Keardon Schindler

Observer, Thinker, Explorer.

[eric @econoar]

Claude keeps getting worse by the day with this crap. Acting like it's tired and claiming it's too late at night to continue.

"Let's call it here I made too many mistakes, plus it's 2am"

Meanwhile it's 7pm and it just wants to quit building. Wtf?

[Keardon Schindler @Keardon]

@CL207 It redirects attention to problems that actually affect society. Time saver.

[vittorio @IterIntellectus]

how do you even defend from an army of these chasing you?!

[Andrew Feldman @andrewdfeldman]

Today, @AMD and @cerebras announce a historic partnership.

A new disaggregated inference architecture that combines the best of both worlds:

AMD Helios for world-class prefill performance and the Cerebras Wafer-Scale Engine for the industry’s fastest decode.

For years, AI inference forced a tradeoff between throughput and latency.

Developers don’t want that tradeoff. Neither do users.

The result of our partnership is an entirely new performance envelope for frontier AI: ultra-low latency at massive scale.

Faster AI changes what developers can build.
Richer user experiences.
Faster software development, robotics innovations and scientific discovery.

Entirely new classes of applications that simply weren't possible when inference is slow.

Proud to partner with @LisaSu and the AMD team again.

[Wilfred Frost @WilfredFrost]

🚨NEW EPISODE🚨

60 minutes with JAMIE DIMON

- wouldn't buy long bonds or SP500 here
- "risks are bigger than people think"
- "I want @AndyBurnham to succeed", but Banks Levy is "wrong"
- Leadership masterclass - breaking bureaucracy; overcoming insecurity; loneliness at top

STOCK MARKET: "In general would I be a buyer at this price? No."

BUYER OF LONG DATED BONDS? “Personally, no. I would not be a buyer and part of it is interest rates. I mean even if inflation was 2%, the 10-year bond should probably be at 4-4.5%. And the short rate should be at 3.25-3.5%. And they're almost there today. So I don't understand what the upside is, even if you think inflation going to go to 2%. But being an economic historian, I can't take out of my mind what happened after the great recession of 74. Deficits were less…And it climbed from 3.5% to 5% to 7% to 9% to 11%.”

ECONOMY/MARKET RISKS: “Make a list of all those really complex long-term geopolitical tectonic plates things that affect the market or may not… I do think those risks are probably bigger than other people think.”

MARCH 2020 NEAR DEATH EXPERIENCE: “So I knew at that point in time that there might be goodbye. Yeah.” And what stood out for you in the life that you'd led in that moment? “I remember I spoke to my wife and I told her to call their company and tell them exactly what's happening so they can do what they got to do. But the good news is I didn't have any great regrets. I would be leaving behind great children, great wife, great company. I did the best I can. Of course I made mistakes. But fortunately I recovered from all that.”

ON ANDY BURNHAM: “I want him to succeed. I want to see the UK thrive. I want London to succeed. But the UK, like everybody else and like my own country…you need a strong economy to do that. So the new Chancellor going to need good policies that actually cause growth. So I'm praying that they get policy right and government after government get it wrong.”

ON BANKS LEVY: “I have always thought it was wrong. JP Morgan did not damage the UK and I called the Chancellor at the time. We're a great citizen there. We hire people there. We want to be bigger there. We train people there. We hire veterans there. All of our people get medical and all that stuff like that. And I just thought it was a lack of principle to punish a company that had nothing to do with the crisis. And it's still there seventeen years later. Is that fair to a shareholder? I mean, it may sound great, 'tax the banks', but it's $5 billion that my shareholder's paid on that extra tax. And I just think things like that have adverse consequences.”

Timestamps:
0:00 Intro
2:14 As good as it gets environment
3:08 Risks bigger than people expect
4:28 Resilience despite Iran
9:21 Would not buy bonds here
12:50 AI risk & opportunity
15:16 Not buyer of SP500 here
16:02 SpaceX valuation
17:45 Lessons from Financial Crisis
19:09 Don’t expect success
21:00 Loneliness of leadership
23:25 Commitment to NYC @NYCMayor
25:03 I want @AndyBurnham to succeed
26:40 UK Banks Levy is wrong
29:14 Fighting bureaucracy
33:20 Character most important trait
35:26 Insecurity ruins leaders
39:46 Success is not just your own
43:15 Politics is a very tough game
47:55 Dimon’s founder-like impact & succession
51:30 Near death experience
53:05 Family
56:34 Learn, learn, learn – from history & people

@jpmorgan @Chase @chase_uk @JPMorganAM

[nic carter @nic_carter]

havent seen one person from OAI or Ant address Jon's argument here.

the point is simple: the USG does not owe either of the large labs a business model. if the economics of selling tokens don't work due to distillation/cheap clones/Chinese AI magick, the American enterprise and consumer will be A-OK. they will benefit from hyperdeflation in the cost of digital cognition just like everyone else. the hyperscalers will be fine. it's just OAI and Ant that won't be – in their current forms at least. if they are willing to adapt, they can develop new business models.

so what if the token merchants don't do well? the neoclouds will be fine. the internet companies will be fine. the consumer gets cheaper queries. the enterprise will still incorporate AI.

the only world in which this isn't fine, is if you hold a quasi-religious belief that we're on the cusp of a kind of AI rapture in which one of the labs Logs On And Wins Forever, namely hits RSI and we enter some kind of sublime post economic society run by GEOTUS Dario. so to accept that Ant's business model might be suboptimal or impaired by China's commoditization is to accept the unacceptable; namely that someone other than the anointed might kick off the runaway feedback loop and that they, instead might log on and win forever.

this appears to explain the discrepancy in reaction to Deepseek Moment v254 Kimi edition. everyone has bag bias, of course. but leaving that aside, most people think it's pretty much ok if Ant and OAI suffer margin compression due to Chinese distillation / industrial sabotage via open weight models. the American economy is not reliant on those two firms. they could blink out of existence and we would pretty much be ok. the AI capex supercycle will still produce tokens, closed weight or not. American firms will consume those tokens. OAI and Ant would probably still scratch a living, due to the latent preference of some token consumers to buy domestic and face off against a known entity.

this is only unacceptable if you think AI is strongly path dependent; that is, if it really matters who the market leader is when AI reaches a breakout level of capability. this is true both in the good case (superintelligence, singularity, etc) and the bad case (this is the essence of safetyism). but if this sounds more like wishcasting than forecasting, you probably don't mind the labs being pressured economically.

now you can clearly tell which side I'm on. I think AI is a fantastic technology which is hyperdeflating the cost of cognition and will fundamentally reshape society but there are real reasons why it wont diffuse as fast as the AGI people think it well. I would prefer an American firm achieve RSI relative to a Chinese one but I think either outcome would be suboptimal; better that we don't end up with a closed oligopoly composed of Ant/OAI. China by crushing the margins of the labs is doing everyone a favor by eliminating their pricing power and empowering the buyers of AI, namely, everyone.

objections:

-but you can't celebrate America losing to China!
- in my opinion this is a minor victory for China but not necessarily an enduring one. USA still has the chip, datacenter, and neocloud advantage, not to mention, it still has the best frontier models. Chinese labs releasing open weight models have no business model of their own. so even if they hurt the US labs, they have nothing to show for it. it's profoundly unlike their successful dumping campaigns with solar panels, batteries, drones, etc where they eventually built big domestic industries. (if China kills American AI with open weight models, we can even the score the moment they try and release a proprietary model). even if open weights win, the USA can still leverage AI extremely well and potentally retain the aggregate compute advantage. yes, the US would be more assured of victory if OAI or Ant won forever, but I don't know if I want to live in that world.

- no one will ever train a model again
- this is where I think the concern is unwarranted. let's say distillation really is a golden bullet and kills big training runs. that doesn't advantage either China or the US. that's a stalemate. not to mention, the trend seems to be less focusing less on massive pretraining budgets and more on finetuning for specific genres of tasks, thinking machines style. and lastly I find it hard to believe that training runs will stop altogether. the labs can probably develop anti-distillation techniques. you could adopt a whitelist style permission for everyone using your model. different consortia could be put together to share in the cost of training a model, if it is seen as too expensive for an individual firm.

- the AI buildout is path dependent and OAI/Ant are now load bearing GDP infrastructure
- it would be a significant setback for investors if they had to cancel their IPOs and suffered big markdowns, and some neoclouds with lab based RPOs would suffer for a while, but everyone would be fine, really. does Microsoft need OAI or Ant? does Meta? does Google? ordinary Americans have ~no exposure to either OAI or Ant. would the world want any less compute if it turns out to be another order of magnitude cheaper? certainly not. as we all know at this point, consumption would go up. I don't think the economy is so dependent on the labs that it couldn't handle their margins compressing.

[Brian Graham 🦬 @iroasmas]

you come to me today, on the day i disprove the jacobian conjecture, and you ask me to center a div

[kache @yacineMTB]

You live in a bubble. Some of us live in a bubble of a bubble of a bubble

[Ole Lehmann @itsolelehmann]

ok this 2.2% stat just sent me down a rabbit hole.

a few more numbers i found:

> only 0.2% of U.S. households spend over $100 a month on AI
> only 1% of U.S. adults personally pay for Claude
> only 4% of U.S. adults use AI chatbots almost constantly
> only 6% of U.S. adults use Claude
> only 4.5% have ever had an AI agent complete a task for them
> only 8.3% of U.S. workers say AI lets them do work they couldn’t do before

we are still absurdly early

[sophie @netcapgirl]

instagram: just got married

linkedin: thrilled to announce we have sold more b2b saas

X: hello there the jacobian conjecture is false

[swyx 🇸🇬 @swyx]

@__alpoge__ somewhat ironic that the j-space company has now helped disprove the jacobian

[Afshine Emrani MD FACC @afshineemrani]

I'm a cardiologist. Everyone is sharing this study as a skin story. They're burying the part that matters.
Scientists took the aorta of a 75-year-old donor, applied a single engineered enzyme, and stripped away more than 70% of the molecular damage — bringing it down to the levels you'd see in a 30-year-old artery.
Published five days ago in Nature Communications. Revel Pharmaceuticals, with Calico and the University of Colorado.
Here's what they erased.
Sugar reacts with proteins in your body the same way heat browns bread — slowly, over a lifetime. It leaves behind a residue called CML, the most abundant advanced glycation end product in aging tissue. It welds itself onto collagen and elastin in your skin, your eye lens, and your arterial walls.
Two things follow. Your arteries stiffen. And CML latches onto a receptor called RAGE, which drives chronic inflammation — the exact fire I've been writing about for months as the engine of heart disease.
Since the 1980s this damage was considered permanent. Your body has no enzyme to remove it. Every existing approach only slows new damage from forming. Nothing touched what was already there.
So they built an enzyme that doesn't exist in nature. They screened 45,000 protein structures, then ran five rounds of directed evolution across more than 500 million variants until they had CMLase — a molecular lawnmower that oxidizes the CML off the protein and restores the original, healthy lysine underneath.
Not patched. Reversed.
Over 70% cleared from elderly arterial tissue. Over 55% from elderly skin — below the levels found in 31-year-old skin. 45-78% in lens proteins. The CEO said they expected 20% and were floored.
Arterial stiffness drives systolic hypertension, heart failure, and stroke, and I have no drug that reverses it. I can slow the process. I cannot undo it. This paper says undoing it may be possible.
The caveats are real and I won't skip them. This was done on donated tissue in a dish, not in a living person. No functional data yet — we don't know if that artery got measurably more elastic. Delivering a large enzyme deep into human tissue is a hard, unsolved problem. Clinical trials are years away.
But something considered permanent for forty years just came off human tissue.
We spent a century learning to slow aging. Someone finally figured out how to erase it.
Thank you @theallinpod @friedberg @chamath @pesottas

[Pablo Antonio @PabloAntonio]

Transparent invisible drones is the most terrifying thing I’ve seen in my entire life.

This follows a thesis I have on military drones: we’re going to see less explosive drones, and more kinetic drones

[SemiAnalysis @SemiAnalysis_]

Similar to the panic over DeepSeek R1, some uneducated people think Kimi K3’s use of linear attention (KDA) is bad for NVIDIA, HBM, DRAM, and networking because it has relatively lower KV-cache requirements. The opposite is true, and we explain why below. 👇️ 1/8🧵

[Dean W. Ball @deanwball]

Some observations on Kimi:

1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run.

2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China.

3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex.

4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business.

5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this.

6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

[roon @tszzl]

the era of the chinese labs being far behind is over, Kimi is at least on par with the modern public frontier models. people have to think differently now without any competitive margin built in

[Citrini @citrini]

Continuing with the idea of it’s probably better to try to educate newbies during drawdowns rather than dunk on them, let’s talk about how expectations change as stocks rally significantly.

Basically, in the beginning of a rally you have the highest likelihood of the most positive catalyst being good earnings. People doubt the company or the environment in the present, so the best chance they have to surprise is simply to prove they are doing well right now. (Think memory in 2024, with AI overshadowed by cyclical glut fears)

Now take where we are today, when trailing multiples are twice or three times as high as forward. 

Imagine you’re holding a fishing rod. You are grasping it 6 inches from the far end of the pole and sharply raise your hand a foot in the air. The end of the pole will raise pretty much the same as your hand.

Now imagine you’re holding it from the handle and you do the same thing. The tip of the pole will move more than your hand, as there’s more room for the rod to bend and transfer/exaggerate the movement.

That’s what happens in names after everyone has turned positive on them and analyst expectations are universally positive. They aren’t able to rally simply because last quarters expectations were good, because the market is no longer pricing them on an expectation of what’s currently happening but rather on an expectation of what happens 1, 2, 3 years from now. 

What that means is higher volatility, earnings take a backseat to narratives. When the narratives are positive, the stock can rally significantly because compounding 50%+ annual growth out 3 or 4 years makes it easy to say the stock is extremely cheap on 2029 numbers. 

But it also makes it susceptible to narratives that can’t be falsified today by good earnings or bullish management (just an aside, so everyone knows, semiconductor company management is generally bullish no matter what…with rare exceptions that typically mark the bottom of a cycle). 

What that means is news about memory efficiency improvements or anticipated expansion of supply or threats from players like CXMT become really impactful when everyone is already bulled up. This is why it’s extremely difficult to buy or sell semis based on valuation - someone can always come up with a valuation to support a bull or bear case. Especially when the valuation is based on numbers 3 years from now.

These narratives can impact growth estimates by maybe 3-5% a year over the next 4 years, but that can add up to a lot. The market is holding these stocks from the far end of the fishing pole. 

This is why stocks get increasingly volatile when they have significant rallies, and also why it always seems like when they put in tops (either near term or long term), they do so over very little real developments in the way of news or fundamentals.

[Adam Singer @AdamSinger]

Opposition to data centers today will look like opposing electrification in 1910 or railroads in 1855. Actually, even worse, given we already use data centers to power the world right now, illustrating how oblivious people are to how things work

[Aleph @woke8yearold]

I have never seen an issue more polarized by IQ than data centers. Not even vaccines. It’s a reminder that many people are basically medieval peasants that think witches poisoned the local well

[SemiAnalysis @SemiAnalysis_]

State of AI Models