In partnership with
12 ways to build real wealth
Most people have exactly one income stream: their job. Marina Mogilko (AKA Silicon Valley Girl) has twelve.
The Hustle's free wealth-building guide reverse-engineers how she got there by turning side projects into a 7-figure portfolio of courses, products, and passive commissions. It's a practical blueprint you can borrow one piece at a time.
Get it for free when you subscribe to The Hustle and join 1.5M+ readers.
Good morning,
Last Tuesday, America voted. Not just for senators and governors. In 41 local ballot measures across ten states, people were asked a much simpler question: do you want a data centre next door?
The answer was no. Forty times no. By CNBC's count of the Ballotpedia list, voters sided against the data centres in every single measure except one, in a village in North Dakota called Harwood. Wilmington, Ohio threw out four re-zonings for server farms by roughly four to one. El Dorado, Kansas voted itself a ten-year ban. Not close. Not even a debate.
Then it went up the ticket. In Ohio, Sherrod Brown took back a Senate seat by branding his opponent “the face of data centers”. The governor's mansion went to a physician who wants a moratorium until the industry covers 100% of its own utility costs. In Texas, on track to become the world's biggest data-centre market, the governor lost his job to a challenger who ran against the buildout. A CNBC voter poll found 56% of voters called data centres a major factor in their choice.
And Elon Musk, who wrote the biggest cheque of the whole election, at least $474 million, lost both of his marquee races in Texas and then posted that data centres in space were “looking pretty appealing right now”. You can buy rockets but not towns, it turns out.
Two days before the vote, Axios reported that OpenAI already had a plan ready for the morning after. Not a plan to use less water. Not a plan to pay for the grid. But a plan to stop saying “data centre” and start saying “AI compute campus”.
When you’re out of ideas, hire a word guy. The humming, windowless box that doubles your electricity bill becomes a campus, and the people who voted against it become, presumably, anti-education.
Let me be clear. I like this technology. I write about it every week and use it every day. But the model is not the box. The box has neighbours. And the neighbours just told us, in the only language a politician understands, what they think about paying for it.
The industry's own numbers do not help its case. OpenAI has eight gigawatts planned on a single site in Ohio, a project CNBC says could cost more than half a trillion dollars. Anthropic's updated prospectus lists $895 billion in compute commitments. Somebody has to live next to all of that, and pay for the grid upgrades long before any dividend shows up, if it ever does.
Watch Washington now. The Democrats who just took the House are already talking about AI executives testifying under oath. One congressman's advice to the industry was two words long: “Lawyer up.”
And before anyone in Europe gets smug: those boxes are coming to a field near you too, with the same pitch. Jobs. Innovation. Campus. Before you nod along, ask one question. Who gets the bill?
Welcome to the Blacklynx Brief
AI News
Voters reject data centres in 40 of 41 local ballots. Tuesday's US midterms put 41 local data-centre measures before voters in ten states; all but one, in Harwood, North Dakota, went against the industry, per Ballotpedia and CNBC. Democrats who campaigned on pausing projects also won the Ohio and Texas governorships. (CNBC)
Google unveils Gemini 4 Carbon, cyber defenders first. Google says its new frontier model tops the Artificial Analysis Intelligence Index and sets a new state of the art on exploitation and vulnerability-finding benchmarks. Access is limited to vetted cyber defenders for now, with no date for general release. (Google DeepMind)
Google ties school Search to accepting its AI. Google told the New York Times that K-12 districts which refuse Gemini will also lose Search for students, because the two are “deeply integrated”. Bernie Sanders called it extortion, Lina Khan called it monopoly abuse, and Google says schools get “more choice, not less”. (TechCrunch)
CrowdStrike ships a kill switch for rogue AI agents. The security firm added an “AI Kill Switch” to Falcon Guardian that cuts a misbehaving agent's access to its underlying model through an AI gateway. Coordinated containment across endpoints, cloud, identities and browsers is promised for later. (CrowdStrike)
MIT shows GPT-6 Astra rewriting another model's weights by hand. In an early MIT CSAIL study, Astra lifted a small Qwen model's MATH-500 score from 7.9% to 48.1% by editing about 0.016% of its weights, with no gradient training. It failed to improve Gemma 3, and the authors call the results preliminary. (MIT CSAIL)
AI Quick News
Anthropic researchers and academics at EPFL and MIT released FlashProve, an open-source prototype that uses cryptographic proofs to show a response really came from a specific committed model.
SpaceX's Grok API will invisibly watermark generated text by default from November 7 to comply with the EU AI Act, with an opt-out for team admins.
Marvell agreed to buy optics maker Lumentum in an all-stock deal valuing it at about $115 billion, with closing not expected before the first half of 2028.
Anthropic's amended IPO prospectus shows about $17 billion of third-quarter revenue, a $567 million operating loss and $895 billion in compute commitments.
Cohere announced an initial close of a Series E targeting up to C$7 billion, led by Schwarz Group, to build a frontier AI lab outside the US and China.
Cisco president Jeetu Patel told Fortune he gave a 27-year-old ex-founder unlimited AI tokens and four weeks to prove a two-person team can rewrite the company's legacy code.
Meta quietly launched Style Sketch, an AI fashion app that builds a wardrobe from your Instagram photos and generates virtual try-ons, so far only in Indonesia, Singapore and Thailand.
Deep Principle released Hydrogen-1.0, an open-weights chemistry and materials model post-trained on GLM-5.3 that the startup says beats GPT-6 Astra on OpenAI's FrontierScience-Research benchmark.
OpenAI hired former US Army Secretary Dan Driscoll as its senior counselor for national security modernisation.
Sam Altman said on the Cheeky Pint podcast that OpenAI will point its robotics programme at building data centres, and that “self-replicating data centers” are “within grasp”.
Startup Particle released Particle-1, a DeepSeek V4.1 Flash post-train that it says uses about 45% fewer tokens on coding tasks at near-identical accuracy.
Google opened a phased preview of its Connected Apps platform, letting developers bring their apps into Gemini and Search via MCP, with partner apps heading to Search in the coming weeks.
OpenAI scrapped a planned cut to ChatGPT Pro allowances and instead doubled usage limits for its Pro and Business Premium tiers after subscriber backlash.
Anthropic is reportedly SpaceX's largest single customer, the unnamed “Customer B” that paid its AI division 34.5% of third-quarter revenue, according to a person familiar with the matter.
Elon Musk said Grok Bot usage was throttled because serving it had become “unsustainably high relative to revenue”, after subscribers complained of hitting their limits early.
Netflix is rolling out AI Discover, a TV tab that builds theme-based recommendations, to US members watching in English first.
Anthropic's Addy Osmani confirmed that Claude Code is now prompted not to recreate specific existing games over IP concerns, and some developers say the rule is refusing projects they own.
OpenAI pulled hundreds of employees into an election operations centre to watch for ChatGPT misuse around the US midterms.
Artificial Analysis launched v5 of its Intelligence Index, with 60% of the score now based on private test sets and the gap between the leader and the best open-weights model widening to 12 points.
Anthropic renamed Max 5x to Max and Max 20x to Ultra at unchanged prices, and removed the five-hour usage limit for verified Ultra accounts.
Anthropic extended its GSA agreement with a 40% discount on Claude for federal, state and local agencies, saying civilian federal usage grew about 20-fold in a year.
Legal AI firm Harvey acquired Aida, a sales-platform startup co-founded by a Glean co-founder, its fifth acquisition this year, to help law firms build their own ontologies.
Inference platform fal raised $1.3 billion in a Bessemer-led Series F, roughly doubling its valuation to $16 billion since April.
Closing Thoughts
That’s it for us this week. Please like and subscribe 🙂
