Facebook's Quiet Atlas ML Acquisition
Over the last few years, Facebook has been busy building out AI capabilities in areas like computer vision, natural language processing (NLP) and 'deep learning,' in part by acquiring promising startups in the space. Understandably, this has seen the U.S. social networking giant look to the U.K. for AI talent, including an acqui-hire of NLP startup Bloomsbury AI in 2018, and most recently acquiring Scape Technologies, a British company using computer vision to offer more accurate location positioning for augmented reality. Now a third U.K. acquisition quietly took place this December, seeing Facebook acquire Deeptide Ltd., the company behind Atlas ML, which is also the custodian of "Papers With Code," the free and open resource for machine learning papers and code. A regulatory filing for Deeptide reveals that Facebook became a majority owner on 13th December 2019. The same day, Atlas ML co-founder Robert Stojnic published a Medium post titled "Papers with Code is joining Facebook AI," which went largely unnoticed outside of the machine learning research community. Terms of the deal — or even that the acquisition took place — weren't announced by Facebook at the time, beyond Stojnic's sanctioned post. However, according to sources within London's tech community, the ballpark price is thought to have been around $40 million or thereabouts. Founded in 2018 by Stojnic and Ross Taylor, Atlas ML wanted to "make it easier to discover and apply deep learning research." The young startup was an alumni of Entrepreneur First (EF) — along with Bloomsbury and Scape — and raised subsequent seed funding from Episode1 and Kindred Capital.
From Deeptide to Atlas ML
The acquisition flew under the radar because Facebook did not issue a press release or blog post at the time. A regulatory filing for Deeptide Ltd., the UK entity behind Atlas ML, showed Facebook becoming a majority owner on December 13, 2019. That same day, Robert Stojnic, co-founder of Atlas ML, published a Medium post titled "Papers with Code is joining Facebook AI." The post announced the custodianship transfer but did not disclose financial terms. Facebook's typical discretion around acquisitions meant the deal went largely unnoticed outside specialized circles. According to my sources within London's tech community, the ballpark price is thought to have been around $40 million or thereabouts. The information comes from industry insiders familiar with the London startup scene, not from Facebook itself.
Atlas ML's Mission and the Papers With Code Custodianship
Atlas ML was founded in 2018 by Stojnic and Ross Taylor with the goal of making it easier to discover and apply deep learning research. The startup built tools that helped machine learning researchers find relevant papers and code implementations, addressing a common pain point in the research workflow. Perhaps more importantly, Atlas ML became the custodian of Papers With Code, a free and open resource that aggregates machine learning papers alongside their published code. The repository has become a staple reference for ML researchers who need to reproduce results or explore new methodologies. By acquiring Atlas ML, Facebook gained stewardship of this widely used resource, signaling an intent to integrate it deeper into its AI research pipeline. The acquisition may also give Facebook privileged access to the latest research trends captured in Papers With Code entries.
Founders, EF Alumni, and Early Backing
Robert Stojnic and Ross Taylor founded Atlas ML in 2018. Before Atlas ML, both founders were part of Entrepreneur First (EF), the prestigious talent accelerator that has produced companies like Bloomsbury AI and Scape. The EF connection places the founders in a familiar ecosystem of UK deep-tech startups, many of which have been acquired by large tech firms in recent years. After its founding, Atlas ML raised seed funding from Episode1 and Kindred Capital, two early-stage firms with a track record of backing AI and machine learning startups. The seed round validated the problem Atlas ML was solving — bridging the gap between research discoveries and practical application. While the exact amount raised was not public, the seed funding helped the small team build and launch their discovery platform before the Facebook acquisition closed the chapter.
Facebook's Pattern of U.K. AI Acquisitions
The Atlas ML deal fits a recognizable pattern in Facebook's expansion strategy. Over the last few years, the company has acquired several U.K.-based AI startups, each targeting a different capability area. Bloomsbury AI, acquired in 2018, focused on natural language processing and helping machines understand written content. Scape Technologies, the most recent acquisition before Atlas ML, used computer vision to improve augmented reality location positioning. Each acquisition has followed a similar quiet path: no public announcement from Facebook, minimal details in regulatory filings, and founder-side communications that hint at the deal without revealing terms. The Atlas ML acquisition is the latest and perhaps most significant in this sequence, given its connection to the broader ML research community through Papers With Code. The pattern suggests Facebook is methodically building a portfolio of AI talent and tools across multiple subfields, often through small UK outfits that specialize in specific research problems.
What the Custodianship Means for the ML Community
Papers With Code has become one of the most referenced resources in machine learning research, listing papers, code repositories, and evaluation benchmarks in one searchable site. Its independence and open-access nature have made it a go-to reference for researchers across academia and industry. Facebook's acquisition of the custodian raises questions about the resource's future independence. Will the repository remain open and neutral, or will it be steered toward Facebook's research priorities? The company's history of quietly integrating acquired assets suggests the transition could happen gradually, with minimal immediate changes to the public-facing site. For now, the Papers With Code catalog remains unchanged, and Robert Stojnic's Medium post remains the primary public record of the handoff. The ML community will be watching for any shifts in editorial direction or funding focus in the months ahead.
Conclusion
Facebook's acquisition of Atlas ML may have slipped past most tech news cycles, but its implications are wider than the quiet acquisition suggests. By adding Deeptide Ltd.'s team and the Papers With Code custodianship to its stable, Facebook gains both talent and a key infrastructure piece for the machine learning research ecosystem. The ~$40 million price tag, while unofficial, aligns with valuations of early-stage AI startups with strong research platforms. The deal also continues a pattern of U.K. acquisitions, each adding a different AI capability to Facebook's growing portfolio. Whether the Papers With Code resource maintains its independent spirit or becomes more tightly integrated into Facebook's research workflow remains to be seen. For now, the ML community has a new custodian — and a reminder that even the most open tools can change hands without fanfare.
