THIS GUY BOUGHT BOXES OF RETIRED NVIDIA TESLA CARDS TO BUILD A LOCAL AI SERVER FOR LESS THAN ONE MONTH OF SUBSCRIPTIONS
these are old datacenter GPUs like the P100, M40 and P4, cards that once cost thousands but now sell used for a fraction of their original price
the best target is a 24GB Tesla M40 for roughly $130, enough VRAM to run serious local models without paying Claude, ChatGPT and transcription fees every month
add a $25 fan shroud and a $10 power adapter, drop it into an old desktop, then run Ollama for private summaries, transcription and overnight AI jobs
the entire box lands near $165 while a $412 monthly AI stack burns $4,944 a year, turning retired server hardware into a machine that pays for itself in weeks
🚨 A hacker group announced at 3:00 AM: "We have locked all Starlink ground stations worldwide. Pay $500 million in Bitcoin, or the world goes offline!"
🚀 Elon Musk replied at 3:01 AM: "Check your firmware. While you were writing that ransom note, Neural-Auto-Patch released a micro-update that routed your entire operation through a dead-end satellite in deep orbit. Thanks for testing our hotfix."
Elon just personally bought a $1 billion gas turbine company, and no one announced it. No press release, no tweet. The deal only surfaced because a firm holding a 5% stake had to disclose its $50.4 million payout in an SEC filing. What he bought tells you where the real bottleneck in AI is.
APR Energy operates a fleet of mobile gas and diesel turbines totaling over 1 gigawatt. Their units arrive on trucks and can be delivered, installed, and commissioned in as little as a month. The fleet was built for blackout zones and disaster response in countries with unreliable grids.
Here's the constraint that makes it worth $1B to one man: Nvidia can deliver 100,000 GPUs in months. A new grid connection for a power plant spends a median of roughly 5 years in the interconnection queue. The chips depreciate while the paperwork sits.
Elon already lived this. xAI's first Memphis cluster ran 100,000 GPUs on about 150 megawatts, much of it from roughly 35 leased mobile turbines while grid power was pending. Environmental groups sued. The DOJ intervened to keep the turbines running. He was renting the most important input to his most important company.
So he bought the landlord. At $1B for 1+ gigawatts, he paid roughly $1 per watt of dispatchable power he can park anywhere. A gigawatt runs on the order of 600,000 H100-class GPUs.
Every AI lab can buy the same chips. Only one of them now owns a power plant fleet that ships by truck.
Tesla decommissioned the entire Model S/X Assembly Lines in just 46 days.
End of an era. Now, it’s time for Optimus. Once fully ramped, Fremont will produce 1M robots per year.
GOOD NEWS 🤖 Several sources deep inside the supply chain have indicated that Tesla has finally handed down concrete procurement targets for Optimus components 🔥
These guidelines push suppliers to ramp up production capacity to an incredible 1,000 units a week by September this year, and scale up even further to between 2,000 and 2,500 units a week by the end of the year 🆒
Two insiders close to the suppliers noted that Tesla normally sorts its orders about two months in advance, and they've already spotted specific requests for hundreds of units queued up for August. Looking at these estimates, Tesla's supply chain will soon have the sheer capacity to churn out components for a massive 100,000 Optimus robots a year by the end of December 😍
Word on the street is that before locking in these massive orders, Elon personally reviewed and signed off on the absolute latest build of Optimus during a high-level meeting in late June. This is huge—it means that after more than three years of hardcore R&D, Optimus Gen 3 is finally leaving the lab and stepping right into the mass production phase 💥
Both supply chain sources mentioned that during this meeting, Elon laid down the law, demanding the team hit these end-of-year production targets, or he would completely sack the entire Optimus procurement department. Classic Elon 😤
Honestly, this hardcore ultimatum gives the supply chain, the markets, and the wider robotics industry way more confidence than any safe, standard forecast ever could 👏
Smart pet technology company Fi has just launched the world's first dog collar with @Starlink.
Basically, you’ll know exactly where your dog is at all times anywhere in the U.S. It’s $199 + $189/year membership for new customers, and just a $299 flat fee for existing members.
More info:
• Up to 5 days of battery life: Operate multiple days on a single charge, supporting extended trips and remote exploration.
• T-Satellite with Starlink GPS: Track a dog's location nearly anywhere in the US, including areas entirely without cellular coverage.
• Dual Band High-accuracy GPS: Deliver precise location data continuously, whether connected via satellite, LTE, Wi-Fi, or Bluetooth.
• Automatic switching: Seamlessly connect to LTE when available and to T-Satellite with Starlink when cellular networks are out of reach — with no action required from the owner.
• Real-time location updates: Deliver accurate, continuous location data whether the dog is in a city park or miles from the nearest cell tower.
• Fi Callback: A proprietary training system based on sound and vibration that uses no static electric shock to bring dogs back to their owners on command.
• Fully waterproof: Tested in salt water and IP68/66K-rated
$CAT
We would be excellent traders.
'Why did Burry Short $CAT here?
Because it is in a clear parabolic move
It is not cheap
Margins are stretched
And it was chased
This will drop hard.
If Burry has signaled the top, then he'll close his short at $579'