Clustered scale-out file system provider Qumulo has changed CEO – from eight-year incumbent Bill Richter to ex-Arista VP and software general manager Doug Gourlay – intending to grow in applied AI markets with its scale-anywhere, hybrid cloud filer software.
Richter took on the CEO role in November 2016, when investors put in $32.5 million in a C-round, followed by $30 million in a C-prime round the next year, a $93 million D-round in 2018, and a $125 million E-round in 2020. He used these cash infusions to develop the business, its software, go-to-market organization, and channels. Today Qumulo has over 1,000 customers. It uses full-featured and performant file system software that runs on-premises and in cloud-native form in the Azure and AWS clouds. In fact, it beats Weka, known for its fast parallel file system performance, in the Azure cloud. Richter now becomes a Qumulo board and executive advisor.
Samir Menon, director at BlackRock, a Qumulo investor, remarked in a statement: “We are thrilled to have Doug build on our recent successes and lead the expansion of the next growth phase for Qumulo. His accomplished career, vision, and passion for customer success make him ideal to lead Qumulo.”
Matt McIlwain, managing director at Madrona Venture Group and a Qumulo board member since 2014, added: “Doug helped scale Arista Networks from an early-stage startup to an over $100 billion market cap company, making him the ideal person to guide Qumulo’s next phase of expansion.”
Gourlay said: “I am delighted to join Qumulo. The combination of an outstanding team, innovative on-premises and cloud software, opportunity in applied AI, and the Scale Anywhere vision drew me to this company. I look forward to executing our unique vision and strategy for our customers, partners, and channels. To quote America’s first astronaut just before launch: Let’s light this candle.”
We asked execs about this leadership change.
Blocks And Files: Why the change from Bill Richter to Douglas Gourlay?
CTO Kiran Bhageshpur: Executive management change, a natural evolution, is often inevitable after eight-plus years as CEO. This transition, while bittersweet, is a testament to our commitment to growth and evolution. We have identified someone who can build on top of the foundation Bill has developed: A great team, strong culture, high-quality product foundation, exceedingly satisfied customers, and a broadly differentiated product targeting mainstream enterprise customers.
B&F: What does Doug bring to Qumulo that Bill could not?
Bhageshpur: Bill’s core competencies are a solid financial background and being incredibly operationally savvy. Doug’s background is in visionary product strategy, strong product/market leadership, and scaling businesses consistently to over $1 billion in revenue. When we interviewed Doug, it became clear that his extensive background in networking and systems while paying keen attention to the engineering, operational, and executive priorities of the mainstream enterprise customer base was incredibly synergistic with Qumulo’s next growth phase.
Of particular note is that we increasingly see interest in the reconvergence of compute, network, and storage in many of our clients – having a leader who views this as a broad systems play, as opposed to a point product development, is a vital enabler for Qumulo.
B&F: Will Qumulo under Doug’s reign be moving quickly to ensure customer information held in its storage will be available for retrieval-augmented generation (RAG) LLMs?
Gourlay: This is one of the first questions I had for Kiran and the executive team when discussing the evolution of Qumulo. When you combine an industry-architecture independent “run anywhere” model with a cloud-native implementation that flattens the consumption economics across public clouds with a modern global namespace implementation, the application of having your “data anywhere, everywhere, all at once” is being able to apply that data to solve complex business problems – especially for mainstream enterprise.
Let’s be direct for a moment – the market for hyperscale AI training/learning and model development is quite real. However, the spartan human capital that is actually capable of developing quality models and the significant infrastructure investment going into this space by a small number of well-funded organizations leaves this in the purview of nation-states, hyperscale technology companies, and specialty trading operators.
Cloud providers are developing capacity for mainstream enterprise consumption using largely “off the shelf” models from OpenAI and others. Enabling an enterprise to use its data securely to interact with a well-formed and mature public model without having its data become part of the public model is transformative to the consumption of AI – moving to a more “applied AI” model. RAG is a key technology we are actively integrating to unlock the potential of the exabytes of data we manage and store across volumes with billions of unstructured and semi-structured data sets.
B&F: That would seem to be a pressing need and one that will be present both on-premises and in the public cloud, perhaps with different implementations?
Gourlay: Qumulo’s strategy and today’s product execution normalize the implementation and management of the client’s data, whether placed in an on-premises datacenter, a remote edge location, or within a public cloud. Arguably, even more important is the economic catalyst – normalizing the unit economics, enabling the enterprise to have the freedom of choice to determine whether to use the cloud, or an existing datacenter, or a hosting facility without the technology foundation inhibiting or pre-determining a specific outcome.
The ability to offer on-premises and cloud-based consumption, consistent unit economics, and removing the legacy barriers and Hobson’s choice of “cloud vs on-premises” is a unique differentiator for Qumulo and one that resonates from the head of infrastructure to the CIO.
B&F: Will Qumulo be developing its own LLMs to assist customers in using their Qumulo data estates?
Gourlay: I’ll never say never, but this is not in our current execution plan. We are guided by a combination of customer priorities, building what is “right” and high quality for our clientele, and paying keen attention to market, secular, and technology trends.
It is much more likely that we will use the same RAG-assisted models we are enabling to leverage the incredible investment in LLM development at OpenAI, Microsoft, Google, and others than to try to develop our own competing LLM in such a crowded space that is building such generally capable technology.
B&F: Will Qumulo in Azure be integrated with the OpenAI Azure facilities?
VP, cloud & strategic partnerships Brandon Whitelaw: We cannot discuss the nature of our strategic partnership with Microsoft and Azure, especially future developments that have not been announced. We have a strong and deep partnership with Microsoft, remain the only storage partner in the Azure Native ISV program, and actively work together along multiple fronts.
B&F: Will Qumulo move quickly to have a cloud-native presence in AWS and get with AWS AI facilities such as Bedrock?
Whitelaw: As you might have seen, we have already announced Cloud Native Qumulo on AWS. We are in an active co-development mode with customers and AWS on various integrations.
