Oriole Networks

Company Overview

Oriole Networks is a new company founded in 2023 that aims to revolutionize the performance of AI systems and speed up data centers while dramatically reducing energy consumption. The company is based in London, UK, with its registered office at 100 Liverpool Street, London EC2M 2AT. Oriole Networks is a spinout from University College London (UCL), with intellectual property licensed through UCL’s technology transfer company UCLB.

The company was founded by four individuals: - Professor George Zervas - James Regan - Alessandro Ottino - Joshua Benjamin

James Regan, who serves as the CEO, has prior experience in building successful tech companies from university spinouts, having previously built EFFECT Photonics.

Products Overview

Oriole Networks is developing a novel approach to connect thousands of AI GPUs directly to each other using light. This technology aims to:

1. Increase the performance of AI systems
2. Speed up data centers
3. Dramatically reduce energy consumption

The company claims that with their approach: - Large Language Models can be trained up to 100 times faster - Energy consumption is reduced to a tiny fraction of current levels - Machine learning algorithms can run with a thousandth of the latency

This technology is expected to revolutionize time-critical tasks such as algorithmic trading and accelerate AI adoption and algorithmic progress.

Founding Team

1. Professor George Zervas - Chief Technology Officer (CTO) Background: UCL scientist, expert in networked systems

2. James Regan - Chief Executive Officer (CEO) Background: Experienced entrepreneur with a track record of building successful tech companies from university spinouts

3. Alessandro Ottino Background: UCL scientist

4. Joshua Benjamin Background: UCL scientist

Problem and Market Fit

Oriole Networks is addressing several key issues in the AI and data center industry:

1. Increasing computational demands: AI computational needs are doubling every 18 months, leading to distributed training and inference across large numbers of xPUs.

2. Data movement bottleneck: Collective data movement across servers in data centers is becoming a bottleneck, limiting training and inference completion time.

3. Unsustainable power consumption: The increasing demands on data centers are leading to systemic problems and unsustainable power consumption.

4. Network constraints in High-Performance Computing (HPC) environments: Over the last decade, compute performance has improved ten times faster than networking performance, creating constraints in HPC environments.

Oriole Networks’ solution aims to address these issues by using light to connect AI GPUs, potentially unlocking significant performance improvements and energy savings in a market that is actively seeking such solutions.

Business Model

While the specific details of Oriole Networks’ business model are not explicitly stated, it can be inferred that the company plans to commercialize its technology by:

1. Developing and selling hardware solutions for AI and data center applications
2. Potentially licensing their technology to existing data center and AI hardware manufacturers
3. Offering consulting services to help companies implement their technology

Funding and Runway

In March 2024, Oriole Networks successfully raised £10 million in seed funding. The round was co-led by:

• UCL Technology Fund
• Clean Growth Fund
• XTX Ventures
• Dorilton Ventures

Additional support was provided by Innovate UK Investor Partnership.

The specific runway information is not provided, but given the substantial seed round, it’s likely that the company has secured funding for at least 18-24 months of operations.

Competitive Landscape

While specific competitors are not mentioned, Oriole Networks is operating in a highly competitive field of AI hardware and data center technology. The company’s unique approach using light to connect AI GPUs sets it apart from traditional networking solutions. However, it’s likely competing with other innovative startups and established tech giants working on improving AI and data center performance.

Customers

No specific customers are mentioned in the provided information. Given the nature of the technology, potential customers would likely include:

• Large tech companies running extensive AI operations
• Data center operators
• Financial institutions involved in high-frequency trading
• Research institutions and universities working on large-scale AI projects

Relevant News

March 27, 2024: Oriole Networks announced a £10 million seed funding round, co-led by UCL Technology Fund, Clean Growth Fund, XTX Ventures, and Dorilton Ventures, with support from Innovate UK Investor Partnership.

Key points from the announcement: - The company claims its approach can train Large Language Models up to 100 times faster while consuming only a tiny fraction of the power. - Oriole Networks’ technology is said to enable machine learning algorithms to run with a thousandth of the latency. - The company reports up to 100 times speed-up in completion time and 40 times improvements in energy consumption.

The funding news highlights the significant interest from investors in Oriole Networks’ technology and its potential to address critical challenges in AI and data center operations.