Revisiting the unicorn concept

Subscribe • Previous Issues

The AI $100 Million Revenue Club

Everyday there’s a new unicorn. It used to be a status that meant that a startup has graduated from being a startup and into a mature company worth being listed on the public stock market backed by revenue. In today’s climate becoming a unicorn is increasingly a signal of investor enthusiasm than how mature a company is, with companies with $1-5 million in revenue being awarded the status. 

Consequently, Shasta Ventures’ Kenn So and I are creating and maintaining a list of flying unicorns (Pegacorns) – AI companies that have reached the $100M revenue milestone and have graduated from startups to mature companies. Find out which companies made the inaugural list and what skills their founders possess.

Read The Post

Data Exchange podcast

  • Why You Need A Time-Series Database:  This conversation with Timescale founders (Ajay Kulkarni and Mike Freedman) took place a few weeks after Timescale raised a massive funding round and achieved unicorn status.
  • The 2022 AI Index:  I discuss key sections of the fifth edition of the AI Index with Jack Clark, co-director of the AI Index Steering Committee. This year the report contains a chapter dedicated to AI Ethics where they focus on new benchmarks and metrics that have been developed to measure bias in AI systems.
[Image by Ben Lorica.]

Data & Machine Learning Tools and Infrastructure

  • Pathways from Google: This is a new system that orchestrates distributed computation for accelerators. A very early example of Pathways in action comes from language models where it was used to train PaLM, a 540-billion parameter, dense decoder-only Transformer model.
  • Large-scale distributed training with TorchX and Ray: Any existing TorchX component can now run on top of Ray. This allows developers to easily run scalable and distributed PyTorch workloads.
  • General Availability of Delta Live Tables: The first data integration framework that uses a declarative approach to build reliable data pipelines while managing infrastructure automatically. This means analysts and engineers will spend a lot less time on tooling.
  • Kubric: A new Python framework for generating high-quality synthetic data for computer vision applications.

If you enjoyed this newsletter please support our work by encouraging your friends and colleagues to subscribe: