The Layer Between Your Database and the Cloud Bill

Most software you can see. You open it, you click around, it does something for you. Silk is the opposite kind of company. If you are a patient at Franciscan Health or a customer at a bank running its books on the public cloud, you have almost certainly touched Silk's software and had no idea. It has no interface most people will ever open. It is a layer - a thin, unglamorous layer that sits between a company's most demanding databases and the cloud those databases run on - and its entire job is to make the cloud behave.

The pitch is deceptively plain: databases and analytics on Amazon Web Services, Microsoft Azure or Google Cloud get roughly ten times the performance, without the overprovisioning that quietly inflates the monthly bill. Silk describes itself as "software-defined cloud storage that delivers predictable performance at a fraction of the cost." That sentence is doing a lot of work, and unpacking it is the story of a company that once sold physical boxes and decided the boxes were the wrong business.

01 / The ReinventionFrom a lightning god to a software layer

Silk did not start as Silk. It started in 2008 as Kaminario, a name borrowed from a Japanese god of lightning, and for its first decade it built all-flash storage arrays - the physical hardware that lived in corporate data centers and made databases fast. Founder and chief executive Dani Golan raised money against that hardware thesis through the 2010s: a $15 million Series A in 2011 with Sequoia, Globespan and Pitango; a $68 million round in 2015 backed by Silicon Valley Bank and Mitsui. By 2017 the company was reported at a $1 billion valuation.

Then it did the thing most hardware companies cannot bring themselves to do. It walked away from the hardware. Around 2017 Kaminario began shifting to storage delivered as software, and in 2020 it completed a full rebrand to Silk. The old business - arrays, appliances, the whole apparatus of shipping metal - was set down in favor of code that could run inside someone else's cloud. It is one thing to add a product line. It is another to retire the thing that got you to a unicorn valuation.

The company killed its own hardware business: after a decade building flash arrays, it kept only the software and re-launched as a pure cloud company.

02 / The ProductWhat the layer actually does

Here is the problem Silk exists to solve. When a demanding database runs on the public cloud, teams often make it fast the blunt way - they overprovision, buying far more capacity and horsepower than they need at any given moment, just to cover the peaks. It works, and it is expensive. The cloud's convenience comes with what amounts to a performance tax.

Silk's answer is to place its own high-performance storage layer directly adjacent to the customer's compute, inside the same cloud. That proximity is the trick: it delivers consistent, low-latency access to live production data at scale, so the database hits its performance targets without the customer buying its way there through overprovisioning. The Silk Cloud Data Platform supports AWS, Azure and Google Cloud, and is sold through each of their marketplaces. In practice it functions as a virtual SAN - a storage network - that the hyperscalers do not offer natively.

03 / Silk EchoCopies that take up no space

In February 2025 Silk extended the platform into the busiest corner of enterprise IT: artificial intelligence. The new product, Silk Echo, is a copy data management tool, which sounds mundane until you understand the constraint it removes. To run AI, analytics or dev/test work, teams normally have to copy production data - and copies cost storage, time and risk. Silk Echo creates instant, fully functional copies that consume no additional storage, effectively a live digital twin of the production database. Teams can spin up environments and run intensive AI pipelines against current data without touching the systems that keep the business running.

Silk Echo empowers unlimited developers to create fully operational environments instantly and without a data footprint.Adik Sokolovski, Chief R&D Officer, Silk

It is a neat inversion of AI's least-discussed bottleneck. Everyone talks about models and GPUs; far fewer people talk about the plumbing that gets real, current data in front of those models. Silk's bet is that the data-logistics layer is where a lot of the practical value - and the practical pain - actually lives.

04 / The CustomersHospitals, banks, and the cost of hours

Silk sells to enterprise IT and data teams - CIOs, cloud architects and database administrators - rather than to consumers, and it has clustered in industries where databases are large, regulated and unforgiving: healthcare, financial services, retail and SaaS. Two customer results appear front and center in its own materials. Franciscan Health reports $2.8 million in annual cloud storage savings from eliminating performance-driven overprovisioning. Sentara, another health system, reports 32x faster analytics and ETL workflows, shrinking processing windows from hours to minutes.

Those two numbers describe the same coin from opposite sides. One customer used Silk to spend less for the same speed; the other used it to go dramatically faster for the same footprint. Both are versions of removing the performance tax.

Total capital raised across equity and debt is reported at roughly $380M+. Figures are approximate and drawn from public filings and trackers.

05 / The BusinessSitting on top of the giants

The most interesting thing about Silk's business model is what it decided not to do. It does not compete with AWS, Azure or Google Cloud. It sits on top of them. Customers keep their cloud; Silk layers in, makes the databases faster and the bills smaller, and takes a subscription in exchange. Distribution runs both directly and through the same cloud marketplaces its customers already buy from. It is a quietly clever place to stand - adjacent to the hyperscalers rather than against them.

That positioning also defines its competition. The real alternative to Silk is usually native cloud storage used the expensive way - Amazon EBS and RDS, Azure managed disks - plus a field of storage and data-management vendors like Pure Storage, NetApp and, for copy data specifically, Rubrik. Silk's argument is not that those tools are wrong, but that a dedicated performance layer changes the math for the most demanding workloads.

Silk deliberately doesn't compete with the hyperscalers. It sits on top of them and sells the performance they don't.

06 / The PeopleAn Israel-to-Needham engineering story

Silk's engineering roots run through Israel's storage-systems community, with the commercial base in Needham, Massachusetts, just outside Boston. In November 2022 the company hired Tom Murphy as chief marketing officer to accelerate its go-to-market. Murphy arrived with two decades of Boston software marketing behind him, including a stint as CMO of Turbonomic before its acquisition by IBM, plus earlier roles at Bit9 and Bradford Networks.

The future of almost any industry - from healthcare to retail and financial services - is in the cloud, and joining a company that supports organizations in their cloud endeavors was a no-brainer.Tom Murphy, CMO, Silk

In January 2026, Silk turned up in Gartner's Hype Cycle for Strategic Cost Management - analyst shorthand for a category that exists because the cloud got expensive and enterprises went looking for ways to make it cheaper without making it slower. Silk has, in effect, been working that problem since before it had a fashionable name.

07 / The TimelineTwo names, one company

08 / Where It FitsThe layer nobody notices

Silk occupies a narrow, useful band of the market: not a cloud, not a database, not an application, but the performance layer between them. It is the kind of company whose success is measured in things not happening - the outage that did not occur, the overprovisioned capacity that was never bought, the analytics window that quietly closed in minutes instead of hours. For a business that once shipped boxes you could point to, that invisibility is the whole point. The product works best when you forget it is there.

Whether the next chapter belongs to real-time AI - Silk's clear bet with Echo - or to the plainer economics of cloud cost, the company is playing the same hand it has played since it stopped being Kaminario: put the right layer in front of the cloud, and let the cloud do the rest.