Y Combinator updated its Requests for Startups for the spring 2026 batch. The list highlights areas where YC actively wants new companies.
AI tools that turn user interviews and product data into clear product decisions and feature plans.
Funds built around AI agents that research filings, synthesize signals, and execute trades from the ground up.
Design, marketing, or legal agencies rebuilt as software-first businesses with scalable margins.
Payments, investing, and cross-border money movement built on stablecoins while regulation is still open.
Tools that help governments handle growing workloads with lower cost and faster execution.
Automation and energy optimization for outdated steel and metal manufacturing systems.
Real-time AI assistance for workers using cameras and step-by-step instructions on the job.
AI systems that can reason about 2D and 3D space and physical manipulation.
Platforms that organize whistleblower data, messy documents, and complex corporate structures.
Better tooling for large datasets and end-to-end LLM training workflows.
The list signals where YC expects new categories to emerge next.
Please open Telegram to view this post
VIEW IN TELEGRAM
π27β€26π17π₯9
Sid Sijbrandij became a billionaire after GitLab went public in fall 2021. A year later, in 2022, he was diagnosed with a spinal tumor. Surgery and chemotherapy led to remission. In 2024, the cancer returned and standard treatment paths were exhausted.
Sijbrandij stepped away from GitLab and switched to what Paul Graham calls founder mode. He built a dedicated team around his own treatment, including an external CEO, computational biology experts, and a scientific advisory board.
The work expanded treatment options from almost none to dozens. After therapy, the tumor shrank enough for surgery. Cancer is currently not detectable, and the team continues working to prevent relapse.
Startup execution, applied to staying alive.
Full article
Please open Telegram to view this post
VIEW IN TELEGRAM
β€37π₯21π15π13
The comparison focus on open models suggests pretraining is largely complete. Meta previously signaled it may move away from open weights, which makes Avocadoβs positioning notable.
Please open Telegram to view this post
VIEW IN TELEGRAM
β€30π27π―22
A pilot launch is planned soon. The model turns training data into a priced input rather than a legal gray zone.
Please open Telegram to view this post
VIEW IN TELEGRAM
π₯34π28β€20
Markets reacted sharply after reports around new Anthropic tooling, with a broad selloff across software, finance, and asset management names.
Bloomberg data showed heavy pressure across the sector:
The trigger was not a new model, but the release of 11 plugins for Claude Cowork on Jan 30. These tools target full workflows like financial modeling, legal research, and sales operations, rather than plugging into existing software.
Stocks tied to those workflows sold off hard:
Wall Street has started calling it a βSaaSpocalypse.β The concern is that foundation model companies are moving beyond APIs and into owning the application layer itself.
For markets built on selling automation, this is no longer theoretical.
Please open Telegram to view this post
VIEW IN TELEGRAM
π25β€23π―22π12
A rare YC solo founder story, coming from far outside the usual startup hubs.
Please open Telegram to view this post
VIEW IN TELEGRAM
β€56π₯33π27
Her last email to Epstein was sent 11 days before his 2019 arrest.
Please open Telegram to view this post
VIEW IN TELEGRAM
π20π’16π―14π10π€―9β€6
Please open Telegram to view this post
VIEW IN TELEGRAM
π39π€36β€32
This media is not supported in your browser
VIEW IN TELEGRAM
Frank Bourassa didnβt just make βfakeβ money he made it better than the government
He served just 6 weeks and paid a fine of $1,350
Thatβs basically the cost of a TV for a $250 million crime
Please open Telegram to view this post
VIEW IN TELEGRAM
π56β€39
Please open Telegram to view this post
VIEW IN TELEGRAM
π₯28π22β€17π€16
βBeing very early and being wrong look exactly the same 99% of the time.β
β Seth Klarman
βοΈ Subscribe on @venture
β Seth Klarman
Please open Telegram to view this post
VIEW IN TELEGRAM
π€―27π19β€17π₯17
Big money for a single URL.
Branding wars in AI are heating up fast.
Please open Telegram to view this post
VIEW IN TELEGRAM
π³31π±23π20β€13
The companies themselves are saying the opposite.
For 2026, internal capex plans are blowing past analyst expectations:
This isnβt defensive spending.
This is an all-in bet on AI infrastructure, data centers, and compute.
When the people writing the cheques are more bullish than Wall Streetβ¦
it usually matters.
Please open Telegram to view this post
VIEW IN TELEGRAM
π28β€17π₯14
Anthropic, the company behind Claude, is still private but already valued around $350B after recent funding rounds. Capital is moving fast, even before an IPO. Here are the main routes investors are using today.
Accredited investors can buy Anthropic shares directly from existing holders on private secondary platforms. This is the closest thing to owning the stock pre-IPO, but access is limited and liquidity is thin.
Some venture funds include Anthropic as part of a broader portfolio of private tech companies. This lowers single-company risk, but Anthropic usually represents only a small percentage of the fund.
Cathie Woodβs ARK Venture Fund holds Anthropic alongside other private AI leaders. It offers exposure without direct ownership, though fees are higher and Anthropic is not the core position.
Google, Amazon, Microsoft, and Nvidia have all invested billions into Anthropic, mostly tied to cloud and compute deals. Buying these public stocks gives indirect exposure, mixed with their much larger core businesses.
Some investors choose a basket approach through AI-focused ETFs or major AI builders. This spreads risk across the sector but removes any pure Anthropic bet.
Reports suggest Anthropic has discussed a possible IPO in 2026, but nothing is confirmed. This remains the cleanest option for many investors, assuming timing and valuation make sense.
The market is already pricing Anthropic like a public giant. The IPO would just make it official.
Please open Telegram to view this post
VIEW IN TELEGRAM
π28π±23π₯15β€8
"Intelligent people make decisions based on opportunity costs."
- Charlie Munger
βοΈ Subscribe on @venture
- Charlie Munger
Please open Telegram to view this post
VIEW IN TELEGRAM
π25β€20π₯17π’15
The gap highlights a mismatch between revenue strength and public market outcomes. Over time, this difference may narrow as more Chinese firms mature or list.
Please open Telegram to view this post
VIEW IN TELEGRAM
2β€15π³14π12π₯10π6
Mars is still the long-term goal. For now, the Moon is back at the front of the roadmap.
Please open Telegram to view this post
VIEW IN TELEGRAM
β€7π6