The fusion of VC and Christian nationalism
Welcome to Cautious Optimism, a newsletter on tech, business, and power.
Friday. Today we’re parsing the latest jobs data, conservative anti-AI sentiment, the stablecoin spat, and closing with notes on vector databases including notes from the CEO of Weaviate. To work! — Alex
📈 Trending Up: European borders … South Korea-USA relations … John Deere’s cost basis … Eagles … ass-kissing … OpenAI building its own chips … Broadcom, after inking an OpenAI chip deal … Australian GDP … AI agents …
DocuSign, after earnings: Despite middling growth, the esig company bested expectations and is up 9% in early trading. Billings came in ahead of expectations, and the company’s net retention is back over the 100% mark. Not a barnburner for SaaS, but far from a flop.
📉 Trending Down: Job openings … Cowboys … interest rates … Danish GDP growth … chances of rate cuts in Turkey … Anthropic’s business in China … LinkedIn? … IP theft? … social media in small markets …
The labor market is weak
This morning, the BLS reported domestic jobs data, showing slower job creation than expected, unemployment continuing to tick higher, and wage growth in-line with expectations (and ahead of inflation). Here are the big numbers:
August jobs growth: +22,000, versus expectations of +75,000
June Revision: -27,000, reducing job growth from +14,000 to -13,000
July Revision: +6,000, improving job growth from +73,000 to + 79,000
Unemployment rate: 4.3%, in line with expectations and up from 4.2% in July
Long-term unemployed: 1.9 million, ~flat from July, but +385,000 compared to the same period of 2024.
Who is hiring? Healthcare companies, which expanded headcount by 31,000 in August. That figure is down from both July (+55,000) and an average gain of around 42,000.
Who is firing? The Federal government (-15,000) in August, along with wholesale trade (-12,000), manufacturing (-12,000), and “mining, quarrying, and oil and gas extraction,” which lost 6,000 jobs.
Despite concerns that POTUS firing BLS Commissioner Erika McEntarfer after revisions to jobs data in a preceding report hurt his feelings, the data doesn’t appear to be massaged. Or, if this is a massaged report, wow.
In the wake of the report, the DJIA is down a fraction, while the S&P500 is up about a quarter point and the tech-heavy Nasdaq is up two-thirds of a point. That spread makes sense, as tech companies tend to trade higher when rates are low, and at present the market is now ever-closer to 100% certain that a 0.25% rate cut is coming.
For our purposes, we can summarize by saying that the labor market is weak, showing negative momentum rather than recovery.
A short note on vector databases
Vector databases. Different from a traditional database, vector databases store information represented by vector embeddings, which show how similar pieces of data are to one another. So, if you want to know that saying “I’m for a cuppa” and “yes, I would love some tea” are very similar, you need a vector database. Vectors, and vector databases underpin much of the AI tech you use today.
Naturally, startups are building vector database technology. One such company, Pinecone, is reportedly up for sale after raising $100 million back in 2023. That surprised me, given that I had a generally bullish view of its market, so I went to my favorite vector database startup CEO, Bob van Luijt of Wevaite, to learn a bit more. (Bob was very clear that he didn’t want to impugn Pinecone or any other competitor by sharing notes with me, so if you detect anything of that sort below, pin it to my chest and not his.)
Bob argues that there is huge demand for vector database technology, citing adoption of Weviate’s open-source software (2.5 Weaviate instantiations per second globally), and commercial products, with its customer count already totalling to “many thousands.”
Given the scale of demand and our understanding that vector databases are important to the AI game, why might Pinecone consider a sale?
Bob thinks that “truly large-scale AI-native use cases don’t yet exist in the dozens or hundreds,” which means that enterprise sales in the vector database game will be a growing sport, instead of one that has already matured (in TCV terms). Bob also thinks that in some cases, vector databases were used early in the genAI boom, where they didn’t perfectly fit, leading to unsustainable customer costs. That sounds like a recipe for churn.
So, I think it’s fair to say that it’s clear that the technology (vector databases) is proven in demand terms, but is still being solidified in enterprise sales terms. Growth in overall AI usage should close that latter gap, but not before some startups fail to leap across under their own steam.
Past Weaviate and Pinecone, startups like Qdrant and Chroma are also working on the matter.
One more reason to care about vector databases comes from Exa, a search technology startup that wants to vectorize web data for quick agentic AI search. It just raised a pile of fresh funding.
Anti-AI agitation is not just a crunchy phenomenon
There’s a quiet fusion forming between some venture capital investors and Christian nationalism, which we covered here. Notably, the venture-Christians do not have a lock on American conservative thought viz technology.
The intra-conservative split on new technologies like AI was on display at this year at the National Conservatism Conference, at which Senator Josh Hawley (R-Mo), gave a speech that you can read here. Hawley argues against the pursuit of immorality, but what I want to focus our attention on are his comments on AI more generally: