Oleg

AI futurist & investor | Early backer @perplexity_ai & @scale_ai & @solana - on my way to the next breakthroughs

[Oleg @dumbematics]

Friday 5:45pm ET / 23:45 CET, SpaceX tries to catch both Starship stages.

Watch live: https://t.co/aBSgHMsyI6
or @SpaceX.

The ship stays up for about 22 hours, so the ship catch is around 4:17pm ET / 22:17 CET on Saturday.

If Friday's double catch works, it's the first orbital rocket ever recovered in full.

Every space startup model built on today's launch prices goes stale overnight.

Odds it works from the first try?

[Oleg @dumbematics]

Seems like bitcoin is protecting family values by falling down due to the Huang affair. Hyper futures indicating that tomorrow going to the the red day

[Oleg @dumbematics]

It's not sexy to hold btc.
Buy Tencent for $52, hold some gold or silver. Be delusional and ape in Anthropic overvalued IPO.

But bro... 2026 and we want to buy BTC for $80k
Insane

[Oleg @dumbematics]

https://t.co/N8Cztp8h6B

[Oleg @dumbematics]

For the past year I was trying to use YouTube in brave browser with feed and shorts banned / just your subscriptions.

It was so bad that I switched back to the “drugged” version of the YouTube few months ago

Let’s try this one!

[Andrej Karpathy @karpathy]

Weekend hack: HackerTube. I was annoyed with YouTube's recommendations so I had Claude filter down 43M comments over 20 years of HN to 6,454 videos that keep being recommended. *Significantly* better viewing experience, ~2 hours with Opus 5.5:
https://t.co/JIhXN7OSh1

- Excellent recs without any logins (state is kept in local browser session)
- Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads
- Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version
- HackerTube Longs ✊

A bit more on the process:
I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing.

Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of.

[Oleg @dumbematics]

Should have read this one first

[Oleg @dumbematics]

You can read a complete “inspiration” of a book and still have a great time (The Creative Act: a way of being) because you ease the mind to enter this welcoming state.

Can you ease your mind while reading “The Commanding heights”? In the final words the authors advise us to:
- “Sell your Heights while they remain so”.

The word “capital” is mentioned 161 times and their main point, quote:
- “Capital will be destroyed in the old Heights.”
Soo, you want me to “Review your portfolio construction."

The book is definitely not just inspirational. They want me to act now. Do they share their personal allocation, aka skin in the game? I guess you know the answer)

And instead of saying “we all know fuck about the future” they hide it in the word salad and with peculiar royal English they declare:
“We have shown that the durable gains do not accrue to the invention, and the specter of capital destruction will visit each who brushes off the necessary reconstruction of their portfolio.”

Since they made a clear prediction of my portfolio waiting for inevitable destruction, we need to review what kind of great wisdom “they have shown”:

The short version of their thesis: AI models get cheap fast, aka don’t keep your capital in the labs. It should go to “toll booths” that AI passing through (distribution, payments, licenses, data). As a bonus to be in “tolls”, government will bailout because everyone’s 401k is in it.

First problem
The core statement: the government need AI so much, they will spend and bail no matter what. “It must happen, so it will” - American future depends on AI spending, stock prices and model rising in “one line”.

And in their forecasting of “You can see the future” where as the market participant “Keep faith, we say unto you, because the line must go up” (I should be invested).

But at the same time bro, you tell me that all my thesis and investments probably wrong any my picks of Googles and indexes will be destroyed because I must find “it is finally time again to buy DoubleClick and Android, not Alphabet or Anthropic. "

I should go try finding private companies that are targeted for acquisition by larger corporations? LOOOL Really? So the readers should become VCs now you say?

It’s good that you’re mentioning the SCALE AI and how Facebook bought it. Because since 2021 I was actually invested in Scale AI and followed “your advice”. It was a 100% return of capital to investors after the Meta deal while still keeping equity. The problem? In 2021 it was so fucking scary that you can’t really keep good stake of your wealth there. And even if you did. To find this one GEM you must fail around 10 times and potentially go broke. Check my screenshot in the comments: every Scale AI you find, there is a complete portfolio destruction in Klarna sitting next to it (good railroad though!).

You can’t tell people to go gamble their savings in private equity because you must have published the book and it had to look good. Holding low cost index fund won’t lead to “portfolio construction” but trying to find the next Android probably will for your average reader.

Ah, sorry I forgot that according to you, SCALE AI deal wasn’t about the company, “as really just a $15B offer to get Alexandr Wang”. And “Equity, as defined in business-school textbooks, begins to lose its meaning.” So now even If I found the next Android I should merry the founder instead of investing, thanks.

So equity is losing its meaning, but the line must go up. I must stay invested, but sell my Google and go find the next “Android,” which they won’t name. And once I find it, the government takes the value and the founders take the votes. Got it, made my evening

[Oleg @dumbematics]

@akbirthko It
Makes
So much
Sense now!

[Oleg @dumbematics]

@nntaleb Always one loop

[Oleg @dumbematics]

@Lightyagami191 Once CZ was in Moscow and organized a great dinner. I’m kinda on the Elon side but the food was good. I just can’t choose

[Oleg @dumbematics]

Venus day (243 Earth days) is longer than the Venus year (225 Earth days). Imagine being a North Korean hacker assigned to the night shift

[Oleg @dumbematics]

@anah_sahh Show us bro’s photo though

[Oleg @dumbematics]

@robiot What they have on the wall now @robiot

[Oleg @dumbematics]

@Mio_Mind NL is the most cared country I’ve been so far

[Oleg @dumbematics]

@naiivememe We secretly want chaos

[Oleg @dumbematics]

@MarsitaTheUltra Wish us luck bro

[Oleg @dumbematics]

About $6.3B walked out of Binance last night. Biggest crypto heist ever.
And that was just one chapter. Here's the whole day in order.

Thursday, 8:08am ET. Stan Druckenmiller posts on X for the first time ever, and it's about crypto.
He's an investor in trade(xyz), which is teaming up with ICE (the NYSE's owner) to bring 24/7 Hyperliquid perps to the US.
$HYPE ran about 20% in 24 hours, to above $96.
Wall Street wants in onchain.

Then the market opened and Wall Street got the AI bill.
Alphabet closed down 12.6%. Worst day in its history, roughly $500B of market cap gone.
Revenue miss, and free cash flow of minus $9.1B for the quarter.
Microsoft cut capex to $170B and still borrowed $25B to keep building.

After the bell, Amazon reports.
AWS up 39%.
Stock down more than 6% after hours anyway. Trailing free cash flow: minus $42.4B.
Nobody's doubting AI demand. People are doubting who pays for it.

Security was already a mess by then. Polymarket had halted trading after a hacked third party script started draining users, with about $2.5M traced to the attacker.

Then around 4:20pm ET, Binance.
Chainalysis puts it at $6.4B.
First Binance blamed one engineer's laptop. Now it says there were more holes, and Richard Teng says it "was not a single 'one-click' action."
Chainalysis and TRM Labs both pin it on North Korea's TraderTraitor, with high confidence.
BTC slipped below $77k.

Overnight it turned into an onchain war.
Tether froze more than 1.567B USDT and Circle more than 800M USDC. Others piled in, Arbitrum's Security Council even locked ETH on its chain.
Binance's latest count: ~$3B frozen.
The hacker's counter: park $60M USDT on Aave, borrow $44.6M of ETH, walk away.
You can freeze a stablecoin. You can't freeze ETH.

My favorite detail: ~80 min in, another Binance wallet topped up the already drained hot wallet with 200M USDT. It got swept in about 2 minutes.
Binance says it "will cover all losses."

Then Elon shows up.
On the hack: "Everyday is Zeroday."
Then he digs up CZ's post from the day before the hack: "Do NOT click links in emails. Do NOT download attachments."
His reply: 🔥😂

CZ: "Elon, have a lot of respect for you, but bad time to troll. Please stop."
Then he does the math in public. After freezes and SAFU, the gap is about $2B, and "we have $1.x billion in SPCX alone."
So plan B is selling Elon's rocket stock lol.

Binance co-founder Yi He, replying to CZ:
"Reporting to the major shareholder: Our company's revenue for the year is sufficient to cover the losses, so we do not yet need to sell the SpaceX shares you are bullish on."
Corporate burn of the year.

This morning the hardware wallets joined in.
Trezor says a Safe 5 bought from CryptoBilis, the Southeast Asian reseller that also sold tampered Ledgers, was reported with a hidden board and a cellular module that sends out your seed.
Bought a Trezor from CryptoBilis since March? Move your funds to a new wallet ASAP.

No spare hardware wallet? Trezor suggests a software wallet or "an exchange you are familiar with."
The morning after the biggest exchange hack ever.

A few hours later Ledger posts its weekend security tips. Step one: "Move your funds into self-custody."
So one wallet maker says try an exchange and the other says get off exchanges. Mattress is looking pretty good tbh.

Enough threats? Hold their beer
Max Planck and Ruhr University researchers claim a subexponential attack on the elliptic curve discrete log problem. That's the math protecting your keys.
Experts say it only works on absurdly huge parameters, and the curves we actually use are untouched.
For now.

[Oleg @dumbematics]

@WillManidis The author must have worked in Long-Term Capital Management (LTCM) find

[Oleg @dumbematics]

@odarwazah not worth reading in full.

[Oleg @dumbematics]

@stillibr yes

[Oleg @dumbematics]

@chrisalbon It's not only bad for kids, it's risky to follow their advice as the market participant
https://t.co/ulHpNXYrpg

[Oleg @dumbematics]

You can read a complete “inspiration” of a book and still have a great time (The Creative Act: a way of being) because you ease the mind to enter this welcoming state.

Can you ease your mind while reading “The Commanding heights”? In the final words the authors advise us to:
- “Sell your Heights while they remain so”.

The word “capital” is mentioned 161 times and their main point, quote:
- “Capital will be destroyed in the old Heights.”
Soo, you want me to “Review your portfolio construction."

The book is definitely not just inspirational. They want me to act now. Do they share their personal allocation, aka skin in the game? I guess you know the answer)

And instead of saying “we all know fuck about the future” they hide it in the word salad and with peculiar royal English they declare:
“We have shown that the durable gains do not accrue to the invention, and the specter of capital destruction will visit each who brushes off the necessary reconstruction of their portfolio.”

Since they made a clear prediction of my portfolio waiting for inevitable destruction, we need to review what kind of great wisdom “they have shown”:

The short version of their thesis: AI models get cheap fast, aka don’t keep your capital in the labs. It should go to “toll booths” that AI passing through (distribution, payments, licenses, data). As a bonus to be in “tolls”, government will bailout because everyone’s 401k is in it.

First problem
The core statement: the government need AI so much, they will spend and bail no matter what. “It must happen, so it will” - American future depends on AI spending, stock prices and model rising in “one line”.

And in their forecasting of “You can see the future” where as the market participant “Keep faith, we say unto you, because the line must go up” (I should be invested).

But at the same time bro, you tell me that all my thesis and investments probably wrong any my picks of Googles and indexes will be destroyed because I must find “it is finally time again to buy DoubleClick and Android, not Alphabet or Anthropic. "

I should go try finding private companies that are targeted for acquisition by larger corporations? LOOOL Really? So the readers should become VCs now you say?

It’s good that you’re mentioning the SCALE AI and how Facebook bought it. Because since 2021 I was actually invested in Scale AI and followed “your advice”. It was a 100% return of capital to investors after the Meta deal while still keeping equity. The problem? In 2021 it was so fucking scary that you can’t really keep good stake of your wealth there. And even if you did. To find this one GEM you must fail around 10 times and potentially go broke. Check my screenshot in the comments: every Scale AI you find, there is a complete portfolio destruction in Klarna sitting next to it (good railroad though!).

You can’t tell people to go gamble their savings in private equity because you must have published the book and it had to look good. Holding low cost index fund won’t lead to “portfolio construction” but trying to find the next Android probably will for your average reader.

Ah, sorry I forgot that according to you, SCALE AI deal wasn’t about the company, “as really just a $15B offer to get Alexandr Wang”. And “Equity, as defined in business-school textbooks, begins to lose its meaning.” So now even If I found the next Android I should merry the founder instead of investing, thanks.

So equity is losing its meaning, but the line must go up. I must stay invested, but sell my Google and go find the next “Android,” which they won’t name. And once I find it, the government takes the value and the founders take the votes. Got it, made my evening