@Evan_ss6@BigTrout300 Hopefully efficiency breakthroughs like deepSeek’s hybrid attention, minimax, nvidia’s latest work, mmamba-style SSMs, sparse KV cache tricks— are not real breakthroughs and won’t slash RAM needs by 70–80%+ - just hoping
@Tradermayne@pinedegen You can do min things like asking ChatGPT to pull the weekly buys against eth price and ask a 5 years old to read it - or even best just ask yourself how’s that possible given the weekly buys vs time spent > 3800$
@Evan_ss6 lol...i think it can be a good idea for you to take a drink with some CFO's / CEO's and stop reading GS / JPM or Mckinsey reports..happy to continue via DM if you want some IRL impacts
Inference demand may be high, but monetizable demand at full cost may be much lower than apparent usage under subsidy. That's all i'm saying and all i'm seeing at different corp levels and at budget discussions. Not saying AI is not a revolution i'm just saying it is costing us more than we though and we are looking for cheaper alternatives.
Seeing more and more posts like this and I’m also seeing it irl at my company - I won’t fade this. There is some shift undergoing. Tokens cost too much, it’s not sustainable and at the same time, hyperscalers can no longer subsidize tokens for ever. We are coming to the conclusion that the economic model of token consumption x price is hard from a budget / forecast perspective and we are taking some step back. There’s a corporate reality and there’s hype. I believe we will slowly start to see the impacts of the former and I won’t exclude a slowdown if corporate demand is not there. I’m long term AI bullish but starting to witness something new irl
Till last month my $40 @github co-pilot subscription could easily last the whole month. With new changes, 50% credits consumed in 4 days. #Vibecoding will fade at this pace and will severely impact many companies.
@Wild_Randomness would you be a QQQ calls buyer here ? At what maturity ? I’m wondering if I should buy some calls after the pull back or buy some protection to hedge QQQ during the next 10 days (SpaceX ipo) really curious on your view
NEW POST (Let The Bubble Wash Over You)
some thoughts on market structure and what the new financial regime looks like including:
> convective weather systems
> the reality of rolling bubbles
> how did we end up here
> what does it mean for the future
link to full piece below
Only one scenario saves bitcoin:native and $MSTR in the short-term.
Saylor has to come out and say (via Monday's 8K, or next Monday's 8K):
"I sold $4 bn of MSTR and BTC... we now have 2.5 years before we have to raise money again... dividends are covered for 2+ years, debt doesn't mature til sept 2028 (puttable in 2027) and that's easy to refinance via more converts".
If he does that, the market rips, and might even rip 20-30%. It once again makes MSTR uninteresting for years, but that's a good thing. STRC probably goes back close to $100, but he won't be able to sell more. And while capital markets might be closed to MSTR for awhile, it at least buys a ton of time, and in that time who knows what other catalysts might pop up.
He should have done this last week instead of that misguided $2.5mm teaser sale.
If he doesn't, and he continues to just wait it out (only has 5 months), or he sells tiny amounts as he goes (just enough to pay each monthly dividend), this selling won't stop
Markets telling us something.
Memory and alike, ripping higher as the war continued are pushing $QQQ down 4%+, $IWM -3%...
All while safe heavens like $GLD and $SLV are down massively today, -7% and -3%.
All while Oil $USO is moving lower back near lows...
And Trump is saying 'markets should be happy'..?
Perhaps the war with Iran is ending.
Otherwise why sell the safe heavens, why is Oil down massively, why is crypto pushing back higher despite $QQQ at lows...
Alternatively we are witnessing the market thinking the USA will attack Iran and we are heading into a world recession but in this case why is oil down today?