I would have won that bet about 7 times out of 10.

We track over 104,000 open startup jobs, and the tech stack data tells a clear story: startups are converging on an increasingly narrow set of tools. That convergence has real benefits — easier hiring, better tooling, more Stack Overflow answers. But it also has costs that the industry isn't reckoning with.

The numbers

Here's what the startup job market actually looks like by technology:

  • Python: 19,395 jobs (19% of all listings)
  • AWS: 11,409 (11%)
  • React: 5,595 (5.3%)
  • Kubernetes: 5,905 (5.6%)
  • Azure: 5,363 (5.1%)
  • Java: 4,991 (4.8%)
  • TypeScript: 4,844 (4.6%)
  • GCP: 4,816 (4.6%)
  • LLM/GenAI: 4,647 (4.4%)
  • Salesforce: 4,547 (4.3%)

Python's dominance is staggering. Nearly 1 in 5 startup jobs mentions it. But the more interesting pattern is in the frontend: React (5,595) versus everything else. Angular appears in roughly 1,800 listings. Vue in about 900. Svelte in under 300. React has effectively won the startup frontend war, and it's not close.

On the backend, the picture is more fragmented but still converging. Python and TypeScript/Node.js together account for the vast majority of new startup backends. Java persists largely in fintech and enterprise-facing startups. Go has a meaningful presence in infrastructure. Rust is growing but still niche.

Why convergence happens

This isn't random. Startups converge on the same tools for rational reasons:

Hiring pool size. If you're a 20-person startup trying to hire 5 engineers, you pick the stack that has the largest pool of available talent. React has more developers than any other frontend framework. Python has more developers than any other language. Choosing something exotic means a smaller hiring pool, longer searches, and higher salaries.

Ecosystem maturity. React has 14 years of libraries, tutorials, and battle-tested patterns. TypeScript has first-class IDE support everywhere. AWS has a service for literally everything. When you're moving fast and can't afford to debug infrastructure, you pick the tools with the most paved roads.

Investor and advisor influence. This one's subtle but real. When a YC partner or a technical advisor reviews your architecture, they're more likely to approve something they recognize. "We're using React, Node, and Postgres" gets a nod. "We're using Elm, Elixir, and CockroachDB" gets questions. Founders learn quickly that conventional choices reduce friction.

Copy-paste culture. Startups copy each other's stacks the way they copy each other's pricing pages. When Vercel, Linear, and Notion all use Next.js, it becomes the "correct" choice for any startup building a similar product. The reasoning is circular — "successful companies use X, therefore X is the right choice" — but it's powerful.

The costs nobody talks about

Convergence has a dark side, and it shows up in three places:

1. The senior engineer shortage is partially self-inflicted

When every startup requires React + TypeScript + AWS, they're all fishing in the same talent pool. This creates artificial scarcity. There are plenty of excellent engineers who work in Vue, or Angular, or who prefer Python backends to Node. But they're filtered out by job requirements that treat a specific framework as a hard requirement rather than a preference.

One CTO I spoke with realized this after spending 4 months trying to hire a senior React engineer. He changed the requirement to "strong frontend experience in any modern framework" and filled the role in 3 weeks — with someone who had been building production apps in Svelte for two years and picked up React in a week.

The framework doesn't matter nearly as much as the fundamentals. A great engineer can switch frameworks in days. But the job market treats framework experience as a proxy for competence, which is expensive for everyone.

2. Monoculture creates systemic risk

When 60% of startups run on AWS, a single AWS outage takes down 60% of startups. When everyone uses the same npm packages, a single supply chain attack affects everyone. When every startup's frontend is React, a breaking change in React's API is an industry-wide event.

This isn't hypothetical. The left-pad incident in 2016 broke thousands of builds. The Log4j vulnerability in 2021 affected virtually every Java application on earth. The more the ecosystem converges, the more correlated the failure modes become.

3. Innovation slows at the edges

The most interesting technical ideas in the last few years — Rust for systems programming, SQLite for edge computing, HTMX for simpler web apps, Elixir for real-time systems — are all happening outside the mainstream stack. But startups that adopt these tools face a hiring penalty, a fundraising penalty, and a support penalty.

The result is a two-tier ecosystem: a safe, boring center where most startups live, and a risky, interesting periphery where the next generation of tools is being developed. The center gets all the talent and funding. The periphery gets all the innovation. And the gap between them is widening.

What this means for your career

If you're a developer choosing what to learn, the convergence creates a clear short-term strategy and a less obvious long-term one.

Short term: learn the stack. Python, React, TypeScript, AWS, PostgreSQL. This combination will qualify you for the largest number of startup jobs. It's not exciting advice, but it's correct. If you're job searching right now, these are the keywords that get you past the filter.

Long term: go deep, not wide. The developers who command the highest salaries aren't the ones who know React and TypeScript — everyone knows React and TypeScript. They're the ones who understand distributed systems, database internals, performance optimization, or security at a level that transcends any specific framework.

The convergence means that framework knowledge is becoming commoditized. When every junior developer knows React, knowing React isn't a differentiator. What differentiates is understanding why React makes the architectural tradeoffs it does, when those tradeoffs are wrong, and what to do about it.

The contrarian bet: If you're early in your career and willing to take a risk, learning a non-mainstream technology deeply — Rust, Elixir, or even something like Zig — could pay off enormously in 3-5 years. The developers who knew Kubernetes in 2018, before it became ubiquitous, had their pick of jobs by 2021. The next Kubernetes is probably something most startups are ignoring right now.

The stack doesn't matter (except when it does)

Here's the paradox: the best startups know that the tech stack is one of the least important decisions they'll make. What matters is the product, the market, and the team. Stripe was built on Ruby. Facebook was built on PHP. Neither would have been more successful with a "better" stack.

But the job market doesn't operate on that logic. It operates on keyword matching, and the keywords are React, Python, TypeScript, and AWS. Until hiring practices catch up with reality — and they won't anytime soon — the convergence will continue, the talent pool will stay artificially constrained, and startups will keep complaining that they can't find engineers while filtering out perfectly qualified candidates who happen to know the wrong framework.