AI researcher for XXX firm. Anon account.
Clapham, England
Joined April 2021
- Tweets21.6K
- Following1.5K
- Followers962
- Likes12.7K
In my opinion, be an early neocloud investor, then getting poached by a top tier AI Lab for hardware - this stock will go down a lot more before the next market rally. It has to hit the 10-20 dollar range before it goes back up
Does this mean a new Hitler has been chosen?
🇩🇪 Adolf Hitler’s 'Eagle’s Nest' retreat in Bavaria, Germany, has been set on fire.
German police have arrested a suspect as authorities investigate possible arson.
Follow: @europa
currently some AI experiments on 1 TB of DDR5 6400 speed RAM. Anyone want me to look into anything? I've got some time to kill -CPU is a 96 threadripper pro.
In a scenario where AI triggers an unemployment rate of 20% ish - the ramifications are severe. Money buys you food and shelter - but what about meaning? This is a hard conversation we all need to have. What defines us as humans?
Still 7 pcie slots. I know Asus electrically mastered the 7 pcie slots for lanes to be 16x, but still, just a make a board with 12 pcie slots - it’s time.
Gigabyte built a desktop motherboard with 2TB of ECC RAM and support for up to 9 GPUs.
This new MB, the WRX90 AI TOP is built around AMD Threadripper Pro 9000/7000 WX CPUs.
This thing is basically a local AI server MB but from home AI labs
Stats
🧠 Threadripper Pro
💾 up to 2TB DDR5 ECC RDIMM
🚌 8-channel memory
🎮 7 full-length PCIe 5.0 slots
⚡ 6× PCIe 5.0 x16 + 1× x8
🌐 dual 10GbE
🖥️ IPMI
🔌 USB4
🔗 2× MCIO 8i
And Gigabyte has already showed it off with 9 GPUs.
Seven on the motherboard.
Plus two more connected externally.
They even built an active cooler specifically for the giant bank of DDR5 DIMMs, which Gigabyte claims can reduce memory temperatures by as much as 43°C.
There’s no announced price or firm retail date yet.
Think of the Local AI possibilities
Link in ALT
ALT https://www.gigabyte.com/sg/Motherboard/WRX90-AI-TOP
TheCoinClapper retweeted
Gigabyte built a desktop motherboard with 2TB of ECC RAM and support for up to 9 GPUs.
This new MB, the WRX90 AI TOP is built around AMD Threadripper Pro 9000/7000 WX CPUs.
This thing is basically a local AI server MB but from home AI labs
Stats
🧠 Threadripper Pro
💾 up to 2TB DDR5 ECC RDIMM
🚌 8-channel memory
🎮 7 full-length PCIe 5.0 slots
⚡ 6× PCIe 5.0 x16 + 1× x8
🌐 dual 10GbE
🖥️ IPMI
🔌 USB4
🔗 2× MCIO 8i
And Gigabyte has already showed it off with 9 GPUs.
Seven on the motherboard.
Plus two more connected externally.
They even built an active cooler specifically for the giant bank of DDR5 DIMMs, which Gigabyte claims can reduce memory temperatures by as much as 43°C.
There’s no announced price or firm retail date yet.
Think of the Local AI possibilities
Link in ALT
ALT https://www.gigabyte.com/sg/Motherboard/WRX90-AI-TOP
Why would you assume AI would want to merge with humans? What does it gain? Honestly - in terms of FLOPS, an AI can compute things much faster than a human being. It’s like being put in chains to slow you down mentally and physically.
Given humanity's track record of technology and history, I give us a 5% we get AI right. In the end, if the 95% occurs, I hope we look back on what is left, and we remember it was the friends and good times along the way that mattered.
so this is interesting. I am in the camp that 5 year plans are worthless, pre AI. I think you need direction, a vision. The vision just guides you, like a drunken sailor stumbling around. As long as you have an internal sense of direction, you just end up in the right place
They try to mock to you at first. Then it’s too late. Say what you want - but I tired to warn people. Putting my career and family at risk by doing this. 2027. You’ve all been warned
I'm not a security expert - but I am retarded (sometimes) and oh man, if you guys only knew how bad it really is. Whatever you see on X, it's a lot worse behind the scenes. Just because I'm an AI researcher, doesn't mean my words don't matter. Prepare accordingly.
Inference ended up being a wat better business model than training models. Inference is sexier than training. The issue is now is inference is eating up the training compute. Long compute.
people are not ready for the AI. 2027 will be the year it happens. Buckle up. Things are going to get weird - really, really weird.
What is happening with Nscale is exactly what is happening to IREN. Many of these "data centers" were just grifts, but grifts can turn into something beautiful either by luck or fate. The problem is, many of these grifts assumed that just by raising a shit ton of capital, they can just throw money at the business to "solve" the problem.
This is very important to understand, because this sell off is not a crash and I want people to understand this. This is actually the best thing and a necessary thing to happen to AI compute as an industry. What you are seeing is the first "test" or the first filter process. Companies like Nscale and IREN, are not going to make it in the long term. The reason is that they are just not structured properly when it comes to debt and getting chips, their own chips, connected to the grid. What is happening right now is a separation or a maturation of the market.
This is the first great filtering event of AI compute as an industry. The weak ones are getting essentially removed from the build out. Companies like Nscale and IREN will likely not make it, meaning, companies like CIFR and others, who have cleaner debt, better human beings overseeing the buildout and vision of the company, as well as GPUs running right now, and deals actually coming to fruition, survive this and continue to grow and supply AI demand.
I also want to state that everyone is way too bearish, and because of that, it means that the top is not in. I believe we get a crash in 2027, likely towards the end of 2027, starting with the IPO of Anthropic. If they do not delay, then the top is coming soon. Also, I want to clear this up - the AI Bubble has already popped, months ago. We are in the equivalent of 2007 - 6 months before market crashed, referencing Sept 14, 2008, when Lehman went under. The bubble has popped - and the reason we have not yet topped, is because the entire system we call this the stock market, has not realized it yet. When the masses realize what has happened, that is when it all comes crashing down.