Deal of the Week - From Oil Tycoons to Data Tycoons
You learn something new every week. This week it was that oil tycoon J. Paul Getty enrolled at Berkeley (but never actually attended).
A quick reminder that J. Paul Getty made his fortune in oil, hitting the big time after securing an oil concession from the ruling family of Saudi Arabia. The Getty family has had its fair share of ups and downs since, including some rather famous kidnappings.
This week, another Berkeley alum and member of the Getty family, Mark Getty, announced a major restructuring of Getty Images. The company, valued at close to $5B just a few years ago, is now preparing for Chapter 11, with the Getty family and Koch Equity Development committing up to $533M to take the business private. Quite a turnaround. Good luck to them on their next chapter!
Fun fact for the new tally I am keeping of Berkeley connections who became the richest people in the world: so far we have Masa Son (check out our old posts on him here and here) and J. Paul Getty. Let me know of any others!
Now, for a small segue from oil to data.
Many are saying data is the new oil, and I posted back in 2025 that Data is the Fossil Fuel of AI.
So it is only apt that another Berkeley alum, Ali Ansari (funnily enough, he also did not complete his Berkeley degree), is now at the forefront of the data industry.
His company, Micro1, has announced a $1B commitment to purchase data to sell to frontier AI labs. Even more remarkably, the company has now surpassed $1B in annual revenue run rate.
There is something quite fitting about going from a Berkeley-connected oil tycoon who made his fortune securing access to one of the world's most valuable resources, to a new generation of Berkeley founders doing much the same with data.
Perhaps data really is the new oil.
Understanding the Machine
This article was very interesting. Note it is fairly dense, so I will attempt to give a layman’s summary (as a layman myself).
Effectively, the idea is looking at mechanistic interpretability (understanding the machine!). One of the slightly scary things about LLMs is that we don't really understand what is happening inside them or how they make decisions. Which is sort of crazy when you think about it. We have created these insanely smart machines but, at the most fundamental level, don't fully understand how they work. A bit of a black box.
What makes this article interesting is the idea that perhaps these models are actually far more understandable than we thought, and we have simply been looking at them the wrong way. What previously looked like random associations might actually make perfect sense if we understand what the model is recognising.
The author even estimates that around 90% of the neuron poles studied have a single main function, suggesting these models might be far more readable than previously thought.
This is obviously early research, but if it proves to be true, it could have significant implications for AI safety and really advance the field. If you can understand what is happening inside the model, you have a much better chance of identifying dangerous behaviours and figuring out how to improve it.
The Great Bear Bull Market
We live in a fascinating time. The S&P 500 is near all-time highs, but beneath the surface the picture is much more uneven. Just ten companies account for 66% of expected Q3 earnings growth.
This is not just a public market phenomenon. In private markets, it is increasingly a story of the have haves and the have-nots. You hear about, or are invested in, companies raising enormous rounds, while companies going from $7M to $21M ARR year over year (3x still feels good to me!) are struggling to raise.
The difference is that, in private markets, no one knows the price someone would pay for a company that cannot raise, even if its valuation has changed materially. Nnamdi Iregbulem captures this well in his article: the valuations we see are conditional on companies successfully raising capital. As the have haves increasingly take more of the share of capital, you have a smaller and more selective part of the market driving the perception of valuations.
There is an important distinction here. Every one of those rounds could be fairly priced, and together they could still give us a misleading picture. If only the strongest companies are raising, reported valuations can go up without the underlying population of companies becoming more valuable.
Then you get a potential feedback loop. Founders and investors look at those rounds and use them as benchmarks. The exceptions start to shape expectations of what “the market” is, while the companies unable to raise are missing from the comparison.
It reminds me of Abraham Wald’s work on aircraft survivability during World War II. Military planners wanted to put more armour on the areas with the most bullet holes in the planes that survived. But this only accounted for the planes that survived. Wald worked out that the best option would be to add armour to the areas with the fewest bullet holes, as this would account for the planes that never came back. This became a classic illustration of survivorship bias.
In private markets, the announced rounds are the planes that made it back. They tell us something important, but they do not tell us the whole story.
The companies that cannot raise are also part of the market. We just do not get a press release telling us what they are worth. In a bull market such as the one we are in, the announced rounds can become benchmarks for the next round, reinforcing expectations of higher valuations. That is where the potential reflexivity comes in: a narrow slice of the market starts to shape how we value the rest.
AI Allows You To Do Fun Stuff - Thank You to All Our Readers You Rock
Quick Takes
I have posted about one of my biggest misses (so far) in venture - as a little cookie I have mentioned them in the above newsletter 🙂
Berkeley is opensource leader so good to this interview, with the founder of vLLM
Nvidia are not just buying opensource companies ($60M Engineers), providing the financing (Nvidia’s $500B finance machine) backstop and they are the most active VC in the big rounds. All in a days work when you have a whopping $190B expected operating cashflows for Fiscal 2027.
Summary by the #️⃣ & 💰:
8 Berkeley-founded companies funded
$231.5 of capital raised from the 12th October to 17th October
💡 Got any ideas or feedback on how to improve this weekly digest? Just hit reply.
Acquisitions / IPOs
📸 Getty Images. Up to $533.0M · Take-private restructuring commitment 🇺🇸 Visual content licensing marketplace. 💰 Koch Equity Development, Getty family.
🐻 Mark Getty, Co-Founder & Chairman. Berkeley attendance. Article
🗣️ micro1. $1.0B · Data-acquisition spending commitment 🇺🇸 Human expertise for AI. 💰 Citi, Hercules Capital.
🐻 Ali Ansari, Founder & CEO. BS Computer Science & Mathematics. Article
💡 Deako. Undisclosed Acquistion🇺🇸 Modular smart lighting systems. 💰 SKYX Platforms.
🐻 Derek Richardson, Co-Founder & CEO. MBA Business Administration (Haas). Article
🐻 Steve Wozniak, Co-Founder. BS Electrical Engineering & Computer Science. Article
🌾 PickTrace. Undisclosed Majority investment 🇺🇸 Agricultural workforce management software. 💰 Terminus Capital Partners.
🐻 Harrison Steed, Co-Founder & President. BS Agricultural Business & Management. Article
Closed Rounds
🧬 Soufflé Therapeutics. $135.0M · Series B 🇺🇸 Cell-selective genetic medicines. 💰 Juvia Group, TCG Crossover, BB Biotech.
🐻 Amir Nashat, Co-Founder & CEO. MS Materials Science; BS Mechanical Engineering. Article
🤖 Ambi. $26.0M · Pre-A 🇺🇸 Personal ambient AI assistant. 💰 Luminous Ventures, Huaqin Technology.
🐻 Sunny Aggarwal, Co-Founder. EECS studies (degree not completed). Article
🩺 Brellium. $23.0M · Series B 🇺🇸 AI clinical compliance platform. 💰 Left Lane Capital, First Round Capital, Digital Health Venture Partners.
🐻 Zachary Rosen, Co-Founder & CEO. BS Business Administration (Haas). Article
🗂️ Tamr. $21.0M · Growth equity 🇺🇸 AI master data management. 💰 Volition Capital.
🐻 Daniel Bruckner, Co-Founder. MS Computer Science. Article
🌐 Limited. $18.5M Seed 🇺🇸 Global business banking platform. 💰 Third Prime, North Island Ventures, The House Fund.
🐻 Hussein Ahmed, Founder & CEO. MB Haas. Article
🧫 Affinity Biosensors. $4.9M 🇺🇸 Rapid antibiotic susceptibility testing. 💰 Founders Capital Partners, Rose Ventures.
🐻 Kenneth Babcock, Co-Founder & CEO. AB Physics. Filing
🧠 CredibleMind. $0.9M 🇺🇸 Digital mental health navigation. 💰 Blue Ash Ventures, Evolve Ventures.
🐻 Deryk Van Brunt, Co-Founder & CEO. DrPH Health Informatics; MPH Epidemiology; BS Natural Resources. Filing
Date Built By Berkeley Started | Companies Funded | Total Raised ($B) |
7/8/24 | 888 | 237.682 |
Our goal is to document the startup ecosystem of Berkeley-founded companies. Please share this newsletter with any Cal Bears in your network so we can crowdsource information about all investment rounds and job opportunities.
Did we miss a company or want to announce a round? Add it here, and we will post it next week.
Do you have a job you want to post from a Berkeley Company? Add here.
Is there someone in the Berkeley ecosystem that you would want us to do a profile on? A founder, funder, or general startup person? Add here.
Built By Berkeley is not affiliated with UC Berkeley, but maybe we will be one day if we get enough subscribers….
Built By Berkeley, where we announce all the funding rounds by Berkeley-founded companies. This is a community effort, so please let us know if we missed a company here. 🐻
