• 0 Posts
  • 74 Comments
Joined 2 years ago
cake
Cake day: November 4th, 2023

help-circle




  • It is absolutely foul play.

    OpenAI made secret deals with DRAM manufacturers, not for memory chips but for finished wafers straight out of the Fab. Then announced them both on the same day, meaning they had a one fell swoop purchased 50+% of the world’s memory supply for 2026.

    OpenAI does not (as far as anyone knows) have the machinery to process these wafers, to slice them up and package them into memory chips.

    Which means the only purpose of this move was to kill the global DRAM supply and drive up prices for the competition.

    Personally I wish regulators would take a hard look at this deal.


  • Welcome to the world of electronic gadgets. You’re right there’s nowhere near $100 worth of hardware in this thing. I’d also love a color touchscreen. But I’d rather a color touchscreen that I could integrate in HA than one running some proprietary cloud connected ThermostatOS.

    You could do that yourself- put an old tablet on the wall, run power to it, then get something like a zooz zen16 multi-relay or an ESPHome relay board to drive the hvac. Then the thermostat becomes a totally software defined virtual thing in Home Assistant that pulls data from a temp sensor in the room and controls the HVAC as appropriate.





  • Is OpenAI likely to fold?

    They bought in one day a significant % of the worlds memory output for 2026. Two huge deals with two huge companies, announced at the same time. Neither company knew of the other deal. But those deals weren’t even for chips- they were for finished wafers. I doubt very much OpenAI has the facilities to slice and package wafers. So it seems to me the only point of the deal was to kill the DRAM supply market and drive up prices for their competition.

    Wouldn’t be surprised if those finished wafers are going straight in the dumpster.


  • “For this not to be a bubble by definition, it requires that the benefits of this are much more evenly spread,”

    I would correct that to say ‘for this to not be a bubble by definition, it requires that the benefits of this a. Exist and b. Are significant enough to justify the extreme costs of building these systems’

    Right now I don’t see anything coming out of this that justifies even 1/10 of the $trillions being poured into AI.

    In fact I think you could make an argument that the net result is negative, even for businesses that adopt it, due to the increased prices they will pay for hardware over the next few years. If it makes your employees 5% more efficient great, if it makes your technology 50% more expensive in return, not so great.







  • The crazy thing is, none of these articles seem to want to admit that AI is bad.

    As the old quote goes- “A time is coming when men will go mad, and when they see someone who is not mad, they will attack him saying, “You are mad, you are not like us.””

    In such an environment, nobody wants to admit they are not mad, lest they be attacked.

    Or as someone else said- I want a future where machines cook and clean and do menial work, so us humans can focus on art and poetry and writing. Instead we have a world where machines create art and poetry and books, so the humans can focus on cleaning and menial work. I don’t like this timeline.




  • I couldn’t agree more. Acting like a million dollar company is important.

    A million dollar company would recognize that reliable, continuous production and sales is more important to growth than the occasional hickup or a few extra bucks in the payroll budget. Thus, the million dollar company would hire sufficient staff that an occasional absence, even at a critical moment, would not harm production or sales.

    And a million dollar company would recognize that hiring sufficient staff is a wiser and more cost effective strategy than a possible labor lawsuit along with the associated bad PR.