Skadoosh

🐼 I build Claude agents | 14-step loops, real results, no theory

[Skadoosh @SkadooshGG]

https://t.co/VMVIUjmW9U

[Skadoosh @SkadooshGG]

BREAKING:

The co-founder of OpenAI disappeared for 2 years.
No interviews. No demos. No announcements.
Just silence.
This week he resurfaced.
And Nvidia — the most valuable company on the planet — didn't just partner with him.
They asked to see his secret research first.
Then immediately bet $5 billion on what they saw.
Nobody outside that room knows what Ilya Sutskever built.
But the company that powers every AI on earth just paid $5,000,000,000 to be part of it.

That's not a partnership.
That's a signal.

[Skadoosh @SkadooshGG]

BREAKING:

The co-founder of OpenAI disappeared for 2 years.
No interviews. No demos. No announcements.
Just silence.
This week he resurfaced.
And Nvidia — the most valuable company on the planet — didn't just partner with him.
They asked to see his secret research first.
Then immediately bet $5 billion on what they saw.
Nobody outside that room knows what Ilya Sutskever built.
But the company that powers every AI on earth just paid $5,000,000,000 to be part of it.

That's not a partnership.
That's a signal.

[Skadoosh @SkadooshGG]

https://t.co/VMVIUjmW9U

[Skadoosh @SkadooshGG]

@0xVampas Saved. The loop inside a loop part is exactly what most people miss.

[Skadoosh @SkadooshGG]

@0x1mpulse Excellent post. This should be required reading for anyone building with agents.

[Skadoosh @SkadooshGG]

@de1lymoon Great breakdown. Live search beats static RAG every time.

[Skadoosh @SkadooshGG]

Netflix co-founder, Anthropic board member - Reed Hastings:

"The father of AI said radiologists would be gone by 2021.
Instead: shortage of 5,000. Salaries hit $500 k.
He was completely wrong. And that changes everything about how you think about AI and jobs."

Timestamps:

4:20 - real 1-on-1 tutoring costs $100k per kid a year
10:43 - inside Anthropic: built almost entirely by AI in a week
12:38 - self-driving cars, 20 years later: still 0.1% of miles
15:01 - "isn't it writing 90% of itself right now?"
15:43 - his real forecast: high GDP, high stocks, high unemployment

[Skadoosh @SkadooshGG]

Netflix co-founder, Anthropic board member - Reed Hastings:

"The father of AI said radiologists would be gone by 2021.
Instead: shortage of 5,000. Salaries hit $500 k.
He was completely wrong. And that changes everything about how you think about AI and jobs."

Timestamps:

4:20 - real 1-on-1 tutoring costs $100k per kid a year
10:43 - inside Anthropic: built almost entirely by AI in a week
12:38 - self-driving cars, 20 years later: still 0.1% of miles
15:01 - "isn't it writing 90% of itself right now?"
15:43 - his real forecast: high GDP, high stocks, high unemployment

[Skadoosh @SkadooshGG]

https://t.co/VMVIUjmW9U

[Skadoosh @SkadooshGG]

@0xChaseTM 19 minutes that explain the next 3 years better than any article I've read

[Skadoosh @SkadooshGG]

Cognition (Devin) co-founder and CEO - Scott Wu:

"Once you've solved AGI, you just take over the entire world and then there's just one entity that rules the world. I just don't think that's the right future for us."

00:02:30 — "Nothing humans do that AI can't do"
00:03:20 — $500M revenue in 3 years, $26B valuation
00:39:00 — not efficiency, but capacity — that's the real unlock
00:40:45 — "when software drives itself — what do you want to create?"
00:43:00 — "the abundance era is real"

[Skadoosh @SkadooshGG]

https://t.co/VMVIUjmW9U

[Skadoosh @SkadooshGG]

@0xVampas Best thing I've read this week. Saving this one.

[Skadoosh @SkadooshGG]

OpenAI co-founder and CEO - Sam Altman іn a recent interview, quietly revealed the actual plan for scaling AI - and it has nothing to do with writing smarter algorithms. It's data centers building more data centers, using robot labor, in a closed physical loop nobody else is describing this bluntly.

Timestamps:

12:41 - bottleneck isn't compute or data. "Transistors, then electrons."
18:33 - "we are now in the singularity"
40:11 - ChatGPT hits 1M users on day 5. Altman goes home: "you have no idea how bad this is about to get."
46:00 - AI won't free up time for math. Everyone's busier, not less.
51:07 - the real plan: data centers building data centers, with robots.
57:26 - OpenAI killed Sora and robotics to fund coding agents.

[Skadoosh @SkadooshGG]

https://t.co/VMVIUjmW9U

[Skadoosh @SkadooshGG]

@OpenAI How long before your sales team is replaced by a voice agent that never sleeps, never misses a follow-up, and closes 24/7?

[Skadoosh @SkadooshGG]

NVIDIA co-founder and CEO - Jensen Huang:

"Someone who uses AI is going to take your job. The end of humanity? Complete nonsense. Destroying half of American jobs? Complete nonsense."

The CEO of the world's most valuable company doesn't hedge on AI doom - he calls the two most repeated fears in the industry "complete nonsense," back to back, on camera.

18:08 — "Literally, we're sitting in a plant..." (це перехідний момент)
18:20 — the jobs numbers: radiologists +20%, paralegals +10%, manufacturing +50%
23:53 — "the bubble will come someday. just not today"
29:16 — Wall Street "got it wrong" on the Kimi selloff
40:10 — China manufactures more AI researchers than the rest of the world combined
45:12 — "it is made up that there's going to be a singularity"
47:15 — "nobody should outsource their alpha. no country should."

[Skadoosh @SkadooshGG]

https://t.co/VMVIUjmW9U

[Skadoosh @SkadooshGG]

@CrpBase Simple and powerful. Love this.

[Skadoosh @SkadooshGG]

Harvard historian, Stanford Hoover Institution senior fellow — Niall Ferguson:

"The US has an edge — not in terms of raw numbers of PhDs. The Chinese throw PhDs at research and development on a massive scale, but in terms of quality, the US is still ahead. In AI, it looks to me as if the United States has a lead, and it's not obvious how China can catch up as long as it can't make its own GPUs."

In a 5-minute clip, Ferguson explains why the AI race isn't the talent contest most people think it is — China is out-producing the US on raw research manpower and still falling behind, because the actual bottleneck sits one layer down, in silicon nobody in China can manufacture yet.

Worth more than most AI-race thinkpieces you'll read this month.

He doesn't stop there. Later in the same clip, he names the exact number that should worry anyone betting on this race lasting decades: this is the first year in US history that Washington is spending more on interest payments on its debt than on its own military.

[Skadoosh @SkadooshGG]

https://t.co/VMVIUjmW9U

[Skadoosh @SkadooshGG]

Terence Tao, UCLA professor and one of the most decorated mathematicians alive, proved the thing that sits underneath every factor model on Wall Street - and makes it dangerous: in any long enough sequence, hidden structure is unavoidable. It always accumulates.

Telling the real structure from the noise that only looks like it is the entire job.

This free lecture is the most decorated mathematician alive explaining the exact problem sitting underneath every factor model, and it costs nothing to watch.

At the board it's simple. Tao's lifelong theme is the line between structure and randomness. The Erdős discrepancy problem asks a deceptively simple thing: can you write an endless string of plus-ones and minus-ones that stays perfectly balanced forever? Tao proved you cannot. No matter how cleverly you try, imbalance - hidden structure - is forced to accumulate as the sequence grows. There is no such thing as a long string of pure, structureless noise. That's the whole idea, minus the jargon.

Which is exactly why a multi-factor model can work, and exactly why it can kill you. Stack enough weak signals and real structure will appear, because at scale structure is unavoidable. But so will fake structure - patterns that exist only because the data is long enough to force them out. Finding structure is guaranteed. Knowing which structure is an edge is the rare, expensive part.

The mathematics is free and public. What nobody can sell you is the judgment to tell the structure the market will pay you for from the structure that exists only because you looked hard enough. That judgment is the alpha, and it takes years to build.

[Skadoosh @SkadooshGG]

https://t.co/sEZNHBxBms

[Skadoosh @SkadooshGG]

Tom Thornton, President at Hedge Fund Telemetry:

"Meta and SpaceX are leasing out their own AI compute to competitors like Anthropic. Possibly their models aren't ready."

In a live Bloomberg interview this week, Thornton laid out the crack nobody in tech wants to talk about — then connected it to something bigger happening across the entire quant industry right now.

The same week he said it, systematic hedge funds — the algorithm-driven quant shops that trade on statistical patterns instead of stories — posted their worst run since 2023. They gave back a quarter of their 2026 gains in about three weeks, as trades everyone had piled into in the same direction started reversing all at once. China's Zhejiang High-Flyer, the quant fund run by the same man who built DeepSeek, dropped 15.7 percent in a single week.

Quants have a name for this: crowding. When enough money chases the same signal, the signal stops being a signal and becomes a stampede. Nobody controls their own actions in a crowd. Everyone controls everyone else's.

The human version of that stampede has a name too: SK Hynix. Investors were cashing out life insurance policies to fund a 2x-leveraged bet on the chipmaker. That position is down 73 percent from its high. South Korea's KOSPI dropped almost 30 percent in a month.

There's a precedent for this exact shape of unraveling. In 2000, telecom companies quietly told Cisco they had enough networking equipment. They kept buying — just a little less. That was the whole signal. Nobody rang a bell. The Nasdaq lost 78 percent of its value over the next two years anyway.

Goldman Sachs has its own read on the current moment: whoever cuts AI spending first wins.

Nobody has cut yet.

[Skadoosh @SkadooshGG]

https://t.co/sEZNHBxBms

[Skadoosh @SkadooshGG]

@Tarik002_eth Exactly. Great models can be copied. Building a team like that is much harder.

[Skadoosh @SkadooshGG]

https://t.co/sEZNHBxBms