Validation and Vision: Thoughts on Our Series B

Today marks a new chapter in MinIO’s story. With our $103M Series B, MinIO’s valuation tops $1B and represents a concrete, meaningful validation of our strategy to build the world’s leading object store. We have achieved this with a remarkable group of around forty people. It is a textbook example of how a small, disciplined, motivated and talented team can achieve amazing results.

Today, however, is for looking forward. I am even more excited as to what lies in front of us and that is what I want to focus on today. I think it makes sense to talk about our world view first, because it informs so much of what we do.

Minimalism

We believe deeply in minimalism in our product strategy. Minimalism is not the least amount, rather is the exact right amount. More adds complexity. Less reduces functionality. We constantly ask ourselves - is this the right answer, is this how we would like to experience this feature or capability. Is it enough, is it too much? Minimalism equates to simplicity. Simplicity, for those who know, is very hard. But it scales - in every way.

Minimalism will remain a driving force for us and will be why MinIO remains focused exclusively on building the world’s leading object storage system - wherever that might take us.

Object Storage

We can afford to be disciplined in our effort to build the world’s leading object storage system because we see object storage becoming the primary storage for data. Object storage has taken over the high-performance SAN/NAS workloads: databases, data warehouses/lakes and AI/ML and the application data such as static web pages, photos, videos, application artifacts and container images. These are all object storage workloads now - accessed through the S3 API, not the legacy POSIX interface.

Object storage is growing while SAN/NAS are shrinking. Kubernetes and the cloud are accelerants. We like the horse we bet on back in 2014. It will win the race.  

Performance

One of the reasons that object storage is primary storage is performance. MinIO can push 325 GiB/s on GETs and 165 GiB/s on PUTs with just 32 nodes of off-the-shelf NVMe SSDs. This delivers the performance at scale needed for the AI/ML workloads. It delivers the performance needed to power Spark, Presto/Trino, Clickhouse and other analytical frameworks enabling MinIO to replace HDFS.

Multicloud

The key theme for MinIO these days is the multicloud. We see it as our path to dominance. Our frictionless open source approach has delivered developer mindshare. We see that in the data - 10,000+ enterprises running MinIO in the last 30 days. A Slack channel with more than 16K members. Massive enterprise adoption.

What we want to take advantage of, is AWS’s monocloud mindset. AWS S3 is a superb product and we owe it a debt of gratitude. It made developers think differently about architecting their applications. We contributed to the adoption of the S3 API - even allowing legacy SAN/NAS/HDFS deployments to speak S3 - driving familiarity of the API beyond AWS.

AWS S3 has won AWS. But AWS has a problem - AWS S3 is stuck inside AWS. Even Outpost does not work independently of AWS. That is not reflective of how enterprises - both big and small, architect today. Those enterprises have multiple clouds and require data infrastructure that operates seamlessly across those clouds.

MinIO is the only company that offers data infrastructure that can operate on the public cloud, private cloud, colos and the edge without changing a single line of code in your application. We are Kubernetes-centric in our design but our Console is designed to make it simple enough that you don’t even need to know how to spell the word to create multi-tenant, object storage as a service from the edge to the core.

We see the multicloud as the ultimate advantage over bigger, more established players like AWS, Google and Azure. They are fundamentally constrained by their business model of offering a tightly integrated set of services. Enterprises are seeking to avoid vendor lock-in and keep control over their software stack and data. They increasingly view the public cloud as a commodity CPU, Network and Drive as a service - and they are right.

AWS S3 can be the dominant public cloud object storage inside AWS. We think that is a small market in comparison to the object storage market at large. MinIO wants to be the dominant data storage player across every cloud - including the public cloud.

Our roadmap is designed around this thinking.

The Private Cloud

Enterprise infrastructure is undergoing a massive transformation - from Virtual Machines, HCI, SAN and NAS hardware appliances to cloud-native architecture built on Kubernetes and object storage. With that shift comes another one - where enterprises reeling the value in data centers and hardware to software and data. While both are “assets” the first depreciates, while the second appreciates. This is why you see the industry (HW vendors and infrastructure providers) moving to on demand subscription based business models.

VMware, Red Hat, Dell and HPE will invest in the private cloud as if their businesses depend on it, because they do. That is not to say they don’t have multi-cloud ambition or capabilities - they do. But the risk today is in the modernization of the private cloud. They have to be part of that effort. It is why they all partner with MinIO. Our Kubernetes-native object storage is S3 compatible, performant, scalable and resilient. We are a key building block in their platform efforts.

We cannot achieve our goal of becoming the leading multicloud object store without including the private cloud. It is why we have invested heavily in our partnerships here, with hardware players and software giants and why we will continue to do so. The future of the private cloud is software-defined, Kubernetes-first and operationally simple - which is exactly what we are today and will be tomorrow.

The Underappreciated Edge

The least appreciated cloud is the edge cloud. Despite current evidence that it is already the largest source of data creation, most vendors can only talk about it in marketing terms. The reason is that it is hard to create a full featured, performance oriented object store that can fit on an ARM chip or an Intel Xeon embedded chip.

MinIO’s obsession with minimalism and simplicity has delivered a binary that is less than 100MB today. That will fit easily on a 5G POP or a factory floor sensor module. This is the same binary that powers the machine learning ambitions at BMW and Comcast or delivers in the cloud for PayPal or FreddieMac.

We are powerful enough to store data and allow it to be processed by AI/ML models in situ - sending the insights up where applicable and optimizing the HW/bandwidth tradeoff.  

Looking Forward

This leads me to one of the areas for which I am most excited - helping enterprises understand their data better. Object storage has a massive advantage over other storage methods when it comes to scale and structure (or lack of structure to be more precise).

The challenge is in how to understand what is inside of that data. We are working on some really interesting new features to help enterprises understand that data (think search for example). We have no interest in re-inventing the wheel when it comes to analytics, we have great partners building great products there (Spark, Trino, Splunk, Clickhouse) but we do see areas of research and development (developing the ability to unlock data using AI functions without requiring ML expertise) that will layer additional value onto our object store.

Final Thoughts
This is an exciting time for us. We will be able to spread our wings like never before with few resource constraints. Our challenge will be to stay disciplined and invest deeply but with the team we have and the team we intend to build, I am confident that we will do just that.

It is so meaningful for the MinIO team, Garima, Harsha and I to see our vision come to life before our eyes. We could not have done it without our community. We are deeply appreciative and look forward to this next chapter together.

To see Garima's post, click here.