A new report<\/span> highlights the demand for startups building open source tools and technologies for the snowballing AI revolution, with the adjacent data infrastructure vertical also heating up.<\/p>\n Runa Capital<\/a>, the venture capital (VC) firm that upped sticks from Silicon Valley and\u00a0moved its HQ to Luxembourg<\/a> in 2022, has published the Runa Open Source Startup (ROSS<\/a>) Index for the past four years<\/a>, shining a light on the fastest-growing commercial open source software<\/a> (COSS) startups. The company publishes quarterly updates, however last year it produced its first annual report<\/a> taking a top-down view of the whole of 2022 \u2014 something it\u2019s repeating now for 2023.<\/p>\n Data is closely aligned with AI because AI relies on data for learning and making predictions, and this requires infrastructure to manage the collection, storage, and processing of that data. And these tangential trends collided in this report.<\/p>\n Hitting top-spot in the ROSS Index for last year was LangChain<\/a>, a two-year-old San Francisco-based startup that has developed an open source framework<\/a> for building apps based on large language models (LLMs). The company\u2019s main project passed 72,500 stars in 2023, with Sequoia going on to lead a $25 million Series A round into LangChain<\/a> just last month.<\/p>\n Top 10 COSS startups in the ROSS Index for 2023 Image Credits<\/strong>: Runa Capital<\/p>\n<\/div>\n Elsewhere in the top 10 is Reflex<\/a>, an open source framework<\/a> for creating web apps in pure Python, with the company behind the product recently securing a $5 million<\/a> seed investment; AITable<\/a>, a spreadsheet-based AI chatbot builder and something akin to an open source Airtable<\/a> competitor; Sismo<\/a>, a privacy-focused platform that allows users to selectively disclose personal data<\/a> to applications; HPC-AI<\/a>, which is building a distributed AI development and deployment platform in a push to become something like the OpenAI of Southeast Asia; and open source vector database Qdrant, which recently secured $28 million<\/a> to capitalize on the burgeoning AI revolution.<\/p>\n A broader look at the \u201ctop 50 trending\u201d open source startups last year reveals that more than half (26) are related to AI and data infrastructure.<\/p>\n Top 50 COSS startups in the ROSS Index for 2023 Image Credits<\/strong>: Runa Capital<\/p>\n<\/div>\n It\u2019s difficult to properly compare the 2023 index with the previous year from a vertical perspective, due largely to the fact that businesses often pivot or change their product-positioning to suit what\u2019s hot today. With the ChatGPT hype train<\/a> going full throttle last year, this may have led earlier-stage startups to alter their focus, or even just place greater emphasis on the existing \u201cAI\u201d element of their product.<\/p>\n But as generative AI\u2019s breakthrough year<\/a>, it\u2019s easy to see why demand for open-source componentry might skyrocket, as companies of all sizes look to keep apace with proprietary AI juggernauts such as OpenAI, Microsoft, and Google.<\/p>\n Geographies<\/em><\/p>\n Open source software has also always been very distributed, with developers from all over the world contributing. This ethos often translates into commercial open source startups which might not have a traditional center of gravity anchored by a brick-and-mortar HQ.<\/p>\n However, the ROSS Index goes some way toward bringing geography into the picture, reporting that 26 companies on the list have an HQ in the U.S., though 10 of these companies originated elsewhere and still have founders or employees based in other locales.<\/p>\n In total, the top 50 hailed from 17 separate countries, with 23 of the companies incorporated in Europe \u2014 a 20% rise on the previous year\u2019s index. France counted the most COSS startups with seven, including Sismo<\/a> and Massa<\/a> which are in the top 10, while the U.K. soared from just one startup in 2022 to six in 2023, placing it in second place from a European perspective.<\/p>\n Other notable tidbits to emerge from the report include programming languages \u2014 the ROSS Index recorded 12 languages used by the top 50 last year, versus 10 in 2022. But Typescript<\/a>, a JavaScript superset developed by Microsoft, remained the most popular, used by 38% of the top 50 startups. Both Python and Rust grew in popularity, with Go and JavaScript dropping.<\/p>\n ROSS Index: Trending programming languages. Image Credits<\/strong>: Runa Capital<\/p>\n<\/div>\n The top 50 ROSS Index participants collectively gained 12,000 contributors in 2023, while the overall GitHub star-count increased by nearly 500,000. The index also reveals that funding into the top 50 COSS startups last year hit $513 million, an increase of 32% on 2022 and 145% on 2021.<\/p>\n ROSS Index: Contributors, stars, and funding Image Credits<\/strong>: Runa Capital<\/p>\n<\/div>\n It\u2019s worth looking at the methodology behind all this<\/a> \u2014 what factors influence whether a company is considered \u201ctop trending\u201d? For starters, all companies included must have at least 1,000 GitHub stars (a GitHub metric similar to a \u201clike\u201d in social media) to be considered. But star-count alone doesn\u2019t tell us much about what\u2019s trending, given that stars are accumulated over time \u2014 so a project that has been on GitHub for 10 years is likely to have accumulated more stars than one that has existed for 10 months. Instead, Runa measures the relative growth of the stars over a given period using an annualised growth rate (AGR) \u2014 this looks at the star value now versus a previous corresponding period to see what has grown most impressively.\u00a0<\/span><\/p>\n A degree of manual curation is involved here, too, given that the goal is to eke out open source \u201cstartups\u201d specifically \u2014 so the Runa investment team pulls out projects that belong to a \u201cproduct-focused commercial organization,\u201d and it has to have been founded fewer than ten years ago with less than $100 million in known funding.<\/p>\n Defining what constitutes \u201copen source\u201d has its own inherent challenges too, as there is a spectrum of how \u201copen source\u201d a startup is \u2014 some are more akin to \u201copen core,\u201d where most of their major features are locked behind a premium paywall, and some have licenses which are more restrictive than others. So for this, the curators at Runa decided that the startup must simply have a product that is \u201creasonably connected to its open-source repositories,\u201d which obviously involves a degree of subjectivity when deciding which ones make the cut.\u00a0<\/span><\/p>\n There are further nuances at play too. The ROSS Index adopts a particularly liberal interpretation of \u201copen source\u201d \u2014 for example, both Elastic<\/a> and MongDB abandoned their open source roots<\/a> for licenses that are \u201csource available,\u201d to protect themselves from being taken advantage of by the major cloud providers. According to the ROSS Index\u2019s methodology, both these companies would qualify as \u201copen source\u201d \u2014 even though their licenses are not formally approved as such by the Open Source Initiative<\/a>, and these specific example companies no longer refer to themselves as \u201copen source.\u201d<\/p>\n Thus, according to Runa\u2019s methodology, it uses what it calls the \u201ccommercial perception of open-source\u201d for its report, rather than the actual license the company attaches to its project. This means that restricted source-available licenses like BSL<\/a> (business source license) and SSPL<\/a> (server side public license), which MongoDB introduced as part of its transition away from open source in 2018, are very much on the menu as far as commercial companies in the ROSS Index is concerned.<\/p>\n \u201cSuch licenses maintain the OSS spirit \u2014 all its freedoms, except for slightly limited redistribution, which does not affect developers but grants original vendors a long-term competitive edge,\u201d Konstantin Vinogradov, Runa Capital\u2019s London-based general partner, explained to TechCrunch. \u201cFrom a VC perspective, it is just an evolved playbook for exactly the same type of companies. The open source definition applies to software products, not companies.\u201d<\/p>\n There are other notable filters in place too. For instance, companies that are mostly focused on providing professional services, or side-projects with limited active support or with no commercial element, are not included in the ROSS Index.<\/p>\n For comparative purposes, there are other indexes and lists out there that give a steer on the \u201cwhats hot\u201d in the open source landscape. Another VC firm called Two Sigma Ventures maintains the Open Source Index<\/a>, for instance, which is similar in concept to Runa\u2019s, except it spans all manner of open source projects (not just startups) and has additional filters in place, including the ability to view by GitHub\u2019s \u201cwatchers\u201d metric, which some argue gives a more accurate picture of a project\u2019s true popularity.<\/p>\nTrends<\/h2>\n
![\"Top](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/03\/Screenshot-2024-03-19-at-11.31.09.png\")
<\/p>\n![\"ROSS](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/03\/Screenshot-2024-03-21-at-10.41.16.png\")
<\/p>\n![\"ROSS](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/03\/Screenshot-2024-03-21-at-10.42.12-e1711019220234.png\")
<\/p>\nMethodology & context<\/h2>\n