AWS Builder Center
AWS Bought DuckDB, NVIDIA Doubled Down, and a Big Rumor

AWS Bought DuckDB, NVIDIA Doubled Down, and a Big Rumor

AWS acquired DuckDB's creators, AWS and NVIDIA expanded their GPU partnership, and a report says NVIDIA may buy Hugging Face next.

Cloud Engineer
On August 26, 2026, three stories landed within hours of each other, and if you build on AWS, all three are worth paying attention to.
AWS announced it has agreed to acquire DuckLabs, the small Amsterdam company behind DuckDB. AWS and NVIDIA announced they are expanding their partnership again, this time adding two million more GPUs and stretching the collaboration into CPUs, networking, and even robots. And a less certain report claims NVIDIA is buying Hugging Face, the site where most of us go to grab a model. Let's walk through what actually happened, then talk about what it might mean for people building on AWS day to day.
If you have not used DuckDB, it is worth a few minutes to understand why AWS wanted it. It is an open source, in-process analytical database that runs inside your own application instead of living on a separate server. Think of it as the tool you reach for when a CSV or parquet file is too big for a spreadsheet but setting up a full data warehouse feels like overkill. It became wildly popular with data engineers because it is fast, free, and stays out of your way. According to DuckLabs, it is now downloaded more than three million times a day.
DuckLabs, the company behind it, has around thirty employees and has existed for a little over five years. AWS says the deal is expected to close in early September 2026, and the financial terms have not been disclosed. The two DuckDB creators, Hannes Mühleisen and Mark Raasveldt, will keep leading the team and the technical direction of the open source project after the move. Just as important for anyone who depends on DuckDB in production, AWS confirmed that DuckDB, DuckLake, and the rest of the ecosystem stay open source under the MIT license, managed independently by the DuckDB Foundation. This is not a case of a beloved open source project quietly getting shut down after an acquisition. AWS and DuckLabs had already been collaborating since 2024 or so, including work on DuckDB support for Amazon S3 Tables and Amazon SageMaker Lakehouse, so this looks more like formalizing a relationship that was already working well.

The AWS and NVIDIA news is a different kind of story, and honestly a much bigger number. AWS plans to deploy an additional two million NVIDIA Blackwell Ultra, Rubin, and Rubin Ultra GPUs in 2027 and 2028. That comes just five months after AWS committed to more than one million GPUs, so the total order has roughly tripled in less than half a year. What stands out is that this is no longer just a GPU purchase order. The partnership now covers NVIDIA's new Vera CPU coming to AWS, deeper networking work, open model collaboration, and even Amazon's warehouse robots adopting NVIDIA's physical AI stack. There is also a plan to build AI infrastructure for the US government using 100,000 GPUs on AWS. AWS and NVIDIA have worked together for sixteen years, and this reads like the moment that relationship shifted from customer and supplier toward something closer to co-building the infrastructure together.
Now for the part that is still a rumor rather than a confirmed deal. The Information reported, citing a source familiar with the matter, that NVIDIA has agreed to acquire Hugging Face, the open model repository that many developers treat as the default home for models, datasets, and machine learning tools, for around 12.9 billion dollars. Bloomberg's version of the story is more cautious, describing it as NVIDIA having held talks about buying Hugging Face rather than a done deal. A few days earlier there were separate reports that Hugging Face was exploring a sale and working with a bank to gauge interest, so something does appear to be moving. But there is no official announcement from either company yet, no confirmed deal structure, and no closing timeline. Treat this one as a strong rumor worth watching rather than settled fact for now.
Here is why these three stories feel connected rather than coincidental. AWS buying DuckDB fits a pattern of AWS trying to turn S3 from a place where you park data into a place where you actually work with it, which puts more competitive pressure on data platforms like Snowflake and Databricks. The NVIDIA expansion shows AWS locking in compute at a scale that goes well beyond simple GPU rental, spreading into CPUs, networking, and physical robotics. And if the Hugging Face rumor turns out to be true, NVIDIA would move from controlling the hardware underneath AI to also controlling the front door where models get discovered and shared, which is a different kind of power than owning chips.

For anyone building on AWS, none of this should change what you ship tomorrow, but it is worth watching where things head. If DuckDB stays healthy and open under AWS, expect tighter integration with S3 Tables, Lakehouse tooling, and possibly new managed offerings built around it over the next year or two. If the GPU buildout plays out as announced, capacity and pricing for large training and inference workloads on AWS should improve, though a surge of that size can also mean waiting lists during the rollout. And if NVIDIA does end up owning the place where most open models live, it is reasonable to expect competing clouds to respond by strengthening their own model hubs, and to expect a wave of consolidation among smaller AI infrastructure and data tooling startups as the big players race to lock down data, compute, and distribution all at once.
None of this is a call to panic or rush into anything. It is just a reminder that the ground under AWS-based AI development keeps shifting, and the three stories from this one week are a decent snapshot of where it is shifting toward.

Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article