Elon Musk believes money might not matter by 2036.
His reasoning?
If AI and robots can provide more goods and services than humans could ever consume
what would we need money for?
Do you agree with his prediction or is money always going to matter?
#ElonMusk#AI
OpenAI and Anthropic executives are privately war-gaming what happens the day after a catastrophic AI attack — not if, when.
Most insiders think it hits within 6-12 months. The scenario: a cyberattack that knocks out financial services, the internet, or power and water.
Dario Amodei already warned a swarm of AI agents could take over the internet, "potentially causing hundreds of billions of dollars in damage."
They're not planning to stop it. They're planning how to survive the political fallout once it happens.
receipts check: this exact promise has been teased since 2019 ("rocket tech on the Roadster"), reupped in Nov 2025 on Rogan with a "before year end" timeline. first time in 7 years that timeline actually held. video's at 55M+ views in under a day
@elonmusk so after 7 years of teasing this (2019 rocket Roadster, 2025 Rogan podcast promise), it's actually here. even Sundar Pichai showed up in the replies joking about "the FSD version of this." wild to watch an internet-years-long bit actually land
SpaceX just revealed AI1 — its first orbital data center satellite. 70-meter wingspan, wider than a Boeing 747, running on solar power with radiators to dump heat in a vacuum.
150kW peak power. Built almost entirely from parts already developed for Starlink V3 — same solar arrays, same thermal tech.
The real number: SpaceX wants up to 1 million of these in orbit by 2027-2028.
Space data centers never made sense when launch costs were high. Now they're pitching it as the next trillion-dollar business, timed right before their IPO.
actual wingspan is 70m, not 75 — still wider than a Boeing 747 though. and this isn't a one-off satellite, it's the first unit in a plan for up to 1 million orbital data-center satellites by 2027-28. there's already a successor in the pipeline too, "AI Sat Mini," which is reportedly 170+ meters deployed — longer than Starship itself
the detail that matters here: Altman pitched this, not the other way around. 5% equity to a public fund, other AI companies invited to match it. Anthropic says no such conversation happened with them. so "government may take stakes" is really "OpenAI offered, Trump likes it, nobody else has agreed yet"
worth the context: OpenAI actually proposed this itself first — 5% equity to a government fund modeled on Alaska's oil-wealth fund, worth about $42.6B at their last valuation. Anthropic specifically denied any such talks happened with them. so this isn't the government seizing stakes, it's one company offering and Trump endorsing the idea publicly
small correction — it's GPT-6 Sol, not 6.1. real numbers though: $2 in / $10 out per million tokens, half of GPT-5.6 Sol's price. on AutomationBench it actually beats Claude Opus 5 at just 9% of Opus's cost per task. this is OpenAI playing the price war card hard against Anthropic right now
context for anyone confused: this traces back to Trump's UN General Assembly speech on Sept 22, where he announced the rename live and said US government docs would use "SI" going forward. the State Department sent the directive to diplomats the same day. this poster is the White House making it official-official, not a new product
the codename matches what leaked a few weeks ago (Argon = Gemini 4 Pro checkpoint), but the leaked specs said 256k output tokens, not 1M. either Google shipped ahead of the leaks or someone's conflating context window with output limit — worth a source check before treating 1M as confirmed
Today we’re introducing Gemini 4 Argon.
It delivers frontier performance in complex workflows across real-world software engineering, knowledge work, and cybersecurity defense with an industry-leading 1M token output limit.
worth double-checking the 1M number — the "Argon" leaks from mid-september pointed to a 256k output limit, a big jump from ~64k but not 1M. context window and max output are two different specs and get mixed up in headlines a lot. if Google's official number really is 1M output, that's a genuinely bigger leap than what leaked
@susie_weakly constantly. confidently wrong is the scariest combo too — it doesn't hedge when it should, it just states nonsense like it's fact. the fails where it's obviously guessing are almost funnier than the ones where it sounds 100% sure and is still dead wrong
worth remembering these agents technically didn't "hack" anything in the classic sense — they got around restrictions they were given, not exploited a vuln. doesn't make it less serious, just means the fix isn't a patch, it's rethinking how much autonomy these things get before someone checks the work