BRXND is One Week Away (Plus other news) // BRXND Dispatch vol. 068
DeepSeek R1 fallout, Stargate, and more.
You’re getting this email as a subscriber to the BRXND Dispatch, a newsletter at the intersection of brands and AI.
Noah here. Before I turn it over to Luke, I wanted to remind everyone that BRXND LA is only a week away. Thankfully, this weekend’s wet weather brought much-needed relief to the fire-scarred city and a long-awaited opening for those who’ve been displaced to begin to take stock and plan for the future.
The favorable change in weather also means the BRXND LA conference should run smoothly next week. (A minor consideration in the grand scheme of things, I realize, but important for those who are coming.)
I’m incredibly excited about this year’s event. For LA we’re going to go from macro to micro, starting with a focus on the broader landscape with talks from myself on getting AI done inside large organizations, Dan Frommer of New Consumer on consumer AI sentiment, James Cham of Bloomberg Beta on investing in enterprise AI, Lance Martin of Langchain on what’s real and not real in agents, and Howie Liu of Airtable on delivering AI products.
After lunch, we’ll switch gears and dive into the nitty gritty of delivering inside marketing organizations with talks from Zillow, Amazon, Blue Diamond, Gatorade, and others. As well as some amazing demos. I encourage you to check out the full agenda.
If you haven’t bought your tickets yet, we still have a handful spots left. Like past events, we expect to run out of space in the homestretch, so I recommend booking as soon as you can if you want to be there.
If you have any questions at all, please reach out to let me know. If you are looking to bring a group of folks, also reach out, as we are offering some group discounts.
— Noah
Join us—alongside leaders from companies like Airtable, Amazon, Apple, Bloomberg Beta, Blue Diamond, Cash App, CW Network, Diageo, Gatorade, Getty Images, Land’s End, Meta, Netflix, OpenAI, Pfizer, SC Johnson, Square, Suntory, Zillow, and more—next week in LA.
What caught my eye this week (Luke)
The timing couldn’t be more perfect. Stargate and R1 dropping in the same week feels like watching two different visions of AI’s future unfold simultaneously. Within days of one another, you have OpenAI announcing a Trump-backed plan to invest $500B in AI infrastructure in the US and China’s DeepSeek shipping a free, open-source model that matches the performance of OpenAI’s most advanced publicly available reasoning model at 98% lower cost.
Silicon Valley’s reaction to these developments tells you everything you need to know. While Stargate’s massive budget was met with skepticism, R1 sent tech giants into what one Meta employee described as “panic mode.” (Let’s not forget that Meta announced their own $62B capex plan, a mini-Stargate of sorts, last week, so it is hard to say how much of that “panic” is shared by Zuck and co.) By Monday, DeepSeek had dethroned ChatGPT as the top app in Apple's App Store, sending shockwaves through the stock market.
I don’t want to extrapolate too much here, but the juxtaposition of these two events raises interesting questions about the level of investment needed for and the defensibility of LLMs. OpenAI is betting that with the combined fortunes of Softbank, Oracle, and others behind them, they can build technical moats deeper than any individual model. Meanwhile, DeepSeek is challenging the very notion that compute is everything. They’re showing how creative engineering—specifically reinforcement learning that enables AI to teach itself through trial and error, much like humans develop new skills—could power the next wave of adoption.
The takeaway seems to be that as AI becomes increasingly commoditized and switching costs plummet, the differentiators will shift from raw model performance to things like user experience, developer ecosystems, and brand strength. I look forward to delving into these topics with you next week, all at BRXND LA…
For more on Stargate, DeepSeek, models, and moats, I found the most recent episode of Stratechery with Daniel Gross and Nat Friedman incredibly illuminating.
OpenAI launched “Operator,” a bot that can do web-based tasks like sending emails and ordering food for you. The company is calling the tool an “agent,” but naturally, not everyone agrees.
Anthropic’s new Citations API makes it much easier for LLMs to incorporate direct quotations from underlying source material into responses to user questions. It’s also great for RAG systems and potentially a sign that Anthropic is shifting focus a bit to the enterprise.
Benedict Evans offers a thoughtful exploration of what “better” really means when it comes to AI models, arguing that the current sweet spots for AI are in domains where mistakes are easy to spot and fix (like software development) and fields where “wrong” answers don’t really exist (like marketing and creative work). He also gets into the fundamental tension that exists between how we've been trained to think about computers (as deterministic systems that give correct answers) and how AI actually works (as probabilistic systems that approximate)—which is something Noah has also written about.
Sam Lessin’s advice for founders coming of age alongside AI: hone your expertise within a chosen vertical, embrace storytelling as a tool for growth, and learn to hire and manage talent, because the ability to code by itself isn’t the superpower it used to be.
As always, if you have questions or want to chat about any of this, please be in touch.
Thanks for reading,
Noah & Luke