services Providers in Europe (CISPE) has filed a formal appeal before the European General Court to seek annulment of the European Commission's decision to approve Broadcom’s acquisition of VMware.…

EU cloud gang challenges Broadcom's $61B VMWare buy in court

Zoom. Take out the forced AI messaging on open. Ffs

Screen shot of a series of buttons displayed on the top right of Google Drive. In the middle is a star symbol, representing Google's Gemini AI.

The same, only now the star is highlighted by a red box, thanks to uBlock's Element Picker Mode

BOOM. The star is no more. The evil is vanquished.

Man, I love uBlock Origin.

I’m on the jubilee obsession, I’m so sorry

I’m only an X-men tourist 🙌🏼

Rereading Ruby Bridges story and I forgot how disgusted I was by the back lash.
"Every morning, as Bridges walked to school, one woman would threaten to poison her, while another held up a black baby doll in a coffin.

Yeah that kind of hate doesn't just dissipate into thin air.

A blue square promoting a talk at the Oxidize Conference in Berlin from September 16 - 18, 2025. This post features an image of Daniel Silverstone, who will be giving a talk titled Safety monitor: Using Rust in mixed-criticality Linux systems.

At Oxidize 2025, Daniel Silverstone (@codethink.co.uk) presents Safety Monitor - an open-source, configurable and Rust-based tool to track system state on Linux that supports defining a safe-state definition in mixed-criticality environments: oxidizeconf.com #RustLang #Safety #Linux

A photograph of a purple hardcover book with ‘Book of Masks by Frayed Symphony’ written on the cover and half of Majora’s Mask on the front. The book is resting in orange and green leaves on the floor

A photograph from the interior of Book of Masks, open at the Great Fairy’s Mask page. The background is purple with Link wearing the mask on the inside. The book is resting in a blueberry bush with green leaves around it

A photograph from the interior of Book of Masks, open at the Fierce Deity Mask page. The background is purple with Link wearing the mask on the inside. The book is resting amongst a patch of leaves.

[ BOOK OF MASKS ]

My 56 page Majora’s Mask fanbook is here! Each double page spread is an illustration for the 24 masks you can collect.

A lot of work went into this love letter to one of my favourite games- please check out and enjoy this hardback collection;

www.lulu.com/shop/michell...

#zelda

I know we weren't taught media literacy and parsing news stories as a part of basic education. But this is BAD.

Because this is not just a "can't read and comprehend" problem. It's a lack of pattern recognition. It's not knowing where and how to verify. It's not knowing what primary sources are.

YouTube video by SukhberS Double Take dance (English)

This movie scene lives rent free in my head
youtu.be/MVBA8ykC3y4?...

the children yearn for fandom edits of vine compilations

EU states to test age-check app to limit children's access to online services The European Commission has unveiled a prototype of the app that Denmark, France, Greece, Italy and Spain will customize and launch within several months.

Five EU countries including France will test an app aimed at preventing children from accessing harmful content online by checking the age of users, the European Commission said Monday.

Carrot cake? Is that the best example of unhealthy food he can think of? He’s not just a eugenics asshole, he’s deeply out of touch.

Never going to get over Sean Paul yelling "five million and forty naughty shorty"

The many lives of Nakagin Capsule Tower on display at New York’s MoMA “The Many Lives of the Nakagin Capsule Tower” showcases the 50-year history of the building through nearly 45 pieces of contextual material and one of the capsules on display.

New York’s MoMA acquired one of the 23 saved capsules after the tower's deconstruction, now on view at the institution's street-facing Manhattan galleries for a full year.

I don't know who needs to hear this. Please clean y'all's cars. My OCD will activate and I'll forget the whole conversation we had.

I'm not talking about clutterbugs. I'm one too. I'm talking about sticky cupholders and shit. Please hit that car with a wipe or 10 😭

Exactly. And the opinions they form, they say as of its fact. Very annoying

It says a lot about people when they look UP to these men. Who otherwise would be ignored if they didn't gather technical skills that made them cool enough for a lifetime pass to do nonsense