AI and data infrastructure drive demand for open source startups

Image credits: Hispanorism/Getty Images

A new report highlights the demand for startups building open source tools and technologies for the snowballing AI revolution, and the adjacent data infrastructure industry is heating up as well.

Runa Capital, a venture capital (VC) firm that raised money from Silicon Valley and moved its headquarters to Luxembourg in 2022, has published the Runa Open Source Startup (ROSS) Index for the past four years and is growing rapidly. Shining a light on the company. Commercial open source software (COSS) startup. The company issues updates quarterly, but last year it produced its first annual report that looked at the entirety of 2022 from the top down, and is repeating this for 2023.

trend

Data goes hand-in-hand with AI because AI relies on data to learn and make predictions, which requires an infrastructure to manage the collection, storage, and processing of data. And these tangential trends collided in this report.

Topping last year’s ROSS Index was LangChain, a two-year-old San Francisco-based startup with an open-source framework for building apps based on large-scale language models (LLMs). has been developed. The company’s flagship project will hit 72,500 stars in 2023, and Sequoia was set to lead a $25 million Series A round into Langchain just last month.

Top 10 COSS Startups in ROSS Index in 2023 image credits: Luna Capital

Also in the top 10 is Reflex, an open source framework for building web apps in pure Python, whose company recently secured a $5 million seed investment. AITable, a spreadsheet-based AI chatbot builder similar to its open source Airtable competitor. Sismo is a privacy-focused platform that allows users to selectively disclose personal data to applications. HPC-AI is building a decentralized AI development and deployment platform with the aim of becoming Southeast Asia’s OpenAI-like entity. And open source vector database Qdrant recently secured $28 million to take advantage of the burgeoning AI revolution.

A broader look at last year’s Trending Top 50 open source startups reveals that more than half (26) are related to AI and data infrastructure.

Top 50 COSS Startups in the ROSS Index in 2023 image credits: Luna Capital

It is difficult to properly compare the 2023 index with previous years from a vertical perspective, as companies often pivot and change their product positioning to match today’s trends. With the ChatGPT hype going into full throttle last year, early-stage startups may have changed their focus or put more emphasis on the existing “AI” elements of their products.

But in a landmark year for generative AI, with companies of all sizes looking to keep pace with proprietary AI giants like OpenAI, Microsoft, and Google, it’s easy to see why demand for open source components will skyrocket. I understand.

geography

Open source software has also always been highly distributed, with developers from all over the world contributing. This ethos is often reflected in commercial open source startups, which don’t have a traditional center of gravity fixed in a brick-and-mortar headquarters.

However, the ROSS index has moved toward taking geography into account, and while 26 companies on the list are headquartered in the United States, 10 of them were born elsewhere and still have founders and employees elsewhere. It reports that it is based in the following regions.

In total, the top 50 companies are from 17 countries, of which 23 are incorporated in Europe, an increase of 20% over the previous year’s index. France has the most COSS startups, with seven companies including Sismo and Massa in the top 10, and the UK has the most COSS startups, jumping from just one in 2022 to six in 2023, giving a European perspective. It came in second place based on.

Other notable information revealed from this report also includes programming languages. In 2022, he had 10 languages ​​in the top 50, compared to 12 in the ROSS index last year. However, Typescript, a JavaScript superset developed by Microsoft, remained the most popular, used by 38%. Among the top 50 startups. Python and Rust have both increased in popularity, while Go and JavaScript have decreased.

ROSS Index: Trending Programming Languages. image credits: Luna Capital

In 2023, the top 50 ROSS Index participants had a combined 12,000 contributors, increasing their overall GitHub star count by nearly 500,000. The index also reveals that funding for the top 50 COSS startups reached $513 million last year, increasing by 32% in 2022 and 145% in 2021.

ROSS Index: Contributors, Stars, Funders image credits: Luna Capital

Methodology and background

It’s worth considering the methodology behind all this.What factors influence whether a company is considered “top trending”? First, all companies are included You must have at least 1,000 GitHub stars (a GitHub metric similar to social media likes) to be considered. But the number of stars alone doesn’t tell you much about what’s trending. Stars accumulate over time, so a project that’s been on GitHub for 10 years may have accumulated more stars than a project that’s been on GitHub for 10 months. Instead, Runa uses annual growth rate (AGR) to measure the relative growth of a star over a given period of time. This compares the current star value with the previous corresponding period to see what has grown most impressively.

Particularly given that the goal is to discover open source “startups,” a certain degree of manual curation is required here as well. Therefore, the Runa investment team needs to pull out projects that belong to “product-centric commercial organizations.” It was founded less than 10 years ago and has raised less than $100 million in known funding.

Defining what is “open source” also presents its own challenges. There is a wide range in how “open source” a startup is. Some startups are closer to “open core” where most of the key features are locked behind premium. Some have paywalls and more restrictive licenses than others. For this reason, Runa’s curators decided that startups simply had to have an “r” product.“We can easily connect to open source repositories,” but it is clear that there is some subjectivity involved in deciding which repositories are adopted.

There are also more subtle nuances. The ROSS index takes a particularly liberal interpretation of “open source.” For example, both Elastic and MongDB have abandoned their open source roots for “source available” licenses to protect themselves from abuse by large cloud providers. According to the ROSS Index methodology, both of these companies would qualify as “open source” even though their licenses have not been formally approved by the Open Source Initiative, and these specific companies Example no longer calls itself “open source”.

Therefore, according to Runa’s methodology, the company uses what it calls “commercial recognition of open source” in its reporting, rather than the actual licenses it grants to projects. This is because source-restricted licenses like BSL (Business Source License) and SSPL (Server Side Public License), which MongoDB introduced as part of its transition away from open source in 2018, have become very difficult for commercial enterprises to use. This means it’s on many menus. The ROSS index is a concern.

“Such a license preserves the spirit of OSS and retains all its freedoms, with the exception of slight restrictions on redistribution. Redistribution does not affect developers. But it gives the original vendor a long-term competitive edge,” Konstantin Vinogradov, general partner at London-based Runa Capital, explained to TechCrunch. “From a VC perspective, this is just an evolved strategy for the exact same type of companies. The open source definition applies to software products, not companies.”

There are other notable filters as well. For example, companies that are primarily focused on providing professional services or side projects with limited active support or no commercial component are not included in the ROSS index.

For comparison purposes, other indexes and lists exist to determine “what’s hot” in the open source space. For example, another VC firm called Two Sigma Ventures manages the Open Source Index. This is similar in concept to Runa’s, but it covers all types of open source projects (not just startups) and has additional filters. Some argue that this is due to GitHub’s “watchers” metric, which gives a more accurate picture of a project’s true popularity.

GitHub itself publishes a trending repository page, but like Two Sigma Ventures, it doesn’t focus on the business behind the project.

So the ROSS Index has emerged as a useful complementary tool to understand which specific open source “startups” are worth monitoring.