Our goal is to provide expert, accurate, value-add research in the LLM infrastructure and application space, accessible to a general business audience.

Over the past few years, a tremendous amount of false information and analysis has been propagated in this area by research providers. This came to a head last week when a complete misinterpretation of a new product from Chinese company DeepSeek led to dramatic market turmoil and publicity. There’s a particularly common theme where research providers just repeat some market rumour, or copy from a new research paper/presentation/podcast/social media without accurate context or proper understanding. Jargon is often heavily used to obfuscate this lack of expertise.

We will focus on the facts, and expert analysis of the facts.

Additionally, we will deliver easily digestible reports that carry a very high density of value and understanding for our readers. We will limit extraneous information and detail as much as possible - of course you can reach out for additional detail or analysis on any point. Much of the good information is hidden in scientific papers and technical specifications - we analyse these documents and interpret them for our clients at a much more accessible level.

What will we provide?

• Important news and developments in the AI ecosystem e.g. our DeepSeek article
• Thematic articles covering important trends that will affect short- and medium-term decision making at companies, investors, and hardware and model providers e.g. our Inference Compute Scaling article

Our coverage will span three key areas:

• Hardware – limitations, advantages and benchmarking of GPUs for LLM training & inference
• Models – LLM architectures, generation methods & compute
• Applications – usage & adoption of LLMs for useful economic work

Contact us