New York Halts Data Centers, Apple Sues OpenAI, & Meta Faces AI Layoff Suit
Moratorium shifts bottleneck from chips to permits, Lawsuit targets former iPhone executive Tang Tan, and Former employees allege algorithmic termination targeting
New York’s Data Center Moratorium Wins on Vibes
New York Governor Kathy Hochul signed an executive order halting permits for data centers 50 megawatts or larger, potentially affecting more than a dozen projects. The state’s Department of Environmental Conservation will not issue any new permits, and the moratorium lifts only after an environmental review that Hochul expects will take about a year.
I grew up in Wisconsin, and visited family back in June. I was shocked how everybody was against data centers. They think it will suck up all the water, crush their property values, and make electricity unaffordable. I pointed out that AI seemed more hated than nuclear power, and everyone thought that was justified.
Two-thirds of respondents in a recent survey said they were concerned about data centers driving up electricity prices, and people told researchers they would rather have an Amazon warehouse in their backyard than a data center. Only 10% of Americans are more excited than concerned about AI in daily life.
In Northern Virginia, data center revenue has enabled it to reduce the real property tax rate every year for the past decade. Semi Analysis calculated that xAI’s Colossus 2 facility in Memphis uses roughly the equivalent of two and a half In-N-Out stores worth of water each year. The resource consumption people fear is a fraction of what they imagine, and the places that have built these facilities show the opposite of what people assume.
This image from Patrick Hedger went viral earlier this week. Estimates are data centers will use 900,000 acres by 2028, which would be less than 4 of the individual squares here. Basically less than flowers.
This is not a popular take, and it is not the best way to get subscribers. But this isn’t my main job, and I do have a bias because of my main job. I have been heads down with generative AI since before ChatGPT, and my company Provyx exists thanks to LLMs. And while I admit the bias, I truly believe that technology has been a great enabler for all of us, and thus feel the vibes are off. People are agreeing with whatever supports their preconceived notions.
Hochul said “Progress shouldn’t arrive with a higher utility bill, deleted water supply, or noise pollution”, and that framing is politically potent regardless of whether the underlying facts support it. Less than a quarter of the public believes AI will give the economy a boost, and less than a third are confident the government will regulate it responsibly. The constituency for a moratorium is broad, and the constituency for facts is narrow.
The developers and hyperscalers most exposed are those with pipeline projects in New York and any state that follows this playbook. Through 2030, nearly a quarter of new data centers will exceed 500 megawatts, and each one of those projects will face a public that has already decided the answer is no.
I wrote last month that “while building on Earth should be cheaper, data centers are requiring more permission, and opposition is getting more fierce.”
The more fierce that local opposition becomes, the more viable it becomes building data centers in space.
Apple’s Trade Secret Lawsuit Against OpenAI Is About the Talent War
Apple filed a 41-page complaint in the U.S. District Court for the Northern District of California alleging OpenAI and its employees engaged in a coordinated effort to steal confidential information and intellectual property. The suit targets Tang Tan, OpenAI’s Chief Hardware Officer, who spent 24 years at Apple as Vice President of Product Design for the iPhone and Apple Watch.
At the end of the day, Apple probably feels betrayed. Jony Ive framed his partnership with OpenAI as an attempt to undo the harms caused by the smartphone, which isn’t exactly a ringing endorsement of the company that made him wealthy and famous.
Apple of course cannot attack Ive directly, so they’re going after Tan instead, Apple cannot attack Ive directly, so it is going after Tan while pointing to the more than 400 former Apple employees now at OpenAI.
The trade secret claims here are impossibly broad. Apple is essentially arguing that any former employee at OpenAI is a walking trade secret, and that employing them entitles OpenAI to jumpstart its hardware efforts using knowledge those people carry in their heads. But from where else would OpenAI hire the personnel to build a consumer electronics company? Apple has never faced real competition for its best hardware people, and they don’t seem happy with this new situation.
OpenAI’s recent acquisition of Ive’s startup io and reports of a screen-free smart speaker described as a “humanlike AI companion” give Apple a concrete competitor to point at. The device reportedly includes moving mechanical elements and is being built with input from former Apple engineers who worked on iPhone and Mac. Apple claims its internal investigation uncovered evidence that OpenAI used confidential information as it develops this product.
When Apple finds a smoking gun in the form of an employee blatantly lifting documents, it must feel satisfying to swing it at the company that represents everything making your competitive position worse.
And while this may seem like a loser move from Apple, I think this is exactly what Steve Jobs would’ve done. He famously went thermonuclear on Android, calling it a “grand theft”.
And maybe that’s what Apple needs. It may not be best for OpenAI, or their employees, but maybe this is the attitude Apple has missed in the era of Tim Cook.
Meta’s AI Layoff Lawsuit Is About Who Gets to Blame the Algorithm
26 former employees are suing Meta, claiming the company used a “constellation” of internal AI tools to score, rank, and select workers for termination, and that the system disproportionately targeted people on medical or parental leave. The layoffs happened in May, when Meta cut 10 percent of its staff. The plaintiffs allege Meta’s tools, including an internal AI assistant called Metamate, employee-trained AI agents, and dashboards displaying AI token usage, failed to exclude protected leave status from their scoring, effectively penalizing employees for exercising their legal rights.
Using AI to decide who gets fired lands poorly, and it feeds the same vibes problem I wrote about when New York hit pause on data center permits. The public already suspects AI companies of cutting corners on people, and watching a trillion-dollar company algorithmically cull workers on medical leave confirms that fear.
But the accuracy question matters just as much as the optics. When we covered the Cloudflare layoffs, I wrote about the doorman fallacy, the tendency to reduce a role to its most visible task and assume the rest has no value. The same fallacy applies to AI performance scoring. These systems measure what is easy to count, like token usage on internal dashboards, and miss what is harder to quantify, like institutional knowledge, mentorship, and the fact that someone on leave is by definition not generating the metrics the system is optimizing for.
But was this really done using only AI? Meta spokesperson Tracy Clayton says “workforce management and organizational decisions were and are made by people, not AI”, and I think that’s closer to the truth. Companies do not fire 8,000 people by algorithm alone. Managers build the criteria, executives approve the lists, and HR signs off.
The AI may have scored employees, but humans decided what the scores meant and whether to act on them. Blaming the algorithm is convenient because it lets everyone involved pretend the machine made the call, and layoffs are often used to shed talent the company already wanted gone. It’s the same reason companies blame the computer when a credit algorithm discriminates.
If this case survives summary judgment, I wouldn’t be surprised if AI hiring tools are the next shoe to drop, because companies use them far more often for hiring than firing.