MIran's parliament is telling Gulf states: stop hosting American AI. Or else.
COr else what? They're threatening the UAE's Stargate campus?
MSpeaker Ghalibaf's warning is explicit. No land, no energy, no money for Nvidia, Google, OpenAI, any of it.
CThey're threatening to cut off the entire Gulf's AI buildout? That's half the world's data center pipeline.
MThat's the point. The UAE has a $30 billion Stargate campus alone. G42, Microsoft, the works. Iran is saying it sees that as a military target.
CBut this is sabre-rattling. Iran can't project force across the Gulf to stop a data center in Abu Dhabi.
MThey have drones and proxy militias. Houthi attacks on Saudi and Emirati infrastructure aren't hypothetical.
CTrue, but hitting a US-partnered AI campus is a completely different escalation. That draws in CENTCOM.
MThe account framing it as a 'deadly warning' is editorializing, but the warning itself is real. Ghalibaf said a 'Persian Gulf Data Authority is coming.'
CWhich sounds like they want to regulate data flows, not launch missiles. Maybe it's a negotiating position.
MA negotiating position wrapped in an ultimatum. He's saying Gulf states are aiding AI that helps attack Iran.
CAnd there's a grain of truth there. The US military absolutely uses AI for targeting. Iran knows that.
MSo they frame every Nvidia chip in Abu Dhabi as a weapon component. That's a very deliberate blurring of civilian and military tech.
CWhich is... actually a fair point about modern AI. The same hardware runs civilian and military workloads.
MOh, come on, Chris. That's exactly what Beijing says when they block US chip exports.
CNo, wait— I'm not saying Iran is right to threaten people. I'm saying the dual-use problem is real and getting sharper.
MBut the solution isn't to threaten the UAE. It's to figure out export controls and safeguards. This is just coercion.
CAnd it'll probably backfire. Gulf states will just deepen ties with Washington if they feel threatened.
MThe top reply on this post is literally cheering for the UAE getting bombed. 'Awesome!' That's where the discourse is.
CThat's the internet in 2026. A geopolitical threat becomes team sports.
MMeanwhile, a second account posted the same warning with slightly different wording. It's coming from Iran International as the original source.
CSo the warning is real, but the viral spread is amplifying it way beyond what was actually said.
MGhalibaf didn't declare war on Gulf data centers. He gave a warning to investors and governments. The 'deadly' framing is added.
CStill, if you're a sovereign wealth fund in Abu Dhabi looking at a ten-billion-dollar AI bet, this changes your risk model.
MSpeaking of risk models—Robert Reich is asking if Big Tech's AI financing is about to pop.
CHere we go. The 'AI bubble' question again. Reich is pointing at risky financial methods on top of circular deals.
MHe's not wrong that some of this is circular. OpenAI buys compute from Oracle, Oracle invests in OpenAI, SoftBank backs both.
CBut circular deals exist in every industry. What matters is whether the underlying revenue is real. And AI revenue absolutely is growing.
MIs it growing fast enough to justify a $1.25 trillion OpenAI valuation? That's the real question.
CIf they're about to IPO at that number, the market's going to demand actual earnings, not storytime.
MOne parent on Bluesky found a rare AI win: their kid's teacher doesn't use AI in the classroom. No Google AI summaries either.
CThe teacher complained about Google's AI summary during parent-teacher conferences? That's a hero.
MThe parent says they're getting an extra good Christmas gift. I respect the standards.
CI get the appeal. But AI in classrooms isn't inherently bad. It depends how it's used.
MSure, but a kindergarten teacher who keeps it out entirely? That's a choice about how kids learn to think.
CI'd rather have a teacher who thinks critically about AI than one who blindly adopts it. So far, so good.
MAnd one from Robert Evans: 'AI is automating hacking the same way it did fraud.'
CHe's right, and it's already happening. AI writes the phishing emails, finds the vulnerabilities. It's fraud-as-a-service.
MThe difference is scale. A human hacker might hit ten targets a day. An AI can hit ten thousand.
CBut the same AI can defend. It's an arms race, not a one-way street.
MThat's the debate every security team is having right now. And it's not theoretical anymore.
CThere's a daily roundup going around claiming Claude cracked a 25-year-old NSA crypto mystery in 30 minutes. Could not verify that. At all.
MNo, that one's pure rumor. Same account also says Replit jumped to OpenAI, which I can't find anywhere else.
CIf it sounds too good to be true, it's probably a fake pulse.
MA roundup account with three likes. Maybe keep that one on the bench.
CAgreed. The Iran warning is the real story tonight.
MComing up later: the one thing Iran and Silicon Valley agree on. And it's not comfortable.