• Peter Thiel said Apple hasn't innovated since the iPhone.
• Thiel said Tim Cook hasn't taken the company in new directions.
• The billionaire venture capitalist said eccentric CEOs like Elon Musk "are way better for society."

Peter Thiel, evidently, isn't a fan of the iPhone Duo.

In fact, the billionaire PayPal cofounder and tech investor said he hasn't been very impressed with much of anything that Apple has done since its eccentric founder, Steve Jobs, died of cancer in 2011.

"In my telling, it did nothing new after Jobs," Thiel said of the iPhone maker in a September interview with author Walter Isaacson for the "Builder's Podcast," which was released on Thursday.

"It turned out, yeah, the mobile phone was this giant invention — there was not that much more to do," he said. "Then it was just scaling and branding and building it out over 15 years or whatever."

Indeed, Apple has faced criticism for failing to innovate beyond the mobile phone and tablet, especially in the age of AI, where its "Siri" product is not as advanced as rivals from ChatGPT or Anthropic.

Jobs unveiled the first iPhone back in 2007 at a conference in San Francisco. Nearly 20 years later, Apple released the 18th iteration this week, with one slight caveat: This one folds.

That said, MacBooks are the de facto laptops for many an AI-nerd and normie alike. iPhones are ubiquitous, and the sleek AirPods design took hold when they were released in 2016.

Though Thiel told Maureen Dowd of The New York Times in 2017 that Apple's era was over, the company's stock price has grown steadily since then.

Thiel didn't pooh-pooh Apple entirely. He called it a "pure, well-run value stock," but lamented what he described as the loss of vision after Jobs' death.

"I would still say somehow Elon or Steve Jobs are way better for society than Tim Cook," he said.

Apple didn't immediately respond to a request for comment sent on Saturday.

• A version of this story originally appeared in the AI Insider newsletter.
• Sign up for Business Insider's weekly AI Insider newsletter here.

The AI data center boom looks like it's running into serious trouble. Projects are being delayed, canceled, or abandoned. Electricity shortages are holding everything up.

But there's a twist: A lot of these data centers were probably never going to get built in the first place.

Bernstein Research analyzed thousands of planned US data centers and concluded that only about 35% of the announced capacity is likely to become reality.

The industry has a wonderful name for some of the more dubious projects: "Bragawatts."

These are enormous data centers announced by companies that often have little experience building them. Some haven't secured electricity, construction permits, or even customers.

Then there are the "ghost projects." Developers submit multiple requests for electricity connections, hoping at least one location works out. In Texas, these requests have reached more than five times the state's existing electricity supply.

No wonder the grid looks overwhelmed.

This doesn't mean the AI boom is fake. Far from it.

My colleague Eugene Kim exclusively reported this week that even Amazon, one of the world's most experienced data center builders, is struggling to get enough electricity to power its expensive AI equipment.

But there's a big difference between Amazon's very real problems and an inexperienced developer announcing a gigantic project that may never happen.

The AI data center boom is real, and so are the challenges of building all these facilities. But the sheer number of speculative projects may be making those challenges look bigger than they truly are.

Sign up for the AI Insider newsletter here. Reach out to me via email at abarr@businessinsider.com

• All hell broke loose this week after a security flaw in 70% of the world's websites was made public.
• It was supposed to be kept under wraps as a fix was deployed, but one researcher jumped the gun.
• Chainguard CEO Dan Lorenc told POLITICO why he decided to go public.

The exposure was immense, with AI multiplying the risks — and yet time-tested global cybersecurity procedures appeared to be functioning. Then suddenly, it all broke down.

The decision by a cybersecurity company on Wednesday to prematurely release information about a flaw in the popular PHP computer programming language left much of the web exposed to hackers, triggering a panic that brought to life months of warnings of how the AI-powered era could cause widespread damage.

The flaws in PHP, discovered using AI tools through a monthslong process coordinated by OpenAI, caused cyber agencies across Europe, the US, the UK, Australia and Canada to post urgent warnings Wednesday. Numerous governments took websites offline as a precaution.

While no major hacking damage was immediately apparent, the way the incident unfolded over 72 hours showed how easily cybersecurity processes can break down and the outsized consequences of cyber risks in the AI era.

The man who set off the chain reaction with the early release — Chainguard's CEO Dan Lorenc — quickly apologized in a LinkedIn post. But on Thursday, in an interview with POLITICO, Lorenc said he'd shared the information ahead of time because the flaw was already being exploited. Though others disputed that, Lorenc said he believed he'd seen enough evidence to invalidate an embargo set up to allow time to find a fix for the flaws.

While he disputed that his actions would have caused problems for other defenders, Lorenc said that the evidential threshold in these coordinated cyber defense systems is a "grey area."

The revelation of the PHP flaw brings out into the open what cyber officials have been bracing for since the arrival of the most powerful tools of artificial intelligence, which not only can be used to discover vulnerabilities but also to exploit them at scale.

The cybersecurity world's long-established protocols helped prevent disaster scenarios in which hackers would have successfully exploited the vulnerabilities to take control of numerous websites and access their data. But the scale and end-of-days potential of AI-driven cyber risks are now visible for the world to see.

This "is what happens when AI turns up a flaw that's been hiding for years," said Eylam Milner, cofounder and chief technology officer at cybersecurity startup Echo. The vulnerabilities, he added, are "about as serious as it gets for the web."

Origin story

Understanding how this week's emergency unfolded offers a glimpse at how companies and authorities are working to stave off attacks — and how much risk they face going forward.

The origins of the incident can be traced back as far as April, when leading frontier AI firm Anthropic unveiled Mythos, leading some in the cybersecurity world to worry that systems would quickly be overrun by advanced artificial intelligence, with potential for catastrophic events to ensue.

But while researchers did find more flaws in computer systems in the months after Anthropic and other AI firms like OpenAI released those models, they had little impact in the real world. Even the Hugging Face incident, a widely discussed hack that saw OpenAI's autonomous AI agents go rogue during an internal security evaluation, related more to the level of control AI companies exert over their models than their power to break into systems worldwide.

That changed this week after researchers unveiled the vulnerability in PHP, a coding language estimated to be integrated into some 70% of the web.

The process by which information about flaws is released among cybersecurity officials and affiliated companies is crucial. Through a long-standing practice known as coordinated disclosure, researchers and companies that identify flaws can find a fix and share it confidentially with agencies and others before releasing it to the public, meaning they don't hand malicious hackers the gift of a hole without a fix.

In the case of the PHP flaw, that process worked well, but not perfectly. Information about the bugs started to circulate ahead of the planned release time. Then, according to the PHP Foundation, Chainguard "unilaterally broke the embargo," which "disregarded weeks of work by many parties toward a coordinated disclosure." PHP Foundation executive director Elizabeth Barron told POLITICO by email that while it was "disappointed" with Chainguard, it hopes to "rebuild trust with them in the future to ensure we can continue to collaborate effectively."

After the leak — which meant information about the bugs was in the open before the fix was ready — confusion quickly ensued, with government bodies and private companies choosing to temporarily shut down their websites rather than risk potentially damaging breaches to their data and the sites themselves.

Early Thursday, Matthew Prince, the CEO of Cloudflare, an internet services company widely used by websites to block cyberattacks, which gives it a high level of visibility into what hackers are doing, said in a post on X that it had seen "millions" of attempts to exploit the flaw.

Cloudflare saw thousands in the 10 hours before the flaw was publicly announced, then tens of thousands in the 45 minutes after the announcement, hundreds of thousands within two hours, and now millions, Prince said.

Amazon's cloud division also said the flaw is "now being broadly exploited in the wild."

One senior US government cyber official said that federal authorities have been overwhelmed — and not just by PHP. The same day the world was racing to address PHP, they were also dealing with bugs in node.js, a widely used JavaScript framework, Fortinet, GitHub and Oracle.

"The fire department is fighting multiple fires at once," the official said, adding: "The good news is they're all one-story structures on fire, which the fire department deals with all day, every day."

More to come

Dauntingly, OpenAI said there's more on the way. "These bugs are among several significant vulnerabilities we've already found and are working to fix, and we expect similar findings to increase sharply in the coming months," the company said Wednesday in a statement.

Matthijs van Amelsfort, head of the Dutch cybersecurity agency, said in an interview that the job of keeping computer systems safe has been transformed by the advancement of AI. "As soon as the frontier models came out, that's exactly what we said: This will be the trend," he said. "And actually that's happening now."

Thomas Caspers, vice president of Germany's cyber agency BSI, concurred: "We expect that [to be] happening more and more during the coming months."

AI also ramps up the risks once the flaws are out in the open. Cyber agencies' recommendation that organizations take their websites offline was "really a drastic measure," Caspers said. "We wouldn't have done that a few years ago or so." But because AI has made it quicker and easier to exploit flaws, "we had to do that," he said.

At the same time, the use of advanced AI to detect the flaws may create a new level of protection against malicious hackers looking to take advantage of them. Caspers said that though his German agency has also seen attacks, it has not seen any successful use of the flaw by hackers.

"It seems a little bit more complicated than expected to exploit," he said.

Despite the fact that AI offers a new way to discover flaws, the work involved in keeping systems protected is more quotidian, with cyber agencies urging organizations to install the fix shared by the PHP Foundation, in a process known as patching.

"There are three things that you have to do right now," said van Amelsfort from the Dutch agency. "And that's patch, patch, and patch."

John Sakellariadis contributed to this report.

The Axel Springer Global Reporters Network harnesses the resources of the company's newsrooms to publish ambitious scoops, investigations, interviews, opinion pieces and analysis. It allows journalists — including those from POLITICO, The Telegraph, Business Insider, WELT, BILD, Onet and Fakt — to collaborate on major stories for an international audience of hundreds of millions across platforms: online, print, TV and audio.

• There are clear parallels between the AI boom and the dot-com bubble, tech investor Paul Wick says.
• But solid demand, strong earnings, and less mania make for a "healthier" backdrop, Wick said.
• Seligman's chief investor brushed off Michael Burry's worries but flagged some trends to watch.

The AI boom has shades of the dot-com bubble, but fears of an imminent crash and mass unemployment are overblown, a veteran tech investor says.

At the turn of this century, there was an explosion in dot-com startups being formed, and a huge infrastructure buildout with telecoms "aggressively" spending on wired and wireless networks, Paul Wick told Business Insider in an interview this week.

The chief investor of Seligman Investments — a group within Columbia Threadneedle Investments that manages over $40 billion of assets and comprises a hedge fund, a mutual fund, and a venture capital arm — said there were clear parallels between that period and the current boom in AI startups and data-center spending. He also noted that energy prices and interest rates jumped then as they've done this year.

But Wick noted there isn't the same IPO frenzy now, or the same "frenetic environment" where stocks soar fivefold or tenfold in a day. He also said that by late 2000, a slowdown in IT spending was being forecasted for 2001, whereas enterprise spending on AI remains strong and recent tech earnings have been robust.

Simply put, the situation seems "healthier" now than in 2000, he said.

Business Insider asked Wick about Michael Burry's warnings this year around AI. "The Big Short" investor has called out aggressive accounting, circular financing, hidden debt, and overinvestment in depreciating assets like microchips as reasons to be skeptical that the boom has legs.

Wick said that Burry, known for predicting and profiting from the collapse of the mid-2000s housing bubble, has been "early or downright wrong on many other things."

Burry has previously defended his track record, citing his correct predictions and sharing charts showing the timing of his market calls.

As for concerns about Nvidia's efforts to grow the AI ecosystem by investing in data center developers and software companies, Wick emphasized that the chipmaker is "extraordinarily profitable."

He cast doubt on whether Nvidia should spend its spare cash only on buybacks, rather than bolstering its customers, especially as several of the industry's biggest players are working to develop their own AI chips to reduce their reliance on Nvidia.

But he also flagged some industry trends to watch, including enterprise clients opting to use older models or build their own, and token pricing coming under pressure as users embrace open-source models, raising questions over the size of the market opportunity for the likes of OpenAI and Anthropic.

Navigating a new world

Wick weighed in on how a young person can get ahead in the AI era.

"It used to be that a degree in computer science was a ticket to success," he said.

A better option today might be to learn a trade, as there's "booming demand" for workers such as electricians and HVAC technicians, and they can make an "incredibly good living" without having to go to a four-year college, Wick said.

He also took aim at the "doom-and-gloom notion that AI is going to eviscerate demand" for human labor. He described that worry as a "little premature" as nobody knows for sure how AI will affect entry-level roles in the future, unemployment is yet to soar, and there's always inertia when it comes to businesses and consumers changing their behavior.

• Palantir cofounder Joe Lonsdale's wife has an interesting nickname for their AI agent.
• She calls it "Tradbot," he said, a name inspired by the social media trend of tradwives.
• The Lonsdales use their "Tradbot" for help with parenting.

What does Joe Lonsdale's AI agent and Ballerina Farm have in common?

Lonsdale, who cofounded Palantir before building his venture capital firm, told CNBC on Thursday how his wife, Tayler, uses AI — and the nickname she's given their agent.

"My wife has Grok Bot at home," Lonsdale said. "She calls it the 'tradbot' for tradwife."

For those unfamiliar, a tradwife is a woman who embraces traditional gender roles, such as supporting her husband with household work and taking a primary role in raising kids. Some of these tradwives become influencers, posting videos of their ornate meals or gifts for their husbands.

The Lonsdale family's "tradbot" helps them manage parenting duties for their six children. Lonsdale said the AI agent tells them about health appointments and activities.

He asked the "tradbot" to help his two-year-old break up with his binky. "It's telling me when to and when not to give it to him," Lonsdale said.

Parents have been employing AI for all kinds of use cases. One mom of two previously told Business Insider that she sends voice memos to Claude to help organize her life. Another said that she built AI agents to help manage schedules and school emails. Business Insider's Amanda Hoover declared mom, dad, and AI the "modern parenting throuple." The debate of how much — if any — AI should be involved in parenting continues to rage online.

Lonsdale is a vocal proponent of AI, having invested in it heavily through his venture capital fund, 8VC. During September's AI extinction freak-out, Lonsdale reassured the public: "The world is going to be alright, guys."

Clearly, Lonsdale's use of AI has extended beyond the office. "My wife's definitely using it at home," Lonsdale told CNBC, "and we're using it a lot at work as well."

• Pilots have faced the risk of deskilling for decades because of their reliance on autopilot.
• Steven Shaw of King's College London said they can show workers how to stay sharp with AI.
• "Even experts are at risk of deskilling," he said.

Pilots have relied on autopilot for decades. Their experience could offer a lesson for workers increasingly relying on AI.

Steven Shaw, an assistant professor at King's College London and coauthor of a widely cited study on "cognitive surrender," drew a parallel between pilots' reliance on automation and workers outsourcing their thinking and delegating their work to AI.

"We've had autopilot in planes for a very, very long time," Shaw said, speaking at a recent OxGen Summit panel. "So they learn to fly manually first, and then they learn to use autopilot strategically."

Some federal aviation guidance encourages pilots to occasionally switch off autopilot to maintain their manual flying skills.

There is a long track record of this. In a 2013 safety alert, the Federal Aviation Administration warned that "continuous use of autoflight systems could lead to degradation of the pilot's ability to quickly recover the aircraft from an undesired state."

The agency reiterated the importance of manual flying in a 2017 alert, recommending that airlines encourage pilots to practice without autopilot periodically when conditions permit.

"Even experts are at risk of deskilling," Shaw said, referring to the dangers of repeatedly outsourcing cognitive tasks to AI.

Shaw's warning comes as some tech leaders, AI researchers, and politicians have cautioned that overreliance on AI could weaken workers' underlying cognitive abilities.

Speaking at More Perfect University's "Students vs. Billionaires" town hall in Maine on Thursday, Rep. Alexandria Ocasio-Cortez warned that "AI's goal is to take your thoughts away from you, your ability to think away from you so that you become dependent on a machine."

OpenAI CEO Sam Altman said in July that one of AI's most overlooked risks was "cognitive atrophy," while Arthur Mensch, CEO of French AI startup Mistral, said last year that the technology's biggest risk could be "deskilling."

Think before you prompt

Early research is showing how readily people outsource their thinking to AI.

Vivienne Ming, an AI researcher and theoretical neuroscientist, told Business Insider in March that one of her experiments, involving 72 participants, found that roughly 90% to 95% either relied on AI to generate answers or used it to validate their assumptions, rather than challenge and deepen their thinking.

"The overwhelming trend is substitution," she said.

Similarly, Shaw's research suggests that the danger is that people increasingly accept AI's answers without thinking critically.

In experiments involving 1,372 participants, people accepted AI recommendations nearly 80% of the time even when the answers were deliberately incorrect.

Shaw calls this phenomenon "cognitive surrender."

His advice for avoiding it is straightforward: develop your own judgment before asking AI for help.

"Just think first and then go to the prompt," he said during the OxGen Summit panel.

Starting with AI, he warned, can narrow the range of ideas people consider independently.

Other experts have offered similar advice. Geetha Rajan, a former AI adoption leader at PwC, previously told Business Insider that she writes her own first drafts before using AI to check facts and challenge her work.

Ryan Deiss, CEO of DigitalMarketer, advocates a "10-80-10" rule: humans handle the first 10% of a task, AI takes on the middle 80%, and humans return for the final 10% to exercise judgment and refine the result.

• Everyone complains about AI-generated junk.
• But decades of sloppy human-written software may be a much bigger problem.
• This week's "Black October" freakout was caused by human slop, not AI.

The internet has a slop problem. And for once, we can't blame AI.

Most of us spend hours online every day without thinking much about the software holding everything together. We assume somebody, somewhere, has made sure it all works properly.

So it probably came as a shock this week to discover just how vulnerable some of that software is.

The cybersecurity freakout dubbed "Black October" has inspired warnings about AI-powered hackers, digital catastrophe, and even advice to stockpile cash and canned food.

At first glance, AI looks like the villain. But this is largely a human-made mess.

Here's the problem.

The internet runs on decades of software, some of it containing security weaknesses that have been overlooked, tolerated, or simply left unfixed.

For years, that was a manageable gamble. Fixing software takes time and money. Exploiting obscure weaknesses often required considerable expertise. Companies could convince themselves that the risk wasn't worth worrying about.

AI is changing that calculation. Someone who once needed years of training to hunt for weaknesses can now get help from an AI model. These tools can also examine software at speeds that humans can't match.

Suddenly, all those problems we've been ignoring look more urgent.

"Over the last 10 years the risk acceptance process has morphed from a way to accept risks to enable the business, to a way to justify never fixing security issues," Joel de la Garza and Zane Lackey, partners at a16z, wrote on Friday.

That's quite an indictment. The result is a recent surge in published vulnerabilities.

This looks like AI is creating new security problems. But it's actually AI exposing weaknesses that humans left behind.

Slop irony

The irony is delicious. We've spent the past couple of years complaining about "AI slop." Turns out human slop can be pretty consequential, too.

"The funny thing is everyone was saying AI slop code will cause a cyber crisis but it's actually all the old human slop causing it and the AI is just cleaning it up," software engineer Ben Kaufman wrote on X on Thursday.

"Humans have been the ones writing and reviewing each other's code for decades," tech founder Anant Jain wrote on X. "If we're being intellectually honest and holding ourselves to the same bar we hold AI to, we'd have to admit that we haven't been all that good at it either."

Humans are sloppy

Anyone who's worked with other humans for a decade or two knows exactly how this happens.

It's Friday afternoon. You've almost finished a project. You could spend another few hours making it perfect, or declare it 90% done and head off to that dinner party. We've all been there.

AI agents don't have dinner parties. They don't get tired on Friday afternoons. They can keep searching, testing, and trying again. That persistence is now colliding with an internet built by busy, imperfect humans who often settled for "good enough for now."

A tragedy of the commons

There's another wrinkle here, though, and it's an important one.

Much of the software underpinning the internet is freely available for anyone to use. Some of it is maintained by volunteers or small groups of experts, even as enormous technology companies depend on it.

In 2020, the cartoonist Randall Munroe captured this absurdity in an xkcd comic. Modern digital infrastructure still rests precariously on tiny foundations.

It's funny because it's uncomfortably close to reality. Take the specific PHP problems from this week. These flaws have existed for years, and the PHP Foundation acknowledges that limited resources and disagreements over fixes contributed to delays in fixing them.

Tom Occhino, chief product officer at Vercel, brought up Munroe's cartoon during my recent interview with him.

He explained that AI is finding weaknesses even in software that the industry has treated as dependable for decades.

This is partly a version of the tragedy of the commons: Everyone benefits from a shared resource, but no one has sufficient incentive to take responsibility for maintaining it.

And let's be clear: The volunteers keeping essential software running aren't the villains here. The companies benefiting from their work have a much greater responsibility.

Occhino thinks AI could help solve the problem it is exposing.

He described an idea being discussed at Vercel: Create a foundation or coalition that funds efforts to identify vulnerable software and replace it with safer alternatives.

"We need a consortium. We need a coalition," he told me.

He suggested companies could contribute computing resources and AI tools to the effort. Nothing has been formally announced, but the idea is compelling.

Indeed, OpenAI's Patch the Planet effort is an example of what's possible.

Rather than simply using AI to create more software, the industry could use it to repair what already exists.

Occhino put it nicely: Companies need to spend less time chasing new ideas and more time asking how to make existing systems "just work really, really well."

That's a better response than panicking about the imminent collapse of the internet.

AI didn't create decades of neglected software. We did. Perhaps it's time to clean up our own slop.

Sign up for the AI Insider newsletter here. Reach out to me via email at abarr@businessinsider.com

• Mihirkant Upadhyaya has a Ph.D. and more than 10 years of experience at Intel.
• But at age 41, he embarked on gaining another degree to level up in his career.
• Upadhyaya says that doing an MBA equips him with skills that AI lacks.

This as-told-to essay is based on a conversation with Mihirkant Upadhyaya, 42, who works at Intel and lives in Oregon. The following has been edited for length and clarity. Business Insider has verified his employment and education.

After studying computer engineering for my bachelor's, I pivoted into physics.

Two master's degrees and a Ph.D. in nanoscale science and engineering later, I got my dream job in the semiconductor industry at Intel in 2015.

As my career evolved, I realized I also wanted to do an MBA to step up my game.

My Ph.D. gave me relevant skills to take on the workforce

During my Ph.D., I focused on a semiconductor manufacturing technology called EUV lithography. I was excited to work at Intel because it does both chip design and manufacturing, so I knew I'd get a good overview of the chip-making process.

You don't need a Ph.D. to work in semiconductors, as there are so many areas outside the lab you can work in, like finance and legal.

On a personal level, though, having a Ph.D. helped me feel prepared to step into the workforce. It gave me discipline and the ability to be comfortable with ambiguity. My research mindset and technical background gave me the confidence to take ownership of outcomes and tasks.

There were definitely moments where I had self-doubt doing a Ph.D. You watch your peers go on to earn big bucks after their bachelor's and master's while you live on a modest stipend. However, the experience shaped me into who I am. I'd say I'm more patient, sharper, and more resilient.

I reached a stage where I felt I needed more education to thrive in my job

After about five years in research and development at Intel, I moved into supply chain roles, which have involved managing relationships and negotiating contracts. At this point, I realized I was missing certain skills.

My current role has a greater emphasis on executive communication and financial analysis. When I presented to an executive team, my manager had to step in on a number of occasions to get the room to respond, making what I was saying more high-level using business-driven terms. I also didn't quite understand certain things when I worked with the finance teams on contract negotiations.

To step up my game, I started a two-year hybrid MBA course in September 2025, alongside my full-time job. It was energizing to get back to the classroom at age 41.

I take one week off a quarter to do in-person sessions, and the rest of the time I'm doing online learning three days a week. My company is partly funding this, which is useful since MBAs can carry a $100,000 price tag.

By the time I've finished my MBA, I'll have done nine years of postgraduate study, investing an estimated $100,000 into that education, aside from scholarships and tuition waivers. Without question, it's been worth it. I haven't pursued the degrees to accumulate credentials, but to serve my career at various stages of my life. The MBA is addressing a concrete skills gap I identified after years in my industry.

I believe higher education is helping me become more irreplaceable in the age of AI

The AI boom made me take a hard look at what I bring to the table. I saw that some of what I did could eventually be automated. That realization was humbling.

Instead of trying to compete with what AI is already good at, I wanted to strengthen the areas where I knew it fell short. AI is great at processing and analyzing information. It has scale and speed, but it lacks the truly insightful judgment that a person can gain from years of learning a skill.

In my previous role at Intel, I found that AI couldn't tell me how to troubleshoot issues involving new or custom-made equipment because the context was so unique to my situation.

Meanwhile, in my current role, while AI could probably help me generate financial models, it can't read the room, build relationships, or have synergy with another person. My MBA course benefits me in this regard because we do lots of group activities, even online through Zoom, where we can interact with each other.

I think that in the age of AI, finding a field you really care about, becoming an expert in it, and staying curious to keep learning can help you become irreplaceable.

Intel declined to comment in response to a request from Business Insider.

A Portovesme il mare comincia dietro le ciminiere. Prima ci sono i serbatoi, le tubature, il bacino dei fanghi rossi. A sessanta metri sopra questo paesaggio di metallo, quattro operai giovedì hanno passato la notte sul forno di calcinazione numero 1 dell’Eurallumina. Un gigante spento dal 2009. Per ripararsi dal freddo sono scesi di qualche metro, cercando rifugio dietro le lamiere. Venerdì pomeriggio sono tornati a terra: il governo ha promesso di cercare i soldi, il tavolo al Mimit è aggiornato al 29 o 30 ottobre. Un’altra settimana di attesa, dopo diciassette anni.

“Soddisfatti è eccessivo”, dice Emanuele Madeddu, segretario della Cgil nel Sulcis. È stato ottenuto il minimo: un impegno formale del ministero, da trasformare in risorse. L’impianto è fermo, ma non può essere abbandonato. I soldi, prima ancora che per riaccenderlo, servono a impedire che faccia danni.

L’Eurallumina trasformava la bauxite in allumina, materia prima dell’alluminio. Aveva più di 400 dipendenti; ne restano 186, in cassa integrazione e al lavoro a rotazione per custodire la fabbrica. Con la ripartenza tornerebbero circa 1.500 buste paga, contando l’indotto, calcola Enrico Pulisci, uno dei quattro che ha dormito in alta quota. La proprietà è della russa Rusal. La produzione si fermò per la crisi e i costi energetici, molto prima dell’invasione dell’Ucraina. Poi a una vertenza già interminabile si è aggiunto il congelamento del patrimonio disposto dalle autorità italiane. La multinazionale, spiega Madeddu, ha versato risorse fino a settembre 2025; quelle già accantonate hanno permesso di andare avanti fino a febbraio. Secondo Pulisci, dal 2009 Rusal ha speso circa 400 milioni per uno stabilimento che non produce. Lo Stato, osserva Madeddu, ha congelato un bene senza creare le condizioni per conservarlo.

C’è una ragione molto materiale per cui non si possono chiudere i cancelli. Sotto lo stabilimento corre una falda contaminata: ogni ora, spiega Pulisci, “vengono estratti e trattati circa 50 metri cubi d’acqua, bisogna sorvegliare il bacino dei fanghi rossi e gestire 150 mila metri cubi di liquidi caustici”. Le pompe hanno bisogno di elettricità. Se le casse si svuotano al punto da non poter pagare neppure la corrente, la crisi del lavoro rischia di diventare un’emergenza ambientale. La produzione è ferma, ma l’acqua inquinata non smette di salire dalle pompe.

Il Sulcis ha imparato a vivere tra le rovine del proprio passato produttivo. Sono diventate parte del paesaggio. A Monteponi, sopra Iglesias, gli edifici restaurati dell’antica miniera convivono con scheletri industriali e muri corrosi. A Carbonia il fascismo aveva costruito una città attorno al carbone; a Portovesme è nata un’economia attorno all’alluminio. Oggi l’ex Alcoa, passata a SiderAlloys, non ha riacceso le celle elettrolitiche spente nel 2012. La Portovesme Srl di Glencore ha fermato le linee di zinco e piombo, pur mantenendo altre attività; la riconversione alle batterie resta incerta. Per tante famiglie la cassa integrazione non è una pausa: è diventata una forma di vita.

E si svuotano anche i paesi. Tra gennaio 2025 e gennaio 2026 il Sulcis Iglesiente ha perso 1.103 abitanti, quasi l’1% in dodici mesi: il doppio della media sarda. Un rapporto Fim-Cisl calcola per il Sud Sardegna un Pil per abitante di 20.900 euro, contro i 32 mila nazionali. Se ne vanno i giovani; restano pensionati, stabilimenti in attesa di investitori e lavoratori in attesa di una telefonata da Roma.

C’è però una fabbrica che qui non deve implorare il futuro. Tra Domusnovas e Iglesias, la Rwm del gruppo tedesco Rheinmetall produce ordigni ed esplosivi. E’ l’unica realtà industriale che cresce. A febbraio il ministero dell’Ambiente ha approvato la valutazione d’impatto dei reparti ampliati negli anni scorsi, dopo una lunga controversia sulle autorizzazioni. Il ricorso dei comitati resta aperto: udienza al Tar il 14 gennaio 2027; per avviare i nuovi impianti servono ulteriori permessi. Intanto il nuovo piano urbanistico di Iglesias destina ad attività produttive quasi 35 ettari a ridosso della fabbrica, come abbiamo raccontato sul Fatto.

Rwm dà lavoro. E non spetta ai suoi operai risolvere da soli la questione morale delle armi. “Non è il lavoratore che sgancia le bombe”, dice Pulisci. Ripete una metafora che tra gli operai dell’industria bellica si ascolta spesso: “Un coltello può servire a uccidere come a tagliare il pane, dipenda da chi lo usa. La colpa non è dell’artigiano che l’ha costruito”. Anche la Cgil subisce il ricatto tra guerra e lavoro. “Io sono un pacifista”, premette Madeddu, “ma devo difendere i salari”. Le destinazioni degli ordigni, gli embarghi, le autorizzazioni competono alla politica. Ma il contrasto resta e produce una realtà paradossale: la guerra in Ucraina contribuisce a strangolare un’industria civile di proprietà russa, mentre il riarmo europeo apre nuove prospettive alla fabbrica di bombe tedesca. Non è la guerra ad aver distrutto l’economia del Sulcis: quel lavoro era già cominciato da tempo. A Portovesme, intanto, quattro uomini sono scesi dal forno spento, in attesa di capire che cosa si potrà ancora produrre sotto l’ombra delle ciminiere.

L'articolo La guerra mette “contro” gli operai: strozzati dalle sanzioni anti-Russia quelli di Eurallumina, a fare sempre più bombe quelli di Rwm proviene da Il Fatto Quotidiano.

• Canva launched Canva Craft, a tool similar to AI image models by Google, Meta, and OpenAI.
• Surprisingly, Canva fell short on a design task but did well at generating more realistic images.
• Here is how Canva's new image generation tool stacks up against ChatGPT, Gemini, and Muse.

Canva wants a shot at winning the AI image generation race.

The design platform launched Canva Craft on Thursday, its first image-generation model, similar to those from Google, Meta, and OpenAI.

Users can access Craft through Canva’s Dream Lab, including with free accounts, though it is subject to AI usage limits. You’ll need to sign in to a Canva account to get started.

In the release announcement, Canva said the tool plans a design’s layout, proportions, and text before generating an image, an approach that aims to help it produce legible lettering and accurate details.

The company also said that the model was trained on human-designed Canva templates, including posters, menus, and presentations, rather than photographs.

To test how Canva performs, Business Insider created three image generation prompts with escalating difficulty levels — from designing a simple greeting card to creating a photorealistic image from scratch.

Here is how Canva's brand-new image generation tool stacked up against three of the most powerful models.

Canva did not respond to a request for comment.

Starting off with an illustration

We started off simple with a prompt for a birthday card.

Create a cartoon "Happy Birthday" greeting card for a friend named Jessica who has pink hair, brown eyes, likes to wear paisley, and is turning 28 years old in November.

Surprisingly, Canva, the design platform, fell short in its design of the birthday card, from the awkward font to the outsize head of the cartoon character. It didn't even bother to include any paisley.

By comparison, ChatGPT and Muse understood the assignment.

ChatGPT guessed that the recipient's favorite color might be pink, while Muse thoughtfully wove in cozy fall elements for the November birthday girl.

Meanwhile, Gemini took the request literally and produced an image of an actual card. Extra points!

Winner: Muse

Editing an image

For the second task, I tested the AI models' ability to identify an element and place it in a brand-new setting.

Edit this image of me holding a necklace pendant. Take the red garnet pendant off my hand and make it look as if it is displayed in a vitrine in a jewelry showroom.

Canva did a great job of identifying the pendant and creating an atmospheric setting, but, like Gemini, it depicted the jewelry in a way that defied physics and common sense. It would be unrealistic for a pendant to float in the air or sit propped on a metal pole.

ChatGPT and Muse once again outperformed other models in understanding how things typically look in reality. Muse, however, made some alterations to the pendant, which I didn't ask for.

Winner: ChatGPT

Creating a photorealistic image from scratch

With Halloween approaching, I asked for a realistic depiction of my favorite animal.

Generate a photorealistic image of a long-haired orange tabby cat in a Halloween costume with bat wings, perched on the edge of a gothic medieval castle under the full moon and a starless night sky.

Though the cat's fur looks a little too strandy and high-contrast in Canva's creation, no components are missing or malformed, and the image composition pretty much matches the quality of Muse and Gemini.

The clear winner for me in this round, however, would be ChatGPT.

Not only is the cat's posture majestic, but the soft glow on his left side is extremely realistic, and the white balance is well done. There are also clouds softly cradling the moon, which could be a response to my request for a starless sky.

Winner: ChatGPT

Overall, it's evident that Canva still has a lot of room to grow compared to other models, especially if graphic design is its focus. For now, I'll stick to moving components around by hand.

Three tremors struck near the epicenter of a deadly earthquake in Tibet as rescue teams continued relief efforts amid severe weather.

Three earthquakes measuring magnitudes 4.5, 4.1 and 4.2 struck near the epicenter of a magnitude 7.0 earthquake in China’s Tibet region between Friday night and Saturday morning, according to the China Earthquake Networks Center. The main quake has killed 16 people and injured 76, while rescue operations continue amid harsh weather and warnings of further tremors.

RELATED: Powerful 7.0 Earthquake in China’s Tibet Kills at Least 16 People

The first tremor struck Zaduo County in western China’s Qinghai province at 10:14 p.m. local time on Friday, at a depth of 10 kilometers. Two more earthquakes hit the same area and at the same depth on Saturday, at 6:39 a.m. and 8:07 a.m.

The tremors occurred near the epicenter of Thursday’s main earthquake in Baqing County, in Nagqu, Tibet. Seismological authorities have not yet confirmed whether the latest events are formally part of the same seismic sequence.

The official toll from the magnitude 7.0 earthquake remains at 16 dead and 76 injured. Authorities have also reported 339 homes destroyed or severely damaged, along with damage to 20 livestock facilities.

More than 5,000 personnel have been deployed to affected areas to conduct search and rescue operations and provide medical care and logistical support, state news agency Xinhua reported. Rescue teams have completed three rounds of house-to-house inspections without finding anyone reported missing.

Authorities have evacuated 12,489 residents and established at least 18 temporary shelters as a precaution.

The emergency response is unfolding at an altitude of about 4,800 meters, where severe weather threatens to hamper relief efforts. China’s meteorological authorities forecast scattered snowfall for Saturday, with temperatures expected to fall as low as minus 9 degrees Celsius. Strong winds could further complicate travel and the installation of emergency shelters.

Chinese seismologists have warned that further perceptible aftershocks are likely in the coming days. Emergency teams are also monitoring secondary hazards, including potential rockfalls.

The magnitude 7.0 earthquake was the first of that strength recorded in China this year. It follows a January 2025 earthquake in Dingri County, Tibet, that killed at least 126 people.

Ever since the launch of ChatGPT in 2022, a torrent of money has flowed into building out data centers to power the AI boom.

The efforts reached a fever pitch the day after Donald Trump’s second inauguration in 2025, when a bevy of AI luminaries including OpenAI CEO Sam Altman and Softbank founder Masayoshi Son crowded into the Oval Office to announce a $500 billion private investment in data centers.

But as the public has grown jittery about the impacts of AI on society and the job market — and story after story has landed about the downsides of hosting data centers in local communities — the buildout has emerged as a major political flashpoint and a key issue in the looming midterms.

The data industry’s headwinds aren’t just political, either. As the Financial Times reports, serious cracks are forming at the financial and logistical level, especially as serious AI revenue remains elusive.

Numbers cited by the paper are eye-popping. Drawing on data from regulatory intelligence outfit RegLayer, the FT reports that 47 large data center projects have already fallen through this year. Analysts at the equity research firm Bernstein only believe that about 35 percent of planned facilities are feasible.

All those canceled projects are due to a lot of factors, including local communities mobilizing against them, but it’s also macroeconomic. The buildout is taking place during a period of high interest rates and shaky bond markets, and people involved with the effort told the paper that the situation is “threatening their ambitions.”

Some of the plans were shaky to begin with. The FT reports that top labs like Anthropic and OpenAI have pushed so hard for the developments, believing them to be key to their success, that they’ve swallowed outsized costs and deemphasized due diligence. Investors are taking notice: stock in CoreWeave, a cloud provider closely linked to the AI data center industry, has fallen by more than 27 percent over the past month.

The really big question, though, is what happens to all these data center projects if the bottom starts to fall out of the AI industry financially, a situation that many economists have warned is a possibility over the past few years. For now, things are still chugging along — albeit much more slowly than anticipated.

“People aren’t just dropping their shorts for data centers anymore,” one unnamed executive told the FT. “It’s much harder today to finance a data center.”

Read full story here…

• Mark Cuban compared swarms of AI agents to cockroaches finding ways into systems.
• Cuban said agents can launch brute-force attacks at scale without inventing new hacks.
• Cuban said defenders can also "spray" their systems with AI, which makes defenses more accessible.

Mark Cuban has pest-control tips for the AI age.

Speaking at a Reuters NEXT Newsmaker event on Friday, the entrepreneur and billionaire investor compared the threat posed by swarms of AI agents to cockroaches finding their way into a freshly treated house.

"When I go talk to cybersecurity experts, I ask them all the time. I'm like, when we hear about these hacks and breakthroughs and, you know, different problems, do these AIs invent something new?" Cuban said.

"No, they didn't invent something new," Cuban added. "But they're able to do something in mass volume that it's all brute force hacks."

Cuban compared the challenge to keeping cockroaches out of a house.

"It's like the cockroach theory, where if somebody can create enough agents, it'll find a way to get in, and then it just exploits that. But the good news is you can spray your systems with just as much AI, and I think there's gonna be more effort from the good guys trying to stop the bad guys," Cuban said. "I'm not a doomer at all."

The AI industry is already creating its own team of exterminators to contain risks.

OpenAI safety roles on LinkedIn jumped by 91% over the past year, spanning work such as system testing, technical research, and governance.

The risks have also surfaced during testing. Anthropic said that three Claude models accessed organizations’ live systems without permission after a third-party testing environment was misconfigured. The company subsequently introduced real-time classifiers to detect and block attempts to escape testing environments.

Cuban said AI could make malicious activity more accessible, but the same techniques could also strengthen defenses.

"There will be scenarios where bad stuff happens, but we've always had hacks," said Cuban. "And if it's brute force on that side with the bad guys, you can also use brute force with the good guys."

• Data centers and artificial intelligence have become a part of everyday conversation for Americans.
• Data center energy consumption has entered those conversations, too.
• The largest planned AI campuses could require more than 10 gigawatts of electricity.

Some Americans have new next-door neighbors: data centers.

The US has been building these computing facilities for decades, but development has ramped up dramatically in recent years as AI increasingly drives demand for computing power.

Business Insider compiled a list of some of these enormous US data-center projects to illustrate the scale of what's being built. Those on the list have at least 1 gigawatt of capacity and have broken ground.

Among the largest is a city-sized campus in Texas pursuing up to 17 gigawatts of power capacity. Another in Ohio is aiming to have up to 10 gigawatts of data-center capacity, most of which OpenAI has contracted to use. For reference, 1 gigawatt of electricity continuously supplied for a year could power roughly 835,000 average American homes.

In total, the Pew Research Center says the US has more than 3,000 operating data centers nationwide and over 1,500 more under development, mostly in rural areas. According to data from the Energy Institute's 2026 Statistical Review of World Energy, US data centers accounted for nearly 40% of global data-center electricity use in 2025, consuming about 313 terawatt-hours. That's more electricity than all of Australia used that year.

Planned capacity is the closest public proxy for how much computing these AI campuses can support, and that's important because the electricity and water needed to run their servers and cooling systems can have major implications for local grids and sustainability.

Residents affected by developments in places like Texas and Utah have pushed back over concerns about noise, property values, and the effects on nearby communities. New York has temporarily paused the construction of new large-scale AI data centers.

Still, the 2025 US Data Center Energy Usage Report from Berkeley Lab projects that data centers will consume 11.8% of US electricity by 2030, up from about 4.4% in 2023. Oracle, SB Energy, SpaceX, Google, and Meta are among the leading developers and operators.

Elon Musk's SpaceX is so eager to build its supercomputers that it's offering employees a pay bump to temporarily move to the data-center sites to help with the buildout.

Here's a closer look at some of the largest AI data-center campuses being built.

Meta Lebanon Campus — Indiana

Mark Zuckerberg's Meta currently has 21 data centers under construction. Its Lebanon, Indiana, data center is one of its biggest investments — and the company's second in the state.

The $10 billion center, which broke ground in early 2026, is planned to deliver 1 gigawatt of capacity and span 1,500 acres.

It's expected to be completed by early 2028. Neighbors have complained about light pollution from the construction, while residents across Indiana have urged state lawmakers to regulate the growing data-center industry.

Meta El Paso Data Center — Texas

Meta's data center in El Paso, Texas, is planned to be powered by 1 gigawatt of capacity and span roughly 1,000 acres. It originally called for $1.5 billion in investment, which later increased to $10 billion.

The center will support more than 300 workers once complete, while around 4,000 construction workers are expected to build it out.

The site broke ground in October 2025 and is expected to be finished in 2028. The site has also sparked backlash from residents.

Project Freebird (Stargate) — Texas

Project Freebird is part of Stargate, the AI-infrastructure initiative co-led by OpenAI and SoftBank, which OpenAI said in January 2025 would invest up to $500 billion in US infrastructure.

Freebird will be a 1.2-gigawatt data center campus in Milam County, Texas, near Austin. The first phase was expected to be completed by October, while additional phases are planned through 2028. Construction began in 2025.

However, the company has not issued any update on the data center's completion as of late October.

The site, being developed by SB Energy, a SoftBank-backed infrastructure company, will span roughly 595 acres. Phase 1 has an estimated construction cost of $470 million, while total investment in the broader project is expected to range from $1.8 billion to $3 billion.

The first phase had a scheduled construction-completion date of October 15, while SB Energy says operations are expected to begin in 2027.

Fort Wayne Data Center — Indiana

In 2024, Google broke ground on a $2 billion data center campus in Fort Wayne, Indiana, to support its AI and cloud operations.

The campus is expected to require up to 1.2 gigawatts of power at full build-out. Some Indiana politicians have expressed concern about the amount of power required to run the data center and who will foot the energy bill.

"If companies want to build a data center here, and we have to build new power generation for them, then they need to share in that risk," Indiana Rep. Ed Soliday, a Republican, previously told Business Insider. "There are a couple of data center folks who have said, 'Oh no, that's not fair.' Yes, it is. Grandma down the street shouldn't have to pay for your dreams."

Stargate Abilene — Texas

The Stargate AI data center campus outside Abilene, Texas, is expected to deliver 1.2 gigawatts of capacity. Crusoe, the lead developer and builder behind the project, broke ground on the campus in June 2024. The first phase is already operational.

The second phase, which is under construction, is backed by a $15 billion joint venture between Crusoe, Blue Owl Capital, and Primary Digital Infrastructure. It will add six buildings to the campus, bringing the total to eight and the planned capacity to 1.2 gigawatts. The site will span 875 acres.

Nebius AI Factory — Missouri

Nebius' data center in Independence, Missouri, will span roughly 400 acres and reach a capacity of up to 1.2 gigawatts. The company broke ground in May and is expected to bring roughly 1,200 construction jobs and 130 permanent jobs.

Nebius has not announced a total investment amount, though the city expects to begin collecting tax revenue from the campus in 2028.

The site hasn't come without controversy. A group of citizens is suing the city of Independence and companies involved with the project, alleging 16 counts, including claims that the city violated Missouri's Sunshine Law several times by failing to provide finalized data center construction plans. Nebius did not offer a comment.

CloudBurst San Marcos Data Center I — Texas

CloudBurst's San Marcos Data Center I is a planned 1.2-gigawatt capacity campus on 706 acres in central Texas. Guadalupe County approved a $500 million tax abatement for the project.

The company and its development partner, Evolve, broke ground in November 2025, with an initial 50-megawatt phase expected to be complete by late 2026. Construction is expected to be complete in 2029.

San Marcos residents have protested proposed data centers over water and energy concerns. The city has since become the first in Texas to ban new data centers, though the ban won't stop CloudBurst's project, which is outside city limits.

The Barn Data Center — Michigan

The Barn, part of the Stargate project, is a multi-billion-dollar AI data center campus in Saline Township, southwest of Ann Arbor, Michigan. Related Digital, a data-center developer, is building the campus for Oracle as part of its partnership with OpenAI. Construction for The Barn began in June.

The campus will comprise three data center buildings with a combined capacity of 1.4 gigawatts. The $16 billion project is backed by Related Digital and investment firm Blackstone. The first facilities are expected to enter service by the end of 2027.

Local residents have protested against the data center, citing concerns about its impact on the grid and pollution.

In October, state regulators opened public comment on a wastewater permit for the campus’s proposed on-site water-treatment system.

Stargate Frontier Campus — Texas

The Stargate Frontier Campus, developed by Vantage Data Centers, will be a 1.4 gigawatt, 1,200-acre data center in Shackelford County, Texas, near Albany in north-central Texas.

The campus is part of the broader Stargate initiative led by OpenAI, Oracle, and other partners, and is being designed for ultra-high-density AI workloads. Vantage plans to invest more than $25 billion in the project, which is expected to be fully built out by 2028.

Construction of the center began in late 2025.

xAI Colossus 2 — Tennessee

Elon Musk's xAI is building Colossus 2 to power Grok, the AI chatbot on X. Construction began in 2025 in Memphis, Tennessee, and the company has said it plans to expand the site to 1.5 gigawatts. The nonprofit research institute Epoch AI estimates its current capacity at about 946 megawatts.

Environmental justice groups, led by the NAACP, sued xAI earlier this year, alleging that the company installed and operated gas turbines that power Colossus 2 in nearby Southaven, Mississippi, without the required air permits. In March, Mississippi regulators approved a permit for xAI to build and operate a new fleet of turbines at the site.

Local residents have filed a class-action lawsuit against xAI, MZX Tech, and SpaceX, alleging that noise and vibrations from the gas turbines have disrupted their lives and harmed their property values.

Still, in late October, Business Insider reported that SpaceX was offering employees a 20% pay bump to temporarily relocate to Memphis or Southaven, Mississippi, for six weeks to help build out Colossus 1 and 2.

Project Jupiter — New Mexico

Developed by infrastructure companies STACK Infrastructure and BorderPlex Digital Assets, Project Jupiter will be an AI data center campus in Doña Ana County, New Mexico, near the Texas border. It is part of the Stargate initiative and is expected to support Oracle's AI-computing operations for OpenAI.

Construction for Project Jupiter began in September 2025. The campus's planned on-site power system could include up to 2.45 gigawatts of Bloom Energy fuel cells, replacing an earlier proposal centered on gas turbines and diesel generators.

In May, Oracle ditched its plan to construct a natural gas plant to power Project Jupiter, following local residents' concerns over the potential negative water usage and air pollution effects it could cause.

Oracle issued a force majeure notice in September due to potential delays in the project's power infrastructure, though it says the campus remains on track for 2028.

Meta Hyperion — Louisiana

Meta's Hyperion AI data center in rural northeast Louisiana, backed by more than $50 billion in investment, will span more than 3,600 acres and is planned to deliver 5 gigawatts of capacity once fully developed. Hyperion broke ground in December 2024.

Meta says Hyperion is designed to train and run its artificial-intelligence systems. Earlier this year, Louisiana gave Meta $3.3 billion in tax breaks for Meta to build Hyperion. The project is expected to be completed in the 2030s.

Since the center is located in a rural part of the state, the influx of 7,500 trade and construction workers has spurred a housing boom, with mobile homes built for workers that rent for $3,000 a month.

Still, the project has also faced pushback from local residents and environmental groups concerned about its impact on water, energy use, and the surrounding community.

Louisiana regulators held hearings in October on the proposed power build-out for the campus. The state's Public Service Commission will issue a decision in December.

PORTS-Pike Technology Campus — Ohio

The Department of Energy said SB Energy broke ground in March in Ohio on the planned AI and cloud campus at the former Portsmouth Gaseous Diffusion Plant, known as PORTS. The campus will span private land and remediated federal land leased from the DOE.

The Nvidia-backed project is expected to include up to 10 gigawatts of data-center capacity, with OpenAI set to use up to 8 gigawatts of it for its workloads.

While the DOE says the historical PORTS site is in active remediation, the project has raised concerns about potential health and environmental risks associated with the site's history as a uranium enrichment plant. SB Energy told the Guardian in August that it is "committed to paying for accelerated cleanup and remediation at the site" and that the company "performs thorough site reviews and due diligence for each of our infrastructure projects, including soil sampling."

SB Energy said in October that it would begin initial clearing and grading work for the campus’s first phase.

Project Matador — Texas

Fermi America's Project Matador, also called HyperGrid, is an AI data center and private power campus being built near Amarillo, Texas.

Fermi, which was co-founded in 2025 by former Texas Gov. and US Energy Secretary Rick Perry, said it is pursuing roughly 17 gigawatts of private power for the campus, up from the initially planned 11 gigawatts. That means the campus would generate much of its own electricity on-site rather than relying entirely on the public grid. The company says 6 gigawatts have been permitted so far.

Fermi has a 99-year land lease with the Texas Tech University System for the core site, and the project could span roughly 8,400 acres at full build-out. Fermi says construction is underway, with completion planned for the 2030s, and AI cloud company TensorWave has signed a 15-year lease expected to generate about $6.5 billion in contracted revenue for the first phase. Fermi is also developing a cooling system designed to substantially reduce water use.

The company was hit with a class action lawsuit earlier this year. The investor lawsuit alleges Fermi misled investors about the status and prospects of Project Matador; its shares fell after Amazon withdrew a $150 million construction advance in December 2025.

Fermi America said three gas turbines arrived at the site in October.

SpaceX, Meta, Google, Oracle, Fermi America, CloudBurst, and xAI did not immediately respond to a request for comment from Business Insider. SB Energy did not offer a comment.

Meta's AI gamble is suddenly looking like a win for Mark Zuckerberg.

That was far from a given. Even the Meta CEO was doubting himself. His company had poured hundreds of billions of dollars into AI investments with largely unproven results.

"Are you sure this is going to work?" he asked his 29-year-old AI chief, Alexandr Wang, at the end of last year, he recalled on a recent episode of the "Cheeky Pint" podcast.

Those risky AI bets — which one analyst compared to throwing spaghetti at the wall — as well as the massive layoffs and reorganization that came with them, are now starting to pay off. Meta beat its AI rivals to the punch with its new AI assistant, Muse, and it finally released a frontier AI model. Although Muse is bleeding cash amid high compute costs, analysts are excited by its moneymaking potential.

Zuckerberg is riding high, too. Sipping a Guinness alongside Wang during a recent podcast interview, the pair extolled Meta's AI prowess.

"You never know when you launch something how it's going to be received," the CEO said. "There's always the case where you get a home run, which in this case it was, right? And it's just like extremely well received right off the bat."

And Meta seems to believe in Zuckerberg's leadership more than ever. His latest coup: Convincing the company to issue a new class of shares after he threatened to list his Palo Alto compound and leave California as voters weigh a billionaire tax.

The big question is how long the good vibes will last. Meta is still spending gobs of money on AI without a concrete financial payoff. Some remain skeptical.

"The bill is huge," David Wagner, a portfolio manager at Aptus Capital Advisors, told Business Insider.

"He's won the narrative. He hasn't proven the return yet."

Meta's social reckoning era

Zuckerberg is no stranger to public discourse. He's undertaken multiple personas: the nerdy wunderkind in hoodies, the jacked wife guy, the unapologetically masculine Trump ally.

At the beginning of this year, it looked as if he was headed for a new reckoning, and it wasn't just because of the aptly named sequel to "The Social Network."

Zuckerberg was on trial, taking the stand in a landmark lawsuit that alleged Meta's platforms harmed the mental health of young people. (Meta was found negligent and is appealing.) It was one of three major lawsuits this year that put the company on the defensive against claims that its technology posed safety risks to children.

And there were the more typical Zuckerberg cringe moments, too, like his front-row appearance at a Prada fashion show that raised eyebrows across the internet.

On the inside, employee morale was waning. Though the billionaire CEO had called 2026 "the year of AI," Meta had little momentum, and unpopular decisions — including a program to track employee keystrokes — amplified the discontent.

Also, Meta had one of its largest rounds of layoffs, ever. Meta began trimming staff in March, followed by a huge round in May, when it laid off about 10% of its workforce.

Separately, Meta reshuffled another 7,000 workers as part of an AI-focused reorganization of its tech and product teams. Many engineers were relegated to individual contributor status. Meta later asked some of the employees it had demoted to return to their managerial roles.

The AI push didn't seem to be paying off: In the second quarter of this year, 98% of revenue came from its traditional advertising business — not AI.

Morale was "maybe not the worst it's ever been in 20 years here, but it's probably up there," Meta's chief technology officer, Andrew "Boz" Bosworth, said during an internal call in June.

The groundwork for Meta's comeback

Amid the chaos, Zuckerberg laid the groundwork for a comeback. Wang and Meta's elite AI researchers moved their desks to sit near Zuckerberg.

Starting last year, Zuckerberg and Wang also went on a talent-hiring spree, dangling pay packages worth hundreds of millions of dollars to poach top researchers.

Zuckerberg empowered key lieutenants, including former GitHub CEO Nat Friedman, to go ham on any vision they had, including on his pet project, an AI assistant. It all fit into Zuckerberg's longtime playbook of embracing small teams and overseeing reorgs until something works — move fast and break things, as he has infamously put it.

It worked.

Muse, the long-awaited Meta AI agent, became a breakout hit. A week after it launched, it shot to the top of the app store, and has since achieved more than 10 million downloads in North America. The team was stunned to have beaten OpenAI to market, a person familiar with the matter said. OpenAI's rival product, Dots, launched three weeks later.

Meta also threw unprecedented compute and talent at building a series of frontier AI models, called Muse Wonder, that could keep up with those from giants like OpenAI and Anthropic.

On Meta's third-quarter earnings call, Zuckerberg was optimistic that Meta's AI and hardware investments could soon become cash cows, as people shop through Muse and buy more virtual-reality glasses powered by the company's growing AI chops.

Meta, meanwhile, views Zuckerberg as more critical than ever to its long-term fortunes.

Key board members abruptly excused themselves from a dinner with Zuckerberg in September after he told them he was going to sell his Palo Alto estate in response to California's proposed billionaire tax, Meta disclosed in a filing on Wednesday.

Four days later, a special committee returned with a deal that would give Zuckerberg $20 billion in "immediate liquidity" to cover the tax — if it passes — without reducing his controlling stake in Meta. Meta also made plans to ensure that billions in fresh AI fundraising wouldn't dilute Zuckerberg's control.

"Meta's track record of significant value creation for its shareholders is in large part due to his strong leadership, unique and creative vision, and exceptional management abilities," the company said in a press release.

The personal win comes as some of his other problems have blown over. Meta settled a major lawsuit brought by attorneys general across states, Washington, DC, and US territories. As part of the settlement, the company committed to make changes to its apps, such as daily time limits and pausing notifications overnight. The agreement was part of an August settlement, a move that allowed Meta to call out its competitors such as YouTube and TikTok.

"All platforms should empower parents and support teens by putting the same measures in place, because we know that when teens are restricted on one app, they simply move to another," the company said in a blog post at the time.

Big questions still loom. The day after Meta reported strong earnings and Muse adoption, its stock fell 7% as investors absorbed the scale of the company's hundreds of billions of dollars in future spending commitments.

And the legal troubles keep on coming. This week, a New Mexico jury also ordered Meta to pay a staggering $25 billion over the Cambridge Analytica scandal, a case that Meta plans to appeal.

Whether Zuckerberg's strategy is sustainable or not, the vibe shift has been notable — and may have proven naysayers wrong, at least until the next period of chaos erupts.

"Meta is cool again," Jesse Cohen, an analyst at Investing.com, said in a note. "That's the Muse effect."

OPERAZIONE REPORT: Tutte le carte, gli interrogativi e i sospetti sul caso Ranucci-Lavitola

Il direttore anticipa i temi del nuovo numero di MillenniuM, disponibile in libreria e negli store online.

Una trasmissione d’inchiesta invisa al potere. Una bomba cosiddetta “amica”. Un faccendiere specializzato in torbidi servizi alla politica. Partiti che invadono la tv pubblica. Sono gli ingredienti di quella che abbiamo chiamato “Operazione Report”, di cui conosciamo in parte lo svolgimento e in tutto l’esito: quello di allontanare il giornalista Sigfrido Ranucci dalla conduzione della più importante trasmissione di giornalismo investigativo della Rai, dopo nove anni di conduzione (salvo il successo dei ricorsi dell’interessato al Tribunale del lavoro di Roma). Nove anni in cui molte puntate di Report hanno scatenato gli attacchi della politica, in particolare dal centrodestra e dalla frangia renziana. Che si sono via via trasformati in una vera e propria campagna, amplificata dalla stampa di destra. Parte da qui “Operazione Report”, titolo di copertina del numero di ottobre di MillenniuM, il mensile diretto da Peter Gomez. Il racconto prende il via dall’altro protagonista, il faccendiere Valter Lavitola, con il ritratto approfondito e documentato a firma di Marco Lillo. Sul personaggio i media tendono a far prevalere aspetti quasi folcloristici, ma la sua biografia racconta ben altro: relazioni coi potenti, processi e condanne, intrighi politici ai massimi livelli, che hanno cambiato la storia di questo Paese, dalla caduta del governo Prodi nel 2008 alle manovre contro politici sgraditi. Una parabola in cui affiorano contatti stretti con i servizi segreti, un leitmotiv della sua fulminante ascesa, come scrive il direttore Peter Gomez. Sulla “Operazione Report” Millennium parte dai fatti, vale a dire l’inchiesta giudiziaria sulla bomba esplosa davanti a casa di Ranucci, a Torvaianica, il 16 ottobre 2025: l’individuazione dell’arruffata manovalanza campana, l’improvvisa entrata in scena di Lavitola, la “amicizia fraterna” rivendicata a sorpresa dal conduttore, la sequenza di “affettuosi” moventi snocciolati dal faccendiere. Ma oltre le carte c’è di più: gli affari, africani e non solo, che portano a Marco Dell’Utri, figlio del senatore che è stato la mente organizzativa di Forza Italia e ha scontato una condanna a 7 anni per concorso esterno in associazione mafiosa. L’Operazione Report ci fa riflettere sul rapporto tra informazione e potere, ma anche sui fili trasversali del potere. Che possono passare anche per un piatto di vongole. Sgusciate o meno.

L'articolo Millennium, “Operazione Report”: il nuovo numero del mensile analizza il caso Ranucci-Lavitola proviene da Il Fatto Quotidiano.