Well lets get a bit technical. He is the CEO of the largest AI chip producing company in the world. For him to say AI doesn't have a future would be to shoot himself in the foot.
The reality is that AI's bubble is about to pop just like the .com bubble. AI is completely overbought right now.
You’re missing the bigger picture with SpaceX.
It’s not just about rockets anymore. Sure, the launch business is crushing it with Starship making reusability the new standard, driving costs way down and opening up frequent access to orbit like never before. But the real driver for its massive valuation potential is Starlink, which already has millions of subscribers, billions in growing revenue, and is becoming the essential backbone for worldwide connectivity.
What most people aren’t seeing yet is the AI space datacenters angle. SpaceX is gearing up for orbital data centers powered by massive constellations of satellites with solar energy in space, sidestepping Earth’s power and cooling limits for AI compute. Starship’s heavy lift capability makes this scalable in ways ground-based setups can’t match. This ties straight into their long-term vision for powering the next wave of AI everything.
Elon’s leadership is the secret sauce here. He cuts through bureaucracy, pushes first-principles thinking, and drives insanely fast innovation and iteration. Just look at how Grok from xAI has grown so quickly even entering the AI race late, climbing benchmarks, attracting users, and advancing fast in reasoning and capabilities. That’s the same mindset fueling everything else.
Zoom out and the synergies become obvious: Starlink connects robots, vehicles, and remote operations. xAI provides the intelligence layer. Tesla builds the cars, Optimus robots, and the huge energy storage and grid solutions to power it all. The Boring Company adds underground infrastructure. It’s not separate companies, it’s one vertically integrated ecosystem for AI, robotics, energy, and becoming multiplanetary.
Put it all together and this is positioned to be the most valuable company on the planet by a wide margin. The growth isn’t linear, it’s compounding at a pace that leaves siloed, bureaucracy-heavy competitors in the dust. SpaceX heading toward IPO isn’t overhyped, it’s betting on the core infrastructure for the next industrial revolution. The momentum is unstoppable. 🚀
Stop and think… The largest IPO ever is launching next week… In order to buy in, you need liquidity… Maybe this is just a market shift for SpaceX buyin… I dont actually think its doom and gloom at all. Next week you will see the money thats probably just sitting in money markets now flood back into the mix…
@Autonomicro@davideciffa As a network engineer, I like being able to set something up that is just mine... Letting someone else do my computing does not give me the same satisfaction.
Ohh absolutely, my 3090 is 3 times faster, but when I am doing creative things, I run alot of workflows / LTX / stable diffusion etc. but I use Ollama geneation that helps throughout my workflow, and I have to load Gemma... Currently I have to unload checkpoints in order to load Gemma because I am FORCED to keep it on NVRAM... I realize that memory is 3 times slower than my 3090 GDDR, BUT its still 10 times faster than my normal 64GB system ram, which I have tried to load the LLM into system ram, and its dog slow... 10 times faster would be acceptable enough that I wouldn't have to worry about even running it on my 3090 build machine, and just dedicate that machine to pure Image / video generation.
I do alot of creative stuff in addition to programming (Program for work, creative for play). And honestly for the most part, I just use Grok anyways for coding... kind of hard to beat 1.7T parameters (or whatever there at now). For creativity / basic agentic stuff, Gemma-4 is actually more creative if she doesn't think...LOL so it works great for chat / creative writing / text 2 img type things. Either way.. It will be interesting to see, unified memory that is still 10 times faster than my current DDR makes at least possible to run the LLM in full system memory. Thanks for advice...
Basically I see this more of a workstation that I can use, have it run on there, and not have to deal with loading it on to my 3090's.. I know it will be about 3 times slower than my 3090's at that speed memory, but without reasoning, I don't think it would be too bad... My coding workhorse with reasoning, I would definitely not want to move off my 3090's
No shame to either of them. She is being showered with love and his money (presumably), and he is chasing his younger years. They are both consenting adults, each getting what they want. Who knows, they may actually be happy in love. Either way, no shame to either one regardless of the reasons behind their relationship.