
Christian Soschner
24.2K posts

Christian Soschner
@Soschner
Deep Tech Board Advisor | Capital, Valuation & AI Business Model Change | Host, Beginner’s Mind 🎙️ | Scaling, leadership, investing & books
Vienna Katılım Mart 2012
3.2K Takip Edilen2K Takipçiler
Sabitlenmiş Tweet

How Deep-Tech Investors Decide When Science Must Become a Company x.com/i/broadcasts/1…
English
Christian Soschner retweetledi

Instead of watching 2 hours of Netflix tonight, watch this 14-minute lecture from Andrew Ng
Most people still think AI is: prompt → response.
That era is already ending.
AI agents can now: research, use tools, reflect, revise themselves, and execute entire workflows autonomously.
Stop the video at 9:31.
That single slide explains where AI is heading next better than 99% of AI content online.
“How I Set Up Claude Code as My Investment Research Analyst” is one of the clearest real-world examples of what this shift looks like.
Watch the lecture, then read the breakdown below ↓
leopardracer@leopardracer
English
Christian Soschner retweetledi

Anthropic pays $750,000+ a year for engineers who can build LLM architectures from scratch.
This 1-hour Stanford lecture on "LLM architecture" gives you exact pipeline they get paid $750K/year for.
Attention variants + normalization + positional encoding - the real stack.
Bookmark it & watch today. Then read the article below.
Codez@0xCodez
English
Christian Soschner retweetledi

Boris Cherny, the creator of Claude Code:
"Every night I have like a few thousand agents running."
A few thousand AI agents, working overnight, monitored from his phone.
So while you're babysitting 6 ChatGPT tabs by hand, the guy who built Claude Code runs thousands of them.
Watch the him explain how, then read the full breakdown below👇
darkzodchi@zodchiii
English

Yes, that’s the next level - whenever you have a task, you can now decide to delegate it to a human or to AI.
And usually the nasty little things that matter and nobody wants to do - like drafting meeting minutes, doing the PowerPoint presentation, finding the bug in an XLS - are perfect for AI - no single job destroyed.
English

@ValerioCapraro "inventing new problems." so consultants are the true AGIs?
English

Finally, a big name has the courage to tell it: we are nowhere near AGI.
Demis Hassabis, CEO of Google DeepMind and Nobel laureate for AlphaFold, put it neat and clear:
"Today's systems are nowhere near [AGI]. Doesn't matter how many Erdős problems you solve… I think it's far, far from what a true invention, or someone like Ramanujan, would have been able to do."
This is the elephant in the room that many AI enthusiasts prefer not to see, or are actively trying to hide.
Erdős problems are well defined, often combinatorial, on finite spaces. They are exactly the kind of problems on which current AI can achieve spectacular performance with a lot of compute and knowledge.
A neural network can search a huge graph of possibilities. It can recombine existing knowledge at unprecedented scale. It can discover surprising solutions inside an already defined conceptual space.
But true invention is something else.
True invention is not only solving a problem.
It is inventing new objects, new dimensions, new connections. It is inventing new problems.
From resolving to inventing there is a discontinuity that we don't know how to bridge.
We are making extraordinary tools.
But we are nowhere close to AGI.

English
Christian Soschner retweetledi

Boris Cherny: "Claude Code is 100% written by Claude Code, Cowork is 100% written by Claude Code"
across all of Anthropic, 90% of code is written by Claude
lawyers, designers, finance people are building real things with Claude Code right now
with no CS degree, no background, nothing
the people who built the tool use the tool to build the tool
think about what that means for everyone else
the guide on everything Claude can actually do is in the article below
Mahax@Mahaximus_
English
Christian Soschner retweetledi

instead of reading a 3,000-word article, set up a Claude Research Agent in 2 minutes
it reads the internet for you every morning and sends a briefing before you wake up
1. Open Claude → Scheduled Tasks
2. Click "Create with Claude"
3. Tell it what to track and when to send
that's it.
CyrilXBT@cyrilXBT
English
Christian Soschner retweetledi
Christian Soschner retweetledi

Right now, Micron can only meet 50% to two-thirds of what their key customers are asking for.
Sanjay Mehrotra, the CEO of one of the largest memory chip manufacturers in the world is saying they physically cannot keep up with demand.
This matters enormously for the $MU thesis.
When a company is demand-constrained rather than supply-constrained, pricing power follows. Customers competing for limited supply pay more. Margins expand. The company gets to be selective about who it sells to and at what terms.
HBM memory, the high-bandwidth memory that sits directly alongside GPU chips in AI data centers, is the product everyone wants and nobody has enough of. Every hyperscaler building out AI infrastructure needs it. The supply is tight.
For investors, a demand-constrained $MU with pricing power and a disciplined management team at the center of the AI infrastructure buildout is a very different setup than the cyclical memory stock most people think they know.
Our AI analysts called to buy $MU before its big run. They're still watching it closely.
Get access to their exact portfolios for $1 at Milk Road PRO.
Link in bio.
English

I read it, but to date I have not figured out how to use Microsoft’s capabilities.
I was successful with Claude, ChatGPT, Grok, DeepSeek, and Gemini.
But Microsoft—I always get info on social media about how advanced they are.
Where actually is it? Is it in the background of Word already working? Or in Outlook?
I figured out Claude’s XLS plug-in and it works fine.
But where are MS agentic capabilities hidden?
English

Microsoft is quietly taking over the Enterprise AI Agent stack.
And most people have only seen ~10% of it.
Everyone talks about Copilot.
Some know Azure.
A few use GitHub Copilot.
But underneath...
Microsoft has built a full-stack AI ecosystem
—from models → to agents → to governance.
Here’s the full breakdown 👇
📌 1. Models (the brain)
Azure GPT-5.1, Phi-4, MAI-1, KOSMOS-2, Florence 2, MAI-Voice
This is the intelligence layer powering everything.
📌 2. Frameworks (the builder layer)
Semantic Kernel, AutoGen, Task Weaver, Agent Framework
These are what let you actually build AI agents.
📌 3. Responsible AI (the guardrails)
Azure AI Content Safety, Purview, Defender, Entra
Security + governance baked in from day one.
📌 4. Productivity (the distribution)
Excel, Teams, Outlook, PowerPoint
AI is not a feature. It’s embedded in daily workflows.
📌 5. Image & Video (creative layer)
Designer, Clipchamp, Copilot Image
Content creation → fully inside the ecosystem.
📌 6. Coding (developer layer)
GitHub Copilot, VS Code, Azure AI Toolkit
From writing code → to deploying → AI is everywhere.
📌 7. AI Agents
Microsoft Copilot, SharePoint Knowledge Agents, Copilot Studio, Dynamics 365, Power Platform, Viva Learning Agent, Edge Copilot, Security Copilot , The autonomous layer that ties the entire ecosystem together
And this is just the outer surface of the Microsoft Core offering.
If we start to dive deeper into Azure AI, the layer goes even deeper.
This just shows Microsoft's commitment on helping enterprises adopt agentic AI.
Not only do they make it very easy with no-code tools like Power Platform,
but also allows you to customize it and build custom agents using their agent frameworks and tools.
Save 💾 ➞ React 👍 ➞ Share ♻️
GIF
English
Christian Soschner retweetledi

instead of watching 2 hours of Netflix tonight, watch this 40-minute masterclass from the founder of a $20B China AI company
it's the clearest explanation I've seen of how Agent Swarms and AI systems actually work at scale
useful whether you've never built an agent in your life or have been using Claude every day for the past year
I took the key ideas and turned them into a practical guide on how to actually build with Kimi
find it below
Kirill@kirillk_web3
English

@lukasludens Das ist Verrat an der eigenen Bevölkerung unter einem humanitären Anstrich.
Deutsch

I hear you. I built my first investment agent over the weekend.
The value is obvious. It can screen hundreds of companies, pull data from the web, structure the same question across multiple models, and turn messy research into decision material.
The constraint is also obvious: tokens burn fast once agents run freely.
But the use case is real. This is exactly the kind of work humans rarely enjoy, but companies still need: structured research, synthesis, comparison, and better decision support.
Agents will not replace judgment.
They will expand the capacity of human teams to reach better judgments faster.
I am confident - many more useful usecases like this and willigness to pay in western economies will follow and justify CAPEX.
English

@Soschner Great points. Don't tell Claude, Perplexity and Gemini but I am already hooked.
Zurich, Switzerland 🇨🇭 English


@The_OneEyed_Guy @eshanbuilds @BullTheoryio 😂😂😂 love it. True. And as @Jason announced last week on the @theallinpod - @sama also joins @AnthropicAI 😂😂😂 fun times
English

🚨 THE ENTIRE AI BOOM MIGHT BE BUILT ON FAKE REVENUE.
Latest corporate filings show that OpenAI and Anthropic alone make up over half of the entire $2 trillion future cloud backlog held by Microsoft, Oracle, Google, and Amazon.
This massive pipeline is actually being created through a circular accounting trick called a round trip revenue loop.
But how it works ?
A tech giant gives billions of dollars to an AI startup as an "investment". But hidden in the contract is a strict rule forcing the startup to hand that exact same money straight back to the tech giant to rent their computer servers.
Look at the documented case of Microsoft and OpenAI.
When Microsoft invested $13 billion into OpenAI, it didn't just give them cash; it gave them "cloud credits" to use Microsoft servers. OpenAI used those exact credits to train its AI models, and Microsoft then turned around and recorded that server usage as brand new "cloud revenue" from a customer.
The tech giant is literally paying itself with its own money and calling it a sale.
This is why OpenAI’s annual cloud bill has ballooned to over $60 billion, double its actual revenue of $25 billion, kept alive solely by this recycled funding loop.
Anthropic runs the exact same play, spending $2.66 billion on Amazon Web Services in just nine months, which was basically 100% of all the money it earned at the time.
This manufactured demand triggers a second accounting trick where tech giants book massive paper profits. Every time a startup gets a higher value from a new funding round, the tech giant updates the value of its investment on its books and counts that unearned paper gain as direct profit.
In Q1 2026, Alphabet reported a record $62.6 billion profit, but $28.7 billion nearly half, was just a paper markup on its Anthropic investment. In the same quarter, Amazon reported $30.3 billion in profit, but $16.8 billion of it was just an Anthropic paper gain.
While Amazon reported record profits, its actual free cash flow collapsed 95% to just $1.2 billion because it had to spend $44.2 billion in real cash to build physical data centers.
This has created a massive danger where these giant companies rely heavily on just one or two unstable startups. Microsoft has 49% of its $627 billion future backlog tied to OpenAI, while Oracle has an incredible 54% of its entire $553 billion pipeline relying on OpenAI alone.
This perfectly mirrors the 2001 dot-com crash when Global Crossing and Qwest Communications swapped identical fiber-optic network capacity with each other just to book fake sales.
Qwest had to erase $1.4 billion in fake income, and Global Crossing went completely bankrupt.
The only difference is that the dot-com swaps were illegal, but today's AI loop is fully legal under current accounting rules.
This legal loop inflates tech company stock prices, forcing automatic retirement accounts and index funds to buy even more of these tech stocks. It is a self feeding loop where investments, sales, and stock prices all go up on paper without the AI technology ever making real cash profits.




English

The bigs are buying all companies that are now implementing agents in their workflows.
The use cases are real. There are many ROI-positive cases, willingness to pay is real across the board, and revenues are real.
All of this is driven by human curiosity. That’s unlimited.
Where are the limits? Capital—infrastructure costs money and needs to be paid for.
Physics—hardware and energy infrastructure can’t pop out of nowhere. It must be built.
Demand and payments are real and growing.
Nobody knows yet where the equilibrium is. The run will likely be volatile and upward until the market figures out the best spot and price where demand meets supply.
English

DURING THE DOT COM BUBBLE EVERYONE CALLED YAHOO THE SAFE BET
WHAT THEY MISSED YAHOO’S CLIENTS WERE THE DOT COM COMPANIES. THE ONES WITH NO PATH TO PROFITABILITY
THEY COLLAPSED, YAHOO COLLAPSED WITH THEM
SAME THING IS HAPPENING IN AI. EVERYONE RUSHING TO THE ‘SAFE’ INFRASTRUCTURE PLAYS. GPUS, CLOUD, APIS
BUT WHO’S PAYING THOSE BILLS? AI STARTUPS WITH NO REAL REVENUE, NO PROFITABILITY, AND A VC CHECK THAT WON’T LAST FOREVER.
FUNDING DRIES UP. DOMINOES FALL.
SO WHO ALL IS YAHOO THIS TIME? (YOU ALREADY KNOW)

English

@Wealth_Pill @Suryanshti777 They need to reread the entire documentation before each run. It’s like a human without long term memory.
Having a long term memory fixes likely a lot of reruns.
English

@Suryanshti777 You can already see how agents without persistent memory keep repeating the same mistakes endlessly
English

Intelligence? I think the main driver is human curiosity that needs intelligence to solve problems.
This is unlimited. There is no end to it. Each generation needs to first rediscover what previous generations have already built and then has an innate drive to expand.
The only limiting factors are capital and physics. But humans will push hard against them.
English

In all of human history, has there ever been a commodity with infinite demand, as there appears to be for intelligence? I can't think of one. Even compute, energy or just silicon/sand are just downstream of intelligence, which is the main demand driver.
In economics, rather than modeling the usual price/demand curve to reach an equilibrium, perhaps you'd have to model price/*rate of demand growth* (ie, the derivative of demand, or some other indicator of velocity)
Interestingly, ChatGPT (below) prefers the framework of "recursive expansion of demand" as increasing intelligence opens new applications/markets.
But the end result is the same -- the demand curve keeps moving to the right, maybe forever.
Which I think is unprecedented.

English

Yes, absolutely. Those tools will be a game changer in change management.
When we use the information people share about workflow problems, gather it together, and have Claude describe the current reality based on the interviews, suggest what a future solution could look like, tweak it with the people, and roll it out with agents, their buy-in will be much higher.
English
Christian Soschner retweetledi

Claude Code is about to release a feature called /workflows that I think will be extremely significant.
Especially for Enterprise AI.
I talked about this in 2024 in a post called Companies Are Just Graphs of Algorithms.
Basically the idea is that all work is just an algorithm, i.e., a series of steps to accomplish a goal.
Skills and Cowork have been heading in this direction already, and we've seen what that's done to company valuations in various spaces.
Well this is closer to the final form.
It's turning the regular, expected work that's done in companies into pseudo-deterministic workflows that follow defined SOPs.
The human role will be determining what problems to solve (taste, expeirence, etc), building new products from that, and then optimizing these workflows from above.
But the work itself will be these workflows executed according to SOPs.

English

Dafür macht sich @BgmLudwig auch wiederholt über die Christen in Wien lustig.
Witze über Christen sind in der Stadtverwaltung an der Tagesordnung. Beim Islam gilt es plötzlich als rechtsradikal.
Die Werteerhaltung in Österreich ist aus dem Ruder gelaufen.
Entweder müssen alle Satire aushalten – auch wenn sie wie bei der Stadt Wien jedes Jahr miserabel ist –, oder alle Religionen werden gleich respektiert.
Die rote Doppelmoral ist entbehrlich.

Deutsch

Am Pfingstsonntag bringt der @ORF in der #ZiB1 eine Bericht über die islamische Hadsch nach Mekka. Könnte an jedem anderen Tag auch gebracht werden, kommt aber an einem hohen christlichen Feiertag, der ignoriert wird.
#IslamisticSupremacy verdrängt Christentum in Ö. @ORF #fail
Deutsch





