SVRGN Weekly Digest #14 š«
The world is upside down, right is wrong and wrong is write. The only thing we can do is move forward.
This week weāve had the pleasure of meeting great founders and potential co-investors, as well as collaborators for our next big event. Stay tuned!
š¤¼Ā People
Eliza Warner - Investor at In-Q-Tel
In-Q-Tel is a non-profit investor with ties to the three letter agencies in the US, the ties are of vital importance to portfolio companies who are relevant to the government. Perfect example of how an innovation ecosystem can be supported by the state, effectively, without bureaucracy, in symbiosis with the venture ecosystem.
Florian Dƶtzer - Entrepreneur and MD at TUM Venture Labs
Florian is a deep tech founder at heart, having founded and run a materials company for years before joining his alma mater TUM, as head of the venture labs department. The springboard for high-tech innovation in southern Germany. UnternehmerTUM is the largest founder-supporting organization in Germany with 400 people active working with entrepreneurs.
šĀ Companies
Radical Aero - High Altitude Platform Systems
Status: Announced Seed Raise!
Source: Direct
Founders: Cyriel Nottebom, James Thomas
Why itās cool:
Inflection led the pre-seed round of Radical and shortly thereafter our friends at Scout ventures did the seed round. Coverage in Techcrunch here. Radical is building a āpseudo-satelliteā platform of persistent coverage, allowing round-the-clock high res video imagery or 5G direct-to-device connectivity, anywhere in the world (except maybe the poles).
Stealth - non-invasive mind-reading capabilities
Status: Pre-seed
Source: Deepchecks.vc - where inflection is now a member
Founders: top of their field
Why itās cool:
Measuring blood flow in the cortex non-invasively is one thing, telling what word youāre thinking of based on that time series data is another. This company can do it.
š”Ā Ideas & Science
An excellent overview of how to learn how to learn, and implement it practically in your everyday practice. The job of a VC happens to involve learning many new things, to the right level of depth, and making use of that knowledge. While digital knowledge systems are helpful, the old fashioned brain cells also need their exercise every now and again. Availability of knowledge matters for analogical thinking.
There is a trend in chip specialization that there are relatively fewer returns from specialization as compared to the general process improvements. This paper from 2019 makes the comparison with bitcoin miners, and some other chips as well. The main metric: Chip Specialization Returns is Gain(algorithm, framework, platform, engineering, and physical)/Gain(physical) and itās trending flat or even down in some cases. The main author made the comparison to LLM training cost in this post and it looks like itās a similar phenomenon. Part of it at least it is that looking at chip performance as an optimization problem, the chip design space is being exhausted at greater and greater cost as well as the software design space becoming better and better known.