Najprej pripeljejo zgodbo do maksimuma, sedaj pravijo, da bodo malo povlekli ročno, ne povedo pa kako... kake bučke.
Evropejci moramo imeti svoj sistem AI, sicer bomo čez 30 let eni navadni loleki. Hvala Lajdri et Co, da nas pelje v to zgodbo: loleki bomo.

za sirsi kontekst in brainstorming...

izabella kaminska (ki po vtisih zadnjih desetletij (yes we're that old) ni po zupi priplavala, ceprav izrecno rada raziskuje skrajni domet idej)

***

I think a lot of people still don’t understand that it was Deepseek that triggered this acute phase of hyper scaling.

By going open weight Deepseek commoditised AI ensuring the “there can be only one” race had no choice but to turn to capex spend instead.

It’s been a simple “who has the ways and means to moat up most expansively?” ever since. If you’re confused by the economics, it’s because the economics aren’t supposed to make sense. The irrationality of the infrastructure spend is the logic. We’re in the world of Highlander.

That means you win by forcing everyone else to go bankrupt. Potentially even yourself (safe in the knowledge that if you do Uncle Sam still inherits the legacy infrastructure.)

In that sense it’s no different from the USSR diverting the nation’s entire consumer budget into vast steel plants that no one else could ever afford to match on cost.

But if loss leading your way to AI domination through control of compute was the plan, last month’s Kimi moonshot disrupted everything. It’s like beating Russian steel with the invention of a Star Trek replicator.

That’s why I’d be surprised if the pace of debt issuance continues on at the projected pace. If anything, Leopold’s blow up implies a pivot towards greater capital efficiency is incoming.

Alternatively this is when “the project” begins in earnest.

Either way, I’d bet on more secrecy and even more centralisation.

The only question now is whether the USG takes control or not. Judging by Anthropic’s previous spat over Pentagon access, this is the point where the statecraft is gonna get complicated.

***

The whole point is the capex spend cycle is over. We were anticipating ongoing spending at the current rate. That is not gonna happen now. The market had priced in that liquidity as going to AI companies. It’s not gonna happen.

They’re presenting it as a safety story. “We will stop the rate of advancement of model design”. But this is a false narrative. The capex spend was about purposeful moating after ai model advancement got matched by deepseek at a fraction of the cost. Models became commoditised at that point. The legacy models still had an edge providing they could throw insane capital at scaling to reduce token cost on the usage side. Loss leading your way to market share. Same as Chinese did with solar. The tech was comparable to US but Chinese still took over the market because they were prepared to operate uneconomically until they had run everyone out of business.

In this case the US hyperscalers also had access to the equivalent of the rare earths that China did not (cutting edge chips).

Hence the tokenization price collapse.

But what happened with Kimi shows that the open weight models can keep up without losing market share even with less powerful chip infrastructure and less capital expenditure.

This kills the point of trying to win the race that way. It’s a waste of capital. And the liquidity was being totally sucked out of Western government bond markets.

So now that eases the rate of liquidity absorption.

The market meanwhile will have to write down their ai assets. To stabilise the market the government will takeover the ai firms. So it’s just a question of what rate does the private sector exchange now totally underwater AI paper for USG paper.

***

They need a credible off ramp from the hyper scaling narrative they’ve been selling for months. Safety concerns are an obvious way to justify the incoming pivot. They also justify taking the whole thing into a closed “Manhattan project” type setting at the government’s expense. This justifies the incoming lockdown.

As I predicted in August the scale narrative had run its course due to kimi. Hyper scaling wasn’t about genuine demand expectations. It was about winning the ai race by outspending and bankrupting your competitors. Kimi made that pointless.

This is no doubt the asymmetric information Bessent has been alluding to for weeks. There’s going to be a helluva lot of liquidity freed by this move. It’s all going to go flooding into USTs. Happy D day bond guys.

Torej so se odločili, da bodo denar uporabili drugje, kjer bo prinašal več in drugače. EU pa, kot ponavadi, kasni in zaostaja potem še več.

[]
Najprej pripeljejo zgodbo do maksimuma, sedaj pravijo, da bodo malo povlekli ročno, ne povedo pa kako... kake bučke.
Evropejci moramo imeti svoj sistem AI, sicer bomo čez 30 let eni navadni loleki. Hvala Lajdri et Co, da nas pelje v to zgodbo: loleki bomo.

Loleki smo že zaradi zmešanih levičarskih fašistov, ki peljejo človeštvo v "woke" pogubo!

[]
za sirsi kontekst in brainstorming...

izabella kaminska (ki po vtisih zadnjih desetletij (yes we're that old) ni po zupi priplavala, ceprav izrecno rada raziskuje skrajni domet idej)

***

I think a lot of people still don’t understand that it was Deepseek that triggered this acute phase of hyper scaling.

By going open weight Deepseek commoditised AI ensuring the “there can be only one” race had no choice but to turn to capex spend instead.

It’s been a simple “who has the ways and means to moat up most expansively?” ever since. If you’re confused by the economics, it’s because the economics aren’t supposed to make sense. The irrationality of the infrastructure spend is the logic. We’re in the world of Highlander.

That means you win by forcing everyone else to go bankrupt. Potentially even yourself (safe in the knowledge that if you do Uncle Sam still inherits the legacy infrastructure.)

In that sense it’s no different from the USSR diverting the nation’s entire consumer budget into vast steel plants that no one else could ever afford to match on cost.

But if loss leading your way to AI domination through control of compute was the plan, last month’s Kimi moonshot disrupted everything. It’s like beating Russian steel with the invention of a Star Trek replicator.

That’s why I’d be surprised if the pace of debt issuance continues on at the projected pace. If anything, Leopold’s blow up implies a pivot towards greater capital efficiency is incoming.

Alternatively this is when “the project” begins in earnest.

Either way, I’d bet on more secrecy and even more centralisation.

The only question now is whether the USG takes control or not. Judging by Anthropic’s previous spat over Pentagon access, this is the point where the statecraft is gonna get complicated.

***

The whole point is the capex spend cycle is over. We were anticipating ongoing spending at the current rate. That is not gonna happen now. The market had priced in that liquidity as going to AI companies. It’s not gonna happen.

They’re presenting it as a safety story. “We will stop the rate of advancement of model design”. But this is a false narrative. The capex spend was about purposeful moating after ai model advancement got matched by deepseek at a fraction of the cost. Models became commoditised at that point. The legacy models still had an edge providing they could throw insane capital at scaling to reduce token cost on the usage side. Loss leading your way to market share. Same as Chinese did with solar. The tech was comparable to US but Chinese still took over the market because they were prepared to operate uneconomically until they had run everyone out of business.

In this case the US hyperscalers also had access to the equivalent of the rare earths that China did not (cutting edge chips).

Hence the tokenization price collapse.

But what happened with Kimi shows that the open weight models can keep up without losing market share even with less powerful chip infrastructure and less capital expenditure.

This kills the point of trying to win the race that way. It’s a waste of capital. And the liquidity was being totally sucked out of Western government bond markets.

So now that eases the rate of liquidity absorption.

The market meanwhile will have to write down their ai assets. To stabilise the market the government will takeover the ai firms. So it’s just a question of what rate does the private sector exchange now totally underwater AI paper for USG paper.

***

They need a credible off ramp from the hyper scaling narrative they’ve been selling for months. Safety concerns are an obvious way to justify the incoming pivot. They also justify taking the whole thing into a closed “Manhattan project” type setting at the government’s expense. This justifies the incoming lockdown.

As I predicted in August the scale narrative had run its course due to kimi. Hyper scaling wasn’t about genuine demand expectations. It was about winning the ai race by outspending and bankrupting your competitors. Kimi made that pointless.

This is no doubt the asymmetric information Bessent has been alluding to for weeks. There’s going to be a helluva lot of liquidity freed by this move. It’s all going to go flooding into USTs. Happy D day bond guys.

Lahko se tvoj povzetek , mnenje?



Moje razumevanje teksta.
Teta pravi, da se je nehalo ai financiranje in bo slo veliko ai podjetij prek propada pod marelo drzave.
Ali je bilo ai financiranja res toliko?
Da je drasticno vplivalo na prodajo ameriskega dolga.
Sam sem slisal o par sto miljardah.
Po moje je to le del manjsega povorasevanja..drugi razlogi so se,
Lucudni trump, dolg brez razloga, yield control kratkorocnih in ne dolgorocnih obveznosti, boljsa donosnost japonskih obveznic, vecletno seljenje kapitala iz obveznic v delnice se pravi nezaupanje v drzave, podobno v kripto in zlato.
Tezko verjamem, da bi propad kakih ai podjetij prinesel hitro vec povprasevanja po ameriskem dolgu.

Sam moje laicno mnenje,pogled ;).
Po prvem branju tvojega citiranega besedila.

[odvisnik]
Moje razumevanje teksta.
Teta pravi, da se je nehalo ai financiranje

da se bo, ker ni vec realisticno niti glede ekonomike podjetij niti glede odnosa trga in postaja tudi sistemsko motece.

in bo slo veliko ai podjetij prek propada pod marelo drzave.

to bi bil skrajni domet tega procesa, ja - torej, kako dalec gre pojem "pod marelo". a gre vse do preteznih lastniskih delezev ali na primer samo do neke hardcore regulacije tempa in dometa razvoja. ob zavedanju, da je to seveda nateg, ker je vsakrsna ostra regulacija z visokimi zahtevami tipicno predvsem utrjevanje polozaja obstojecih igralcev.

Ali je bilo ai financiranja res toliko?
Da je drasticno vplivalo na prodajo ameriskega dolga.

po mojem razumevanju v letosnjem letu 35 do 40% velikosti emisije drzavnega dolga.

Po moje je to le del manjsega povorasevanja..drugi razlogi so se,
Lucudni trump, dolg brez razloga, yield control kratkorocnih in ne dolgorocnih obveznosti, boljsa donosnost japonskih obveznic, vecletno seljenje kapitala iz obveznic v delnice se pravi nezaupanje v drzave, podobno v kripto in zlato.

absolutno so tudi drugi dobri razlogi za drazje zadolzevanje drzave. tud pod razno jih ne manjka. vendar vsak, ki ga lahko omejis ali odpravis, magari samo zato, ker imas nad njim vec kontrole kot nad drugimi, steje. vsaka taka poteza je vec proste likvidnosti, ki jo lahko zakladnica pritegne k sebi. obstaja seveda nek poudarek na "lahko", dopustni in zadostni pogoj ni ista stvar, vendar nekje pac zacnes.

po moje trenutno gledamo tekmovanje med franckom piromanckom trumpom in gasilcem bessentom, ali lahko en vec zakuri ali drugi vec pogasi. oba imata svoje mocne plati :)

osebno sem dolgorocno skepticen glede ameriskih obveznic (poleg drugih stvari) in imam poudarke portfelja cisto drugje. vendar v tem izinem razmisljanju za moje pojme dobro skupaj zleti vec silnic, ki tudi po moje obstajajo in je v tem smislu brez veze ne riskirati stave na TLT v obliki opcij, na primer. (sploh po tem, ko so levicarski komunajzarji odpravili noro obdavcitev opcij, ki jo je uvedla desna, kapitalu prijazna vlada.)

ti si kakor razumem zlasti tehnicar, pa si poglej dolgorocni graf TLT.

(lahko pa pride soros iz penzije za en zadnji ples, kot film space cowboys, in razjebe celo zgodbo :)

politicno tricky del je, da je tako velika transformacija ai zgodbe tezko brez posledic za tecaje delnic, gospodinjstva so zgodovinsko visoko investirana, volitve pa blizu. lahko da to pomeni, da se premiki zacnejo sele po novembru. dokler pridemo do tja, so pa mozna tudi cisto politicna stopnjevanja na podlagi trumpove make or break potrebe ne biti impeachan, ki povecujejo negotovost in sirijo razpon, kaj vse je dobro vracunati kot mozno, tako da o tom potom...

AI je že sprivatiziran (beri: pod kontrolo velikih firm ali vlagateljev in IT pogodbenikov) in ker za sedaj buri duhove, so malo umirili žogico.

Open AI, Anthropic it, ne potrebujejo borze, so že posredno obvladovane s strani mlrd vlagateljev in hi-tech firm, Gemini pa je itak last Googlea...

Zakaj bi jo dajali v roke državi, če pa ji lahko dajo na posodo usluge? USA kot država je še Potemkinova vas, če ne bo imela v rokah AI. In Trump je naredil vse, kar je lahko, da bo AI ostal privat last. Torej v rokah tiste klike, ki ji je ful "mar" za Palestince... ali nasploh človekove pravice.

Usa so (že) ugrabljena dežela. Vsi nastavki so že tu, z obveznicami vred. Trump le pospešuje ta proces in njegva familja bo izdatno nagrajena. Razkroj kapitalizma, demokracije in pravne države at its best.

Ampak tle res eni proizvajajo bučke za izvoz...

Kaj pa EU? Američani in Angleži delajo vse na tem, da nam to "demokracijo" uvozijo še nam. Renska skupina z Draghijem je že dober primer... en Merz pa krasno sodi v to kategorijo.

"The craziest example of this is the "catfish dispute" where a tiny industry that employs only a few hundred people in Mississippi and Alabama managed to lobby the federal government into implementing a regulatory program over Vietnamese catfish imports that cost taxpayers more than the total value of the industry it was designed to protect. If catfish farmers can use this technique with such success you sound ridiculous when you say that supposed trillion dollar AI companies couldn't possibly be doing it too"

x.com/ripplebrain/st...7423765634

Ja sem bolj tehnicne narave in vlaganje mi je zgolj hobi.
Glede na svetovne razmere bo v prihodnosti marsikaj mozno.
In ameriske obveznice znajo bit kot nepremecnine npr. ali bolje delnica Krka.
Forex , dividenda in ne more propast oz bo nazadnje ;).

Glede delnic pricakujem da se bo v prihodnosti obdavcilo imetnistvo podobno kot ostala lastnina.
Oz se bo naredilo nekaj v stilu zasega zlata v tridesetih.

Sem pa mnenja, da je zlatih vlagateljskih dni konec, to bi sovpadalo z odhodom warrena v penzijo npr ;).
In bo glavno vlogo prevzela drzava, kot jo ima kitajska na svojem trgu npr.

Ai , koncentracija proizvodnje v aziji bo ljudem pustilo zgolj se kmetijstvo ;)

Prevod. Mi bomo še naprej razvijali, ampak bomo obdržali zase (če nam bo to uspelo obvladovati).

Žal je to neustavljiv vlak tako kot kockanje na borzah, ki trguje z ogromnimi vsotami a ne povečuje blaginje (ne proizvodnje in ne distribucije storitev in blaga) navadnih ljudi.

Tukaj so nas rešili Kitajci. Skoraj vsako podjetje lahko za nekaj tisoč evrov poganja svoj odprtokodni model. Ne pa, da se zanašamo na dobrohotnost in integriteto fašističnih grupacij zlizanih z državo ali komisijo.

"the big tell is how the AI companies with preexisting highly profitable business lines (Meta, MSFT, Google, Amazon) aren't joining Sam & Dario. This safety charade is about frontier LLMs scrambling b/c they don't have a viable business and can't continue burning cash on training." x.com/markaprovost/s...76124?s=20

[odvisnik]
Glede delnic pricakujem da se bo v prihodnosti obdavcilo imetnistvo podobno kot ostala lastnina.
Oz se bo naredilo nekaj v stilu zasega zlata v tridesetih.
Sem pa mnenja, da je zlatih vlagateljskih dni konec, to bi sovpadalo z odhodom warrena v penzijo npr ;).
In bo glavno vlogo prevzela drzava, kot jo ima kitajska na svojem trgu npr.

to je v grobem tudi moj vtis. ampak je trenutno sele na horizontu, ni treba preventivno obupati. ceprav morda velja tisto z avtomobilskih ogledal: objekti so morda blizje, kot se zdi.

[]
> [odvisnik]
> Glede delnic pricakujem da se bo v prihodnosti obdavcilo imetnistvo podobno kot ostala lastnina.
> Oz se bo naredilo nekaj v stilu zasega zlata v tridesetih.
> Sem pa mnenja, da je zlatih vlagateljskih dni konec, to bi sovpadalo z odhodom warrena v penzijo npr ;).
> In bo glavno vlogo prevzela drzava, kot jo ima kitajska na svojem trgu npr.

to je v grobem tudi moj vtis. ampak je trenutno sele na horizontu, ni treba preventivno obupati. ceprav morda velja tisto z avtomobilskih ogledal: objekti so morda blizje, kot se zdi.

To je pa jeba, ce oba podobno misliva ;).

Mas kako srecko, delnico, panogo, drzavo ki bo kot Nokia, amazone , microsoft, nvidia?

ne, v bistvu racunam samo se z rumsfeldovimi unknown unknowns. ker vse kar je o obrisih prihodnosti known, je precej kurac.