Insights from conversations with almost two hundred startups — what they are building with, and what changed when AI arrived.
We’ve talked to almost two hundred startups in the last several months, and that’s given us insight into what tech stacks startups are using in 2023. Here are the trends we’re seeing at Fraction — and what changed when generative AI arrived.
We originally published this piece based on conversations with startups in late 2022 — but a lot has changed in the tech world since then. In a word: AI. The explosive rise of LLMs and generative AI, with ChatGPT storming onto the scene, has impacted software tech stack choices as developers seek to integrate both OpenAI and open-source LLMs into their products.
React remains king of the front end in late 2023, but JavaScript’s growth on the server side has paused. SQL is far from dead, and major vendors like Snowflake have breathed new life into it in an era of ever-larger data sets. Before I go further, a note on the data: this is based on conversations with almost 200 tech startups since September 2022. We work with US-based companies with fewer than 100 employees — the vast majority have fewer than 30, true early-stage startups, with a smattering of existing businesses putting their toe into the software game.
Before jumping into traditional technology stacks, the rise of low-code and no-code solutions deserves its own attention. We’ve seen substantial growth in the use of low-code tools by early-stage startups — whether to build mobile apps (FlutterFlow), to build highly customized general apps (Bubble), to build more B2B-centric apps (Retool + Refine.dev), or to go fully no-code for the lowest-cost go-to-market (Softr). No one vendor or approach rules the low-code market just yet.
The essential pattern we see: early-stage startups are getting to market faster and cheaper via low-code, and then pivoting to traditional code after they see real traction with customers. It is a sensible sequencing — validate first, then invest in engineering infrastructure.
An increasing percentage of startups are leveraging GitHub Copilot and its numerous competitors to boost their developers’ productivity. At Fraction, we’re close to mandating the usage of such tools — there’s no excuse for not using ML-powered assistance. One word of warning, though: these tools cannot magically turn junior developers into senior developers, and shipping machine-written code to production without testing carries real risk.
Fraction embeds experienced engineers, architects, and PMs directly into your team — on a fractional basis.
See Available TalentNo commitment required to browse profiles.
React leads by a wide margin. Approximately 65% of startups in our sample use React, often in conjunction with full-stack frameworks like Next.js. React’s market dominance is bolstered by an abundance of associated frameworks, including React Native support for mobile apps. Next.js is emerging as the de-facto standard for new React applications — the majority of React-using startups we’ve spoken to in the last quarter all use or plan to use it.
Roughly one in five startups uses Vue instead of React. Many find its templating system easier to use, and it integrates more seamlessly into existing projects. Svelte and SvelteKit are rapidly gaining ground as well, though they don’t yet have as many projects in production.
The balance are using Angular or other JavaScript frameworks. Angular is ceding ground rapidly to React and Vue, with some teams actively migrating away — which makes sense given Vue’s origins as a project founded by an ex-Angular developer.
By the end of 2022, it seemed like JavaScript was going to be the one language to rule them all. Then the generative AI explosion happened. We were seeing over 60% of startups using JavaScript via Node.js at the end of 2022 — today that number has fallen to around 40%. Python has surged forward into a tie with JavaScript, because machine learning is now front-and-center at so many startups. Even when teams are just using OpenAI APIs, tools like LangChain tend to be Python-first, driving more interest in the language.
Python has jumped from 25% of our pool in late 2022 to 40% today. Django still gets usage, but the rise of microservice architectures means we hear about Flask and FastAPI at least as often. Python gets the nod when a startup has data science or analytics needs. On the JavaScript side, TypeScript is overtaking traditional JavaScript for server-side development.
C# and the .NET framework remain alive and well among startups building on the Microsoft stack — critical when interfacing with Excel or other Microsoft products is important. Java seems to be fading. I built HiddenLevers in Java, but there are far simpler options for startups today. Mobile apps are increasingly being authored via Flutter and React Native, or Kotlin / Swift when native, leaving Java further on the outside.
The vast majority of startups are still using SQL, with PostgreSQL the most common database we encounter. NoSQL options like MongoDB and DynamoDB occasionally come up, but the power of modern SQL databases to cover both approaches — storing large JSON in columns is now trivially easy — makes them the more flexible choice for most use cases. SQL is not going away anytime soon, and vendors like Snowflake have given it new life for large-scale analytics workloads.
No big surprises here: it is AWS, AWS, AWS — with Azure rising due to its OpenAI relationship, and GCP rounding out the big three. The cloud hosting providers are increasingly convergent in terms of providing serverless capabilities, managed databases, authentication and authorization, and dozens of other managed services. The more you depend on a single provider, the deeper the vendor lock-in — but the savings in avoided DevOps and infrastructure management can be enormous at early scale.
Worth noting: we see a lot of mixed environments, particularly when serverless architectures are involved. AWS Lambda, Azure Functions, and their equivalents let you use the language of your choice for each service, so more teams are mixing and matching by use case or developer preference.
To summarize what we’re seeing across nearly 200 startups: React / Node.js / PostgreSQL was the default stack for the majority at the start of 2023 — but as of September 2023, it is a tie between that and React / Python / PostgreSQL, particularly for startups building data analytics or machine learning in-house. The best programming language is still the one your team already knows.
React is the dominant front-end framework, used by approximately 65% of startups. It is often paired with Next.js, which is emerging as the de-facto standard for building new React applications. Vue.js holds roughly a 20% share, while Svelte and SvelteKit are gaining ground rapidly. Angular is ceding market share to React and Vue.
They are now roughly tied at around 40% each. JavaScript via Node.js led at over 60% in late 2022, but Python surged forward due to the explosion of generative AI and LLMs. Python is the dominant language for machine learning and data science, and tools like LangChain are Python-first, driving its growth among AI-focused startups.
The vast majority of startups still use SQL, with PostgreSQL being the most common choice. NoSQL options like MongoDB and DynamoDB come up occasionally, but modern SQL databases are flexible enough to handle most use cases, including storing large JSON blobs in columns. This makes them the preferred default for early-stage teams.
The rise of LLMs and generative AI — led by ChatGPT — has significantly shifted server-side language choices, pushing Python from 25% to roughly 40% of startups. Teams integrating OpenAI APIs or open-source LLMs gravitate toward Python because most AI tooling, including LangChain, is Python-first. AI coding assistants like GitHub Copilot are also widely adopted to boost developer productivity.
Yes, increasingly so. Early-stage startups are using tools like FlutterFlow, Bubble, Retool, Refine.dev, and Softr to reach market faster and cheaper. The common pattern is to validate product-market fit with low-code first, then migrate to traditional code once real customer traction is established.
AWS is the dominant choice by a wide margin. Azure is rising, largely due to its relationship with OpenAI. GCP rounds out the big three. All major cloud providers offer increasingly convergent capabilities: serverless compute, managed databases, and authentication. The deeper you commit to one provider, the more vendor lock-in you accept — but the DevOps savings at early stages are substantial.
Praveen Ghanta is a five-time founder and serial entrepreneur. He is the founder of DevHawk.ai, an AI-powered engineering management platform, and Fraction.work, which connects fast-growing companies with top fractional tech and growth marketing talent. Previously, he founded HiddenLevers, a risk analytics platform for wealth management that he bootstrapped from inception to acquisition by Orion Advisor Solutions in 2021, serving thousands of advisors and $600B in assets. He earlier founded SmartWorkGroups, acquired by Intralinks in 2000.
Connect on LinkedIn →Describe your software or AI project. Get a full scope with story-point pricing, sprint estimates, and a downloadable plan in minutes. No calls, no waiting.
Scope Your Project for FreeWorking on a data strategy? Talk to a Fraction CTO. → Book an intro call
