Nebius AI - The Ultimate Cloud for AI Explorers 🌏

Plus: CMO Anastasiia on bridging the gap between cloud providers and AI disruptors...

September 17, 2024

CV Deep Dive

Today, we’re talking with Anastasiia Zemskova, Chief Marketing Officer of Nebius.

Nebius has become one of the largest providers of GPU capacity in Europe since its launch in November 2023, with a robust infrastructure designed specifically for AI workloads. The company has global ambitions, with top AI startups among its customers and plans to announce new data centers in addition to its first one in Finland.

Nebius combines hardware expertise, advanced software, and machine learning capabilities to deliver an AI-centric cloud infrastructure tailored for machine learning engineers and data scientists. With a dedicated team that has built top-tier supercomputers and AI-specific data centers, Nebius is redefining how cloud platforms serve AI professionals. Since its launch in November 2023, Nebius has supported companies like Mistral, Luma AI, JetBrains and many more in creating and scaling ML models and AI-driven applications, and it has quickly become a go-to solution for both model builders and application developers.

In this conversation, Anastasiia discusses Nebius' unique approach to bridging the gap between cloud providers and AI disruptors, the challenges of building a GPU-centric cloud platform, and their vision for the future as they strive to become a leader in the European AI infrastructure market.

Let’s dive in ⚡️

Read time: 8 mins

Our Chat with Anastasiia 💬

Anastasiia - welcome to Cerebral Valley. First off, can you share how Nebius is driving innovation and growth within the overall compute market?

Nebius is uniquely AI-centric, combining three critical elements: hardware expertise, advanced software, and machine learning capabilities. By bridging the gap between what cloud providers traditionally offer and what AI enthusiasts and practitioners need, Nebius has positioned itself at the forefront of innovation in the AI and cloud computing landscape.

Our hardware expertise is unrivaled. With a team of experts who have built supercomputers and AI-specific data centers, we’ve achieved notable successes, such as ranking #19 in the top 500 supercomputers list in June 2024. This achievement underscores our commitment to deliver optimized, high-performance hardware solutions tailored for AI.

In terms of software, our team is largely composed of professionals, experienced in developing world-class cloud infrastructure. This allows us to develop IaaS and PaaS layers of our cloud in-house, supporting the unique demands of AI workloads and offering high scalability and robustness across our services.

Our in-house machine learning team truly sets us apart from our competitors, as we are among an incredibly small number of AI-specialized cloud providers that pre-train LLMs from scratch, using our own platform. This hands-on approach, known as ‘dogfooding’, allows us to continually refine and enhance our offerings based on real-world usage and insights. 

By owning the entire stack and staying ahead of market developments, Nebius ensures a seamless and efficient AI implementation environment. This vertical integration allows us to control every aspect of the AI process, enabling rapid innovation and optimal performance. As a result, we can deliver cutting-edge AI solutions that are secure, cohesive, and perfectly aligned with the evolving needs of our clients - seamlessly driving the integration of AI into everyday life.

What are the key technical challenges in bridging the gap between GPU providers and AI disruptors?

When we launched our product in November 2023, we conducted over 100 interviews with machine learning professionals to understand their real-world challenges. What we found is that cloud providers often treat these practitioners as traditional engineers, overlooking that they are primarily data scientists with very different requirements - highlighting a disconnect between what cloud providers offer and what AI and machine learning practitioners need.

As AI continues to evolve at pace, the roles within it - like data engineers, machine learning engineers, and DevOps - are often bundled under a single title, causing further complication. These professionals are dealing with complex tasks like large-scale distributed pre-training, where there are no established best practices yet, and as generalists, they end up spending too much time on infrastructure issues, focusing less on improving model accuracy and achieving business outcomes. 

To bridge this gap, Nebius provides solutions tailored to the demands of machine learning work and helps to create environments where AI professionals can focus on refining models and delivering insights without being hindered by infrastructure concerns. By meeting these specific needs, we are helping our customers unlock AI's full potential and drive innovation across a variety of industries.

We also leverage internal practices to ensure our platform's robustness and ability to meet those needs. For example, our in-house LLM team has significantly contributed to refining our platform through extensive 'dogfooding' - using our own technology to improve performance. 

Customisation is a key point of differentiation Nebius offers over its competitors, allowing customers to tailor technology to their specific needs and workflows. Can you provide any case studies where this has worked well for one of your clients?

Customisation is a key point of differentiation for Nebius, as we allow our customers the ability to tailor our technology to their specific needs and workflows. One notable example of this is our recent collaboration with Luma AI, a leading company in generative AI video technology.

Luma AI has been leveraging Nebius’ vast infrastructure capabilities showcasing the robustness and reliability of their services. When Luma AI launched their Dream Machine, Nebius was the key cloud provider for inference. This partnership shows our dedication to delivering customized solutions that meet the specific needs of our clients. By providing the right storage, fast network speeds, and reliable infrastructure, we've helped Luma AI produce high-quality generative AI videos quickly and efficiently. 

What is the strategic vision for Nebius over the next 6-12 months?

As AI's popularity continues to rise, we are aware of the economic challenges and skepticism surrounding its viability. Our commitment is to deliver solutions that are not only innovative but economically sustainable, ensuring we meet the needs of the rapidly evolving AI market, to both model builders and application builders.

Since our product launch in November 2023, we've mainly catered to technical machine learning professionals who require robust infrastructure for large-scale pre-training. While this remains a crucial part of our offering, we’ve also recognised a growing demand for further support amongst application builders leveraging existing open-source models to create business solutions - eliminating the need to develop models from scratch. This shift in the market will allow us to address a broader range of AI needs and leverage our recent insights and strong foundation in hardware, software, and machine learning to achieve 10-15x growth.

Given the success of our recent M&A deal and our unique position in the market, we see a perfect window of opportunity to become the leading AI infrastructure provider in Europe and beyond. By catering to both model builders and application builders, Nebius can truly drive innovation amongst developers and support the diverse needs of the AI industry.

How does Nebius plan to approach competitor positioning and public relations, especially competitors like CoreWeave and Lambda Labs?

The fact that we build our platform from the ground up helps us to seize new opportunities – our DC and HW R&D, our experienced dev team, LLM R&D folks and strategic partnership with AI market players - this all gives us a lot of advantages. We already have a number of big customers and our immediate goal is to establish a strong market presence, build our customer base even further to better understand the emerging needs of the AI engineers.

Conclusion

To stay up to date on the latest with Nebius, learn more about them here.

Read our past few Deep Dives below:

If you would like us to ‘Deep Dive’ a founder, team or product launch, please reply to this email ([email protected]) or DM us on Twitter or LinkedIn.