Author:Go to the incubator at sea
The rules of the game for entrepreneurship have completely changed.
In Y Combinator's (YC) latest Spring 2026 "Startup Wish List" (RFS), we see a clear signal: AI-native is no longer just a marketing term, but the foundational logic for building the next generation of giants. Startups today can challenge fields once considered "unshakable" at a faster speed and lower cost.
This time, YC is not only focusing on software, but also turning its attention to industrial systems, financial infrastructure, and government governance. If the previous wave of AI was about "generating content," the next wave will be about "solving complex problems" and "reshaping the physical world."
Here are 10 core areas that YC is closely watching and eager to invest in.
1. "Cursor" for Product Managers (Cursor for Product Managers)
In the past few years, tools like Cursor and Claude Code have completely transformed the way code is written. But this boom has masked a more fundamental question: writing code is merely a means, figuring out "what exactly should be built" is the core.
Currently, the product discovery process remains in the "Stone Age." We rely on fragmented user interviews, hard-to-quantify market feedback, and countless Jira tickets. This process is highly manual and full of gaps.
The market urgently needs an AI-native system that can assist product managers the way Cursor assists programmers. Imagine a tool where you upload all your customer interview recordings and product usage data, and then ask it, "What should we do next?"
It doesn't just give you a vague suggestion, but instead outputs a complete feature outline and justifies the decision through specific customer feedback. Furthermore, it can even directly generate UI prototypes, adjust data models, and break down specific development tasks to be executed by the AI Coding Agent.
As AI gradually takes over specific code implementation, the ability to "define the product" will become more important than ever. We need a super tool that can close the loop from "requirement discovery" to "product definition."
2. Next-Generation AI-Native Hedge Funds
In the 1980s, when a few funds began experimenting with using computers to analyze the market, Wall Street scoffed at the idea. Today, quantitative trading is standard. If you haven't realized we are at a similar turning point now, you might miss the next Renaissance Technologies or Bridgewater.
This wave of opportunity is not about "plugging in" AI to existing fund strategies, but about building AI-native investment strategies from scratch.
Although existing quantitative giants have vast resources, their actions are too slow in the tug-of-war between compliance and innovation. Future hedge funds will be driven by swarms of AI agents—they can comb through 10-K financial reports, listen to earnings call conferences, analyze SEC filings, and synthesize views from various analysts for trading, nonstop 24 hours a day, just like human traders.
In this field, real alpha returns will belong to the new players who dare to let AI deeply take over investment decisions.
3. Software Transformation of Service-Oriented Companies (AI-Native Agencies)
All along, whether it's design companies, advertising agencies, or law firms, all agency models have faced a dead end: difficulty in scaling. Because they sell "human hours," their profit margins are low, and growth must rely on hiring.
AI is breaking this deadlock.
The new generation of agents will no longer sell software tools to customers, but instead use AI tools themselves to produce results 100 times more efficiently, and then directly sell the final products. This means:
Design companies can use AI to generate a complete set of customized solutions before signing contracts, thus outperforming traditional competitors.
Advertising companies can use AI to generate cinematic video ads without the need for expensive on-site filming.
Law firms can complete the drafting of complex legal documents in minutes rather than weeks.
Future service-oriented companies will resemble software companies in their business models: having the high gross margins of software companies, as well as infinite scalability.
4. Stablecoin-Derived Financial Services
Stablecoins are rapidly becoming a key infrastructure of global finance, but the service layer on top of them remains a wasteland. With the advancement of bills such as GENIUS and CLARITY, stablecoins are at the intersection of DeFi (decentralized finance) and TradFi (traditional finance).
This is a huge regulatory arbitrage and innovation window.
Currently, users often have to choose between "compliant but low-yield traditional financial products" and "high-yield but high-risk cryptocurrencies" in a single-choice scenario. The market needs an intermediate form: a new type of financial service built on stablecoins that is both compliant and possesses the advantages of DeFi.
Whether it is providing savings accounts with higher returns, tokenized real-world assets (RWA), or more efficient cross-border payment infrastructure, now is the best time to connect these two parallel worlds.
5. Restructuring the Old Industrial System: Modern Metal Mills
When people talk about "reindustrialization of America," they often focus on labor costs, yet ignore the elephant in the room: traditional industrial system designs are extremely inefficient.
Take the procurement of aluminum or steel pipe in the United States as an example, a delivery lead time of 8 to 30 weeks is the norm. This is not because the workers are lazy, but because the entire production management system was designed decades ago. These old factories have sacrificed speed and flexibility in pursuit of "tonnage" and "utilization rate." In addition, high energy consumption is a major pain point, and the factories often lack modern energy management solutions.
The opportunity for restructuring is ripe.
By leveraging AI-driven production planning, real-time manufacturing execution systems (MES), and modern automation technologies, we can fundamentally compress delivery cycles and increase profit margins. This is not just about making factories run faster, but about making local metal production cheaper, more flexible, and more profitable through software-defined manufacturing processes. This is a key component in rebuilding the industrial foundation.
6. AI Upgrade for Government Governance (AI for Government)
The first wave of AI companies has enabled businesses and individuals to fill out forms at an astonishing speed, but this efficiency comes to an abrupt halt when dealing with government agencies. A large number of digital applications ultimately end up in government back-ends that still rely on manual printing and processing.
Government agencies urgently need AI tools to cope with the impending data deluge. While countries like Estonia have already demonstrated a prototype of a "digital government," this logic needs to be replicated around the world.
Selling software to the government is indeed a tough nut to crack, but the rewards are equally substantial: once you secure the first client, it often means high customer retention and significant potential for expansion. This is not just a business opportunity, but also a public service that enhances the efficiency of society's operations.
7. Real-time AI Tutor for Physical Work (AI Guidance for Physical Work)
Do you remember the scene in The Matrix where Neo instantly learns kung fu by plugging in the tubes? A real-life "skill injection" is coming, and the medium is not a brain-computer interface, but real-time AI guidance.
Instead of discussing all day which white-collar jobs AI will replace, we should look at how it empowers blue-collar work. In fields such as on-site services, manufacturing, and healthcare, AI may not be able to "do" things directly, but it can "see" and "think."
Imagine a worker wearing smart glasses repairing equipment, the AI sees the valve through the camera and directly says to him in his ear: "Turn off that red valve, use a 3/8 inch wrench, that part is worn out and needs to be replaced."
The maturation of multimodal models, the popularity of smart hardware (smartphones, earbuds, glasses), and the shortage of skilled labor have together created this huge demand. Whether it's providing training systems for existing companies or building a brand new "super blue-collar" labor platform, there is enormous room for imagination here.
8. Large Spatial Models that Break Language Limitations
Large language models (LLMs) have driven the AI boom, but their wisdom is limited to what can be described in "language." To achieve artificial general intelligence (AGI), AI must understand the physical world and spatial relationships.
Current AI is still clumsy when handling spatial tasks such as geometry, 3D structures, and physical rotations. This limits their ability to interact with the physical world.
What we are looking for is a team that can build large spatial reasoning models (Large Spatial Models). These models should not treat geometry as an afterthought of language, but rather as a first principle. Whoever can enable AI to truly understand and design physical structures will have the opportunity to build the next OpenAI-level foundational model.
9. Digital Arsenal for Fraud Hunters (Infrastructure for Government Fraud Hunters)
The government is the largest buyer in the world, spending trillions of dollars each year, and also suffers heavy losses due to fraud. Just the U.S. Medicare alone loses hundreds of billions of dollars annually due to improper payments.
The U.S. False Claims Act allows private citizens to sue fraudulent companies on behalf of the government and receive a share of the recovered funds. It is one of the most effective means of combating fraud, but the current process is extremely primitive: whistleblowers provide tips to law firms, which then spend years manually organizing documents.
We need an intelligent system specifically designed for this. It is not just a simple dashboard, but an AI detective that can automatically parse messy PDFs, track complex shell company structures, and package scattered evidence into prosecutable documents.
If you can make the speed of fraud recovery 10 times faster, you will not only build a huge business empire, but also save billions of losses for taxpayers.
10. Make LLM Training Easy (Make LLMs Easy to Train)
Although AI is all the rage, the experience of training large models remains abysmally bad.
Developers struggle daily with broken SDKs, spending hours debugging GPU instances that crash as soon as they start, or discovering critical bugs in open-source tools. Not to mention the nightmare of handling terabytes of data.
Just as the cloud computing era gave birth to Datadog and Snowflake, the AI era also urgently needs better "shovels." We need:
An API that fully abstracts the training process.
A database that can easily manage ultra-large datasets.
A development environment designed specifically for machine learning research.
As "post-training" and model specialization become increasingly important, these infrastructures will become the cornerstone of future software development.