Why we invested in XONAI: Building a Universal Compute Fabric

We are excited to announce our recent investment in XONAI, a UK-based cloud optimization startup that helps enterprise organizations reduce data infrastructure costs with a non-invasive approach that sits between the hardware and the execution layer of the software it integrates with – providing immediate acceleration to data pipelines without any migrations or code changes.

Adara participated in the $3.5M seed funding round led by Kadmos Capital. Deep Science Ventures, Nauta Capital, Notion Capital, and notable angels such as Mehdi Ghissassi, Director of Product Management for Google DeepMind, Martin Gould, former Head of Product for Spotify’s Content Platform, and others, participated. 

The challenge: Meeting the demands of big data and AI processing workloads while maintaining cost efficiency

Organizations have historically been drawn to cloud computing for its cost-saving benefits – paying for computing resources according to their specific needs. However, as data volumes increase, accurately predicting cloud service costs and managing cloud spend have become major challenges. 

According to the Flexera 2023 State of the Cloud Report, 82% of organizations indicated that managing cloud spend is their top challenge. Cloud cost management tasks are often distributed across different teams within an organization. This can slow development and result in significant time spent on data preparation and right-sizing resources – rather than more innovative, value-adding tasks. 

As big data and AI processing demands skyrocket, organizations now face yet another surge in cloud costs. Some companies using AI are already burning up to 80% of capital raised on compute resources. 

Amidst the AI industry’s rapid advancements, Generative AI data center server infrastructure and operating costs alone are forecast to exceed $76 billion by 2028. Even with an aggressive 4X improvement in hardware compute performance between now and then, this increase is overrun by a predicted 50X increase in processing workloads. In short, demand will exponentially outstrip performance capabilities without innovative compute infrastructure.

With most cloud optimization solutions on the market today, margins for cost reduction are tied to how much overprovisioned cloud resources can be right-sized, leaving no room for further improvement. Many solutions are not optimizing what’s allocating those resources to begin with – the data pipelines with demanding resource requirements. There is a real opportunity to accelerate these resource-intensive tasks and produce further cost savings through automatic hardware selection. 

This is where XONAI comes in.

Accelerating resource-intensive tasks through transparent hardware allocation

Founded in 2021 by Leandro Vaz and Brock Doiron at Deep Science Ventures in London, XONAI tackles the challenge of optimizing what fundamentally drives up compute costs: hardware allocation. 

Through a non-invasive approach that sits between the hardware and the execution layer of the software it integrates with, XONAI immediately accelerates data pipelines without any migrations or code changes. Leveraging novel compiler infrastructure designed for executing on multiple hardware, XONAI enables this acceleration from a single source of truth – a domain-specific language designed for incremental integration with subprograms in data pipelines. 

When integrated with big data analytics engines – such as Apache Spark – XONAI has achieved incredible results on cost optimization, including a baseline 3x ROI for organizations upon activation. This immediately frees up engineering teams to focus on other value-adding activities. 

XONAI’s solution is also compatible with modified Spark runtimes, such as Amazon EMR and Databricks, so performance is never lost in existing platforms. Organizations can reliably accelerate any of their existing data pipelines – even if these are already highly tuned for the environment they are being deployed, as is often the case.

The Adara view: A seamless, scalable approach to data infrastructure optimization

Most cloud cost optimization solutions today force customers to move their data and workflows to a closed-managed platform in exchange for promised cost reduction. This ties organizations to lengthy and costly migration processes that erode said cost reduction and leaves little flexibility to adjust  workflows as they grow – a realization that often comes after the migration is already complete. 

We believe XONAI has developed a superior approach, working with any existing distributed computing infrastructure – whether on the cloud or on-premises – and without requiring code or infrastructure changes. This ensures optimal performance for organizations regardless of their existing setup and enables them to immediately start reducing infrastructure compute costs and accelerating time to value. We are thrilled to work with the XONAI team during this next stage of growth.