Deal of the Week

It seems we may be entering the compute crunch that many people have been forecasting.

I have noticed Anthropic’s Claude has been a little slow recently, even as revenues appear to be going vertical. This may simply reflect a comparative lack of compute as inference demand explodes. It is also interesting that Sam Altman has recently been touting how aggressively OpenAI invested in compute early on, and how that is now proving to be a strategic advantage.

We are seeing the rising compute crunch reflected in the current price of GPU’s that spiked nearly 30-40% over the last month or so.

All of which brings us to our Berkeley Deal of the Week — a platform trying to help ease this bottleneck.

SiFive just raised raised a monster $400M round aimed at lowering the energy constraints and cost of silicon. Congrats to Yunsup Lee and the team.

Cyber Explosion

Back in November I wrote about the Golden Age of Cyber ushered in by LLMs. My thinking was that these models dramatically expand the attack surface: millions of potential vectors, easier entry points through highly realistic phishing (text, voice, video), and far more efficient ways to probe code for vulnerabilities.

That trend seems to be accelerating and so does the insane growth rate of Anthropic they ended 2025 at $9B and they are now at $30B in run rate. Let that growth sink in, because it is unprecedented they added more run rate than the total revenue of Databricks + Palantir in a couple of months. WOW.

Anthropic recently announced Project Glasswing, partnering with leading companies to use its new Mythos model to hunt for security flaws. The model reportedly identified vulnerabilities in 27-year-old critical infrastructure. More interestingly, it appears particularly good at chaining together multiple small, seemingly inconsequential vulnerabilities into something meaningful.

Fascinating — and slightly terrifying. There’s a good breakdown of it here.

A cynic might say this is clever marketing, or a way to delay as they add more compute. Luckily, I am an optimist 😀

Meanwhile the big cyber stocks have sold off this year: CrowdStrike (~-20% YTD) and Palo Alto Networks (~-15% YTD). To me this feels somewhat overdone, and I am thinking about dipping my toe in these waters.

These are massively important companies with enormous surface areas to defend. If they disappear, most of the rest of the software ecosystem probably disappears with them (which, admittedly, could happen 😁).

Is Meta Finally Showing a Return for all that CAPEX spend?

Meta has been lampooned recently for over spending on CAPEX, they are spending similar amounts to Google/Amazon, and Microsoft but with no cloud revenues. But we may be seeing the tides turning with a release of a new AI model that seems to be garnering good reviews and to the above compute constraints Meta have a ton of compute.

I am extremely bullish on Meta in this AI wave, they are well set up:

On top of this they have some interesting optionality in the meta glasses. It is crazy that you hear people say they like Instagram just for the ads now, maybe all that AI compute is delivering better targeting.

A nod to the bear case that Meta is maybe entering it’s big tobacco moment.

Quick Takes

Wow — and congrats to the Berkeley startup. Aravind’s Perplexity is also on a heater with its new computer-use product, and they’re now reportedly on a $500M run rate. I hear they’re building a huge B2B business as well.

I’ve always thought Perplexity was an execution and product machine fighting against the tide of the big model companies, and it increasingly seems like they’re simply out-executing.

💡 Got any ideas or feedback on how to improve this weekly digest? Just hit reply.

🧠 SiFive. $400.0M Series G 🇺🇸 Custom semiconductor platform. 💰 Apollo Global Management, Coatue Management, T. Rowe Price

🐻 Yunsup Lee, Co-founder & CTO. PhD & MS EECS Article

🔋 Coreshell. $46.3M Series B 🇺🇸 Advanced battery materials. 💰 Foothill Ventures, Helios Climate Ventures, Industry Ventures

🐻 Jonathan Tan, CEO & Co-Founder. BS Chemical Eng Article

🧬 Iambic. $27.8M  🇺🇸 AI drug discovery platform. 💰 ARK Ventures*, Coatue Management, Pegasus Tech Ventures

🐻 Thomas Miller, Co-founder & CEO. Post Doc Article

⚙️ Hybron. $20.0M Seed 🇺🇸 Composite industrial components. 💰 Marque Ventures, Northrop Grumman, Ultratech Capital Partners

🐻 Aaron Guo, Co-founder & CTO. BA EECS Article

🚽 wasted*. $3.2M Seed 🇺🇸 Circular sanitation technology. 💰 Collaborative Fund, Day One Ventures, Gratitude Railroad

🐻 Taylor Zehren, Co-founder & CEO. MPH Article

📦 Two Boxes. $3.2M Early Stage 🇺🇸 Returns management software. 💰 Assembly Ventures, Matchstick Ventures, Range Ventures

🐻 Kyle Bertin, Co-founder & CEO. Haas MBA Article

📊 Ridge AI. $2.6M Seed 🇺🇸 Embedded analytics AI. 💰 Madrona Venture Group, The House Fund

🐻 Jeffrey Heer, Co-founder & Chief Scientist. PhD & MS EECS Article

🏭 Muun AI. $0.7M Seed 🇸🇬 Industrial intelligence software. 💰 Wavemaker Impact

🐻 Kathryn Knight, Founder & CEO. BA Article

🛡️ Trussed AI. $0.1M Accelerator 🇺🇸 Enterprise AI governance platform. 💰 BrokerTech Ventures, First Trust Capital Partners*, ManchesterStory

🐻 Branden McIntyre, Co-founder. Haas MBA

🏦 Marr Labs. Undisclosed 🇺🇸 AI mortgage automation. 💰 Rebel Fund, SurgePoint Capital, Y Combinator

🐻 Dave Grannan, Co-founder & CEO. Haas MBA

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.

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. 🐻