At Altimeter, we love investing in companies solving foundational infrastructure problems, and few problems today are as pressing or as universal as unstructured data. That’s why we’re thrilled to back Hammerspace and lead their $100m Series B, the company building a new kind of data platform – a parallel global file system with automated data orchestration for the AI and hybrid cloud era.
Hammerspace is led by David Flynn, a visionary founder whose track record speaks for itself. As the founder and CEO of Fusion-io, David redefined how storage interacted with compute by bringing flash storage closer to the processor. His work and products at Fusion-io powered performance breakthroughs at companies like Facebook and Apple and helped lay the groundwork for modern data center architectures. Their CTO, Trond Myklebust is equally impressive. Trond is one of the original architects and lead maintainers of NFS (Network File System) in the Linux kernel, technology that underpins modern file-based data access across distributed systems. He was a core contributor to the development of NFS 4.2 which introduced key features like server-side copy, sparse file support, application I/O hints, the ability to network mount remote files as if they were local, and space reservation. All aimed at making file access over networks more efficient and enterprise-ready.
With Hammerspace, David, Trond and team are taking on the next great challenge: not just accelerating storage, but virtualizing storage and making data access ubiquitous to any user, application, or compute environment, anywhere. Hammerspace breaks data free from the constraints of hardware, geography, and legacy systems.
The problem Hammerspace addresses is both technical and strategic. Today, enterprises generate and store massive amounts of unstructured data, videos, logs, model checkpoints, 3D files, sensor data, and more, spread across on-prem systems, cloud regions, and edge environments. But this data is fragmented, siloed, and lacks sufficient descriptive metadata to identify the valuable files and objects. Teams waste time copying files, moving data to where compute lives, managing storage vendor complexity, and trying to avoid cost overruns in the cloud. For workloads like AI/ML, media rendering, genomics, and high-performance computing, that fragmentation slows down productivity and inflates infrastructure costs.
This problem is especially acute in the era of enterprise AI. While general-purpose models are powerful, the real value for many companies lies in building AI solutions tailored to their own data, models that understand their products, their workflows, and their customers. But preparing enterprise data for training is enormously difficult. The data is scattered across teams, formats, regions, and storage systems, and the performance requirements of training infrastructure (especially GPU clusters) are unforgiving. Companies don’t just need data, they need a high-performance global data platform that can unify, organize, and deliver that data to their AI pipelines. That’s exactly what Hammerspace provides.
Making matters worse, many of today’s storage systems lock customers into proprietary file systems and closed hardware ecosystems. Whether it’s a legacy NAS appliance or a cloud vendor’s file interface, customers often find themselves forced into one vendor’s tools, protocols, and upgrade paths, reducing flexibility and driving up switching costs. Hammerspace breaks this lock-in. It’s entirely standards-based, supporting pNFS, NFS, SMB, and object protocols, and integrates with the data and hardware from virtually any storage, on-prem or in the cloud vendor. That means customers can use the infrastructure they already have and avoid getting boxed in by any one vendor, all while gaining a unified control plane for their data – the Hammerspace global namespace.
Hammerspace solves this with a parallel global file system: a software layer that makes all data accessible from anywhere, no matter where it’s stored. It virtualizes storage across data centers, storage appliances and cloud providers, presenting a single, unified namespace that users and applications can interact with as if the data were local. And critically, it’s built on a parallel file architecture, meaning data can be accessed concurrently and at high throughput across many clients and nodes. This is a significant leap forward. For decades, file systems came with a performance tax, especially at scale, which pushed many workloads toward object storage. But with Hammerspace, enterprises can now have the best of both worlds: the usability and file semantics developers love, combined with the kind of raw, parallel performance AI and HPC workloads demand.
Hammerspace does not stop at file workloads. Its data platform incorporates file and object data into a single platform and makes all data available to any interface an application prefers. This puts this file vs object data argument to rest. The Hammerspace solution seems simple, and obvious. Unify all the data into a single platform and present it across any interface. The solution is obvious but the technology to make it possible is complex and unique to Hammerspace.
The use cases are compelling. Companies use Hammerspace to power tier-0 storage—feeding GPUs with checkpointed models and training data fast enough to keep multi-node clusters saturated. Media studios use it to share enormous files across globally distributed teams without syncing or duplication. Research institutions use it to burst into the cloud for large-scale computation and decommission the environment without losing access to the data. The global namespace means teams in London, New York, and Tokyo can all work from the same dataset, with local-like performance and real-time visibility—no matter where the data physically lives.
Importantly, Hammerspace doesn’t require customers to rip and replace their infrastructure. It integrates with existing storage systems, abstracts the complexity of hybrid and multi-cloud environments, and provides IT teams with the control and visibility they’ve been missing.
The tailwinds behind Hammerspace are strong. The shift toward hybrid cloud, the explosion of AI training data, and the need for real-time global collaboration are all trends that demand a better way to manage unstructured data. We believe Hammerspace is uniquely positioned to lead this category—both because of the strength of the technology, and because of David’s deep understanding of how infrastructure evolves.
We’re excited to partner with David and the Hammerspace team as they bring order, performance, and flexibility to the future of data infrastructure.
David and team saw this market coming long before most. It takes many many years to build an enterprise hardened parallel file system that is highly reliable and can scale to exabytes performantly. For newcomers to the company like myself, it feels like an overnight success. But to the people on the ground, it has been anything but (like most successful businesses). One of my favorite moments in my time working with the Hammerspace team was their company kick off earlier this year. At the end of a dinner one night they took a picture with all the people who worked with David at Fusion-io who were also at Hammerspace (for reference, David left Fusion-io over 10 years ago). You can see this image at the top of this blog post. I was shocked at how many people were in that picture, and it showed me something important. An investment in Hammerspace isn’t just a bet on the product and market, it’s a bet on their leader David. And there are a lot of people who have worked with David over the last few decades who are thrilled to make that bet. Add me to that list!
* * *
The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (“Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training.
This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.