SF is back, and we are back in SF
Gil Dibner

As I write this, I’m coming to the end of a nine-day sojourn in San Franciso. After a bit too much time away, I felt it was important to spend some quality professional time in the Bay Area just meeting with people and strengthening connections new and old. I wanted to spend some time just listening to the rhythm of the US tech ecosystem, trying to understand the current moment a bit better.

Over the past two weeks, I’ve been to Helsinki, Copenhagen, and Munich. European tech is stronger and more exciting than ever. Munich is particularly exciting right now. Israeli tech (see the article linked below) is - despite the ongoing conflict there - soaring to new heights and will, no doubt, emerge from this period stronger than ever. And yet, the US remains the ultimate technology ecosystem. It is here in San Francisco that this energy can be most easily felt. My overwhelming feeling after over a week here is that I need to be spending more time here. Probably, we all do.

It’s easy in the flurry of news about big non-US financing rounds and crazy US election-related shenanigans to conclude that maybe we non-American techies can go it alone. But it’s not true at all. While I am an Israeli living in London, I am also American enough by birth and education and experience to be able to read the signs. There is so much energy and dynamism in the Bay Area right now that anyone anywhere would be crazy to ignore it. Whatever you think about the current wave of enthusiasm for everything AI-related, that future is being written here in San Francisco at a pace and intensity that outpaces everything in London or Paris or Tel Aviv.

It is absolutely true that great companies can be started and built to greatness from anywhere, but the US in general, and the Bay Area in particular, set the tone for the entire global technology industry. The concentration of talent and experience and intensity here is simply unbeatable. Founders from Europe and Israel should strive to tap into this talent pool as much as possible if they want to get serious about scale. Helping them do that efficiently and successfully is a big part of what we do at Angular Ventures. We strive to be a bridge to success which means, almost always, a bridge to the US.

This morning, around a conference table in San Francisco, I joined a board meeting for one of our companies. The CEO had relocated to San Francisco not too long ago, and the American VC in the room was encouraging the founding team to concentrate their new senior executive hires in San Francisco as they build out the next phase of their journey. Like many of our companies, this company has many employees in their country of origin and will continue to hire there. But ultimately, it’s their presence in the US market and their US-based employees that seem to be unlocking their full growth potential.

“At the end of the day,” the American VC said, “companies that are based in San Francisco just outperform their peers by a wide margin on average.” Sitting here and feeling the energy that runs through this place, it was impossible to disagree.

No Sleepwalk to Success
Engineering success in a technical startup.

Revenue Durability in the LLM World
Everything about LLMs seems to make revenue durability more challenging than ever.

A Robotic Future for Retail Grocery, Finally.
Why we invested in Finally.

A Digital Fabric for Maritime Trade
Why we invested in Portchain.

More blogs

Denmark / ClimateTech. Again raised $43M, led by Google Ventures, for its biomanufacturing plant that turns CO₂ into carbon negative chemicals.

Sweden / LegalTech. Leya raised $25M, led by Redpoint, for its LLM-powered legal workflow and automation platform.

Italy / Cloud Storage. Cubbit raised $12.5M, led by LocalGlobe, for its distributed, green, and hyper-resilient storage provider solution.

UK / SaaS. CultureAI raised $10M, led by Smedvig Ventures and Mercia Ventures, for its human risk management platform.

Germany / SaaS. doinstruct raised $7.6M, led by Creandum, for its training platform tailor-made for frontline workers.

France / GenerativeAI. Presti raised $3.5M, led by Partech, for its generative AI-powered iamge production platform that removes the need for furniture photo shoots.

ENTERPRISE/TECH NEWS

Strike out. On Friday morning, some of the largest airlines, banks and hospitals came grinding to a halt as a faulty Windows update pushed by cybersecurity company Crowdstrike took down millions of machines. The impact of this mistake was far-reaching, and it will be, unfortunately, long-lasting. While Crowdstrike has already published a fix, it will take time to resolve because IT administrators may often have to manually reboot the machine to get it working again. A typically cogent and thoughtful take on the whole situation, with a lot of context from the early days of Windows, from Ben Thompson here.

Regulatory consequences. Meta announced that it won’t release its multimodal Llama AI model in the EU due to uncertainty around regulatory compliance. This follows Apple’s announcement that it would not release Apple Intelligence in the EU in 2024 due to uncertainty related to the Digital Markets Act. For what it’s worth, Meta’s issue isn’t with the newly announced AI Act (at least not yet), but rather with how it can train its models with customer data (e.g. Facebook or Instagram posts) while still remaining GDPR-compliant. Long term, this has the possibility of hurting European companies’ competitiveness as they will not be able to use Llama’s multimodal models despite the models being released under an open license.

Rise of the minis. OpenAI released GPT-4o mini and it’s benchmarking surprisingly well. From Artificial Analysis: “With a MMLU score of 82% (reported by TechCrunch), it surpasses the quality of other smaller models including Gemini 1.5 Flash (79%) and Claude 3 Haiku (75%). What is particularly exciting is that it is also to be offered at a cheaper price than these models. The reported price is $0.15/1M input tokens and $0.6/1M output tokens. With such a cheap price for input tokens and its large 128k context window, it will be very compelling for long context use-cases (including large document RAG).” This is a tiny model, small enough to fit on a device, which suggests we may soon all have foundational models at the ready in our phones.

HOW TO STARTUP

War for retention. Garry Tan argues that the critical challenge for LLM-powered product companies will be retention. It’s easy to sell the vision, especially with a flashy demo, but it’s hard to build a full-coverage, enterprise-ready and valuable solution. That, however, is what will drive retention.

HOW TO VENTURE

US, Israel & Europe by the numbers. Is too much cybersecurity a bad thing? So argues this piece from Calcalist. Of the $5.1B raised by Israeli startups in the first half of 2024, half went to cybersecurity companies. That being said, the momentum in Israeli tech is palpable. Israeli startups raised 31% more in the first half of 2024 than the first half of 2023. US startups raised 28%. European startups, however, raised 6% less.

Creative DPI. Sequoia has offered the LPs in funds raised between 2009 and 2012 to buy up to $861M of their Stripe shares at a valuation of $70B. This purchase, aimed at providing long standing LPs with liquidity on older vintages, will be funded by capital from more recent fund vintages. Axios has more here.

Venturification of research. Thoughtful piece from Compound’s Michael Dempsey on how the influx of capital into academia has created a “prisoner’s dilemma” for researchers who now have to decide to stay in the lab or go start (or join) a startup. This isn’t a great dynamic, as it leads to fragmentation of talent (among many other downsides), but it’s also inevitable, and capitalism will eventually identify the winners. As Dempsey writes: “hyper-fragmentation will slow progress temporarily before the waves of death concentrates talent and capital into the core 1-3 Applied Research Companies in a given area that benefit from the virtues of capitalism and competition and thus re-accelerate progress in a given area.”

Forwrd.ai announced its acquisition of LoudnClear.ai, to broaden its product offering of no code data science automation for RevOps.

Lunar.dev is cutting API costs by up to 50%. “Lunar.Dev is the first enterprise-grade API Consumption Platform designed to provide businesses with great visibility, control, and optimization of their API traffic. In an ecosystem where API dependencies are crucial, Lunar.Dev ensures that organizations can manage these dependencies efficiently and cost-effectively.”

Groundcover, Finally and Tensorleap were selected as winners of the Future 50, the world's highest-potential startups by The Generalist.

Aquant
Enterprise Business Development Representative (New York)

Januar
Enhanced KYC Specialist (Copenhagen)

LightSolver
Physicist – Laser Researcher (Tel Aviv)

Reco
Senior Data Engineer (Tel Aviv)

More jobs