B (4:07)

Speaking of algorithms, how do you feel about algorithmic pricing? A new study has found that instacart users across four US Cities saw different prices for the same items from the same store at the same time. Quoting the Times, the shoppers were volunteers participating in a study published on Tuesday and organized by the Groundwork Collaborative, a progressive policy group, and Consumer Reports, a non profit consumer publication. In tests in four cities across the country, nearly 200 volunteers checked prices on 20 grocery items on Instacart. On item after item, they found significant differences. In a Target in North Canton, Ohio, some shoppers were charged $3.59 for a jar of Skippy Peanut Butter that others could get for $2.99. At a Safeway in Seattle, some people paid $3.99 for a box of Wheat Thins, while others paid $4.89. And at a Target in St. Paul, Minnesota, some people were charged $4.59 for a box of Cheerios that others could get for $3. 99. Shoppers who are buying the exact same item from the exact same store at the exact same time are getting different prices, said Lindsay Owens, executive director of the Groundwork Collaborative. The data really backs up how extraordinarily pervasive this is. An Instacart spokeswoman said that stores on its platform set their own prices and that some of them engaged in pricing tests to learn what matters most to consumers and how to keep essential items affordable. The pricing tests are short term, randomized and designed so that people may see slightly lower prices and some may see slightly higher prices with the goal of helping retail partners understand consumer preferences and identify categories where they should invest in lower prices, the spokeswoman said. Meredith Kopit Levian, president and chief executive of the New York Times company is a member of Instacart's board of directors. A Target spokesman said the company, quote, is not affiliated with Instacart and is not responsible for prices on the Instacart platform. Instacart said that during the period covered by the Groundwork study, it was, quote, evaluating different approaches to covering its costs, but has since ended pricing tests on target orders. Groundworks findings are the latest example of how the notion of a single price offered to all customers for a predictable period is breaking down. In the digital age, companies are using sophisticated algorithms to adjust prices quickly in response to competitors, offers and consumer behavior. Dynamic pricing strategies in which companies raise prices during periods of intense demand have spread beyond sectors where they have become familiar, such as air travel and ride hailing services to other parts of the economy, including restaurants and retailers. Prices are definitely more flexible across categories than they were 10 to 15 years ago, and that has to do with the rise of E commerce and these technologies that have allowed firms over time to react more quickly, said Alberto Cavallo, a Harvard economist who has documented the rise of algorithm based pricing strategies. High inflation after the pandemic accelerated the trend, Mr. Cavallo said. By encouraging companies to adjust prices more quickly, these strategies are making prices more volatile and less predictable for consumers, research suggests. And there is some evidence they are pushing up prices, at least in certain categories, at a time when the rising cost of living has emerged as a political point of contention. Grocery prices are up more than 25% over the past five years and continue to rise faster than before the pandemic. In surveys, voters consistently rank food prices among their top affordability concerns, end quote.

B (7:37)

We're starting to get a groundswell of rumblings about this now, so it looks like it's gonna happen. Sources are telling Bloomberg that SpaceX plans an IPO in mid to late 2026, seeking to raise far above $30 billion and targeting a 1 1/2 trillion dollar valuation, which would make it potentially the largest IPO in history, Quoting Bloomberg, the Elon Musk led company is targeting a valuation of about 1.5 trillion for the entire company, which would leave SpaceX near the market value that Saudi AR established during its record 2019 listing. The oil major raised $29 billion at the time. SpaceX is exploring a possible IPO as soon as mid to late next year, people familiar with the matter said Last week. Musk and the company's board of directors advanced plans for the listing and fundraising, including hiring for key roles and how it would spend the capital. In recent days, as SpaceX firmed up its latest insider share sale, one of the people said. SpaceX's faster path to public markets is in part fueled by the strength of its fast growing Starlink Satel Internet service, including the promise of a direct to mobile business, as well as the development of its Starship, Moon and Mars rockets. The company is expected to produce about $15 billion in revenue in 2025, increasing to between 22 and $24 billion in 2026, one of the people said. With the majority of sales coming from Starlink, SpaceX expects to use some of the funds raised in an IPO to develop space based data centers, including purchasing the chips required to run them, two of the people said, an idea that Musk expressed interest in during Barren Capital. Yeah, by the way, data centers in space are all the rage all of a sudden. Sundar Pichai talks about them every chance he gets lately. This quote from Gavin Baker on the Invest like the Best podcast explains why people are interested in this quote in every way. Data centers in space from a first principles perspective are superior to data centers on Earth. In space you can keep a satellite in the sun 24 hours a day. The sun is 30% more intense which results in 6 times more irrad than on earth, so you don't need a battery. The cooling in these data centers is incredibly complicated. Space cooling is free. You just put a radiator on the dark side of the satellite. The only thing faster than a laser going through a fiber optic cable is a laser going through absolute vacuum. Link satellites with lasers and you have a faster and more coherent network than any data center on Earth. End quote.

B (10:16)

We all remember the choices that shaped the course of our lives in business. World renowned venture capital firm Sequoia Capital calls them Crucible moments. Their podcast brings you inside the pivotal decisions that defined some of today's most influential Companies hosted by Sequoia's Rule of Botha Crucible Moments Season 3 pulls back the curtain on the untold stories behind companies like Stripe, Zipline and Palo Alto Networks, Klarna Supercell, and more. I loved the recent episode with the founder of Zipline and how even late to the game, they are leapfrogging the bigger players to bring true autonomous drone delivery not just to hospitals, but now to customers of the likes of Walmart and Chipotle. I might win that burrito delivery bet someday soon. Tune in to Sequoia's new season of Crucible Moments to discover how some of the most transformational companies of the modern era were built. Crucible Moments is available everywhere you get your podcasts and@CrucibleMoments.com.

B (11:11)

Keeping pace with data growth in the age of AI is like trying to find enough shelf space after a trip to a big box store. AI and data growth have outpaced the old storage model. Manual management of traditional storage can't keep up, so it's time for a new, unified approach from Pure Storage. They help organizations simplify and automate how data is stored and managed, eliminating silos and putting intelligence at the center of operations. When you don't know where data lives or how it's used, governance slips, visibility and compliance can become constant challenges. The Pure Storage platform unifies storage into a single intelligent layer that can turn data into a governed, virtualized cloud of data with guaranteed outcomes. Learn more@PureStorage.com Morning Brew that's PureStorage.com Morningbrew.

B (12:04)

Sources have told the Information that Deep Seq is developing its new AI model using several thousand Nvidia Blackwell chips, which were smuggled into China via third party countries. Quote Deepseek's ability to access the chip via one or more Chinese data center companies underscores the difficulties the US faces in enforcing export controls in the intricate globalized semiconductor industry. Most Nvidia chips are manufactured in Taiwan and sold through a complicated network of distributors and resellers around the world. With access to Nvidia's latest chips, DeepSeek is better able to stay competitive in the AI race both in China and globally. Despite Beijing's push for local companies to use homegrown alternatives, Chinese AI chips are still not good enough for training AI models, a process in which models ingest copious amounts of data to learn connections between them, according to executives and employees of Chinese AI companies. Since it burst onto the scene roughly a year ago, Deepseek hasn't launched a completely new model and has only released incremental upgrades to existing existing ones. One factor that likely complicated Deep Seq's model development is that it has been using a method called sparse attention, which only uses certain parts of a model to answer questions rather than the entire model, according to one person with direct knowledge of the process. The technique could significantly reduce the costs of inference when AI models generate answers or perform tasks, therefore lowering the financial barrier for AI adoption. Blackwell chips are particularly useful for this approach because the chips include specialized hardware designed to accelerate sparse computing, running such calculations up to twice as fast as traditional methods. Deepseek's focus on the sparse attention technique has made its model development more challenging and time consuming, according to the person. The company in September released the v3.2Exp, which it described as an experimental model serving as an intermediate step toward its next generation model. But applying sparse attention to bigger models is proving to be more complicated, the person said. Some Deep SEQ employees are hoping to roll out the next generation model by the Lunar New Year holiday in mid February, according to the person. However, Deep Seq founder Liang Wen Feng, who prior advertiser's performance over the timeline hasn't set a hard deadline for the new model, the person said. End quote.

B (14:20)

Boom Aerospace, which is building a supersonic plane, has done a little side hustle by announcing superpower, a 42 megawatt natural gas turbine using Boom's existing supersonic engine to power AI data centers. Quoting TechCrunch aircraft startup boom Supersonic said Tuesday it will start selling a version of its turbine engine as a stationary power plant and that its first customer will be the data center startup Caruso. Caruso will buy 29 of Boom's 42 megawatt turbines for $1.25 billion to generate 1.21 gigawatts for its data centers. Boom said it will announce more details about a turbine factory next year, with first deliveries occurring in 2027. To commercialize its superpower stationary turbine, Boom raised $300 million in a round led by Darsana Capital Partners, with participation from Altimeter Capital, Ark Invest, Bessemer Venture Partners, Robinhood Ventures and Y Combinator. Profits from the sale of Superpower units will go toward funding continued development of the company's Overture Supersonic aircraft. Boom founder and CEO Blake Scholl told TechCrunch launch it's an arrangement that Shoal likens to SpaceX's Starlink satellite constellation. The satellite Internet service is reportedly profitable, helping the company to bankroll the development of its rockets. I've kind of been keeping my eyes open for 10 years for what could be our Starlink, he said. I said no to a thousand things because I concluded they were distractions. This one we're saying yes to because it's so clearly on path, boom said. Superpower and its airborne engine called symphony, share 80% of their parts earlier this year. Boom's XB1 demonstrator was civil aircraft developed by a private company to break the sound barrier. Crusoe is paying $1,033 per kilowatt of generating capacity for that. Boom will deliver the turbines, generators, control systems and preventative maintenance. Crusoe will have to provide everything else, including pollution controls, electrical connections and so on. That's on the higher side for that type of power plant. A typical airplane derived or aero derivative turbine costs around $1,600 per kilowatt, a price that also includes pollution controls, ENG construction, land acquisition permitting, pipelines and more. In a typical project, the turbine and pollution controls contribute about 46% of a project's total cost. Applying that percentage to Boom's figures would likely push the total cost over $2,000 per kilowatt. That's pricey for a simple cycle gas turbine and more. In line with costs for combined cycle gas turbines slated to come online in the early2030s, Boom's superpower is targeting 39% efficiency. Similar to competitors, combined cycle turbines can recover heat from the exhaust to boost boost efficiency above 60%. Shoal said the power plants should be no louder than existing aero derivative turbines, though that's not exactly quiet. Residents near Xai's Colossus data center report hearing similarly sized turbines from at least half a mile away. The first few stationary turbines will be made at Boom's existing facilities while the company builds a larger factory. The goal is to produce 1 gigawatts worth in 2028, 2 gigawatts worth in 2029 and 4 gigawatts worth in 2030. If boom can hit those numbers, it would significant expansion in the turbines available to be deployed. End quote so it turns out that this is actually not that uncommon. Apparently this is quoting from the Odd Lots newsletter. I'm quoting I believe Joe here. It's worth noting how this is yet another example of how some key industrial input, which was originally envisioned for something else, is plugging itself into the AI supply chain. I don't have any view on whether AI is a bubble or not. I don't have any view on whether the multiples being paid for AI companies are too high or not either. I don't have any view on whether all this capital investment will produce satisfactory returns for equity and debt investors. That's for other people to figure out. One risk though, is this crowding out effect that the AI boom is sucking up resources that might have been better allocated to something else. People often associate crowding out with fiscal policy, but when there's a huge investment boom being undertaken by the private sector with a highly uncertain payoff, then the same concept can apply. All that being said, it seems like an unalloyed good thing if AI investment helps accelerate the development of jet engine technology. In fact, if anything, this might be seen as a positive crowding in. And it's not the end of the world if some grid connections get reallocated away from bitcoin mining. Still, it's worth thinking about this general phenomenon by which capacity industrial, electric, labor, et cetera is increasingly being shifted to this one project of building out AI tech. End quote. As Deva Hazarika joked on X every VC emailing their portfolio companies today, did you see this? Have you thought about how our Tinder for Dog's product might be used by data centers? End quote.

B (19:24)

Nothing more for you today. Talk to you tomorrow.