COINCoinbaseNeutralCoinbase

3d ago

[scuttleslops] Cloudflare wants bots to pay; OpenAI’s ad ambitions; Memory chip competition; Muse is a proactive assistant; the Wintel of cars and robots; Atlassian as a coordinating layer

Upcoming posts (in no particular order): Otis, Gartner, Tractor Supply, Intuitive Surgical, Fortive, and Veralto What are scuttleslops? Decoder with Nilay Patel; Can Cloudflare save the web from AI? Sept. 26, 2026 Matthew Prince’s broad thesis is that Cloudflare is evolving from “the company that makes the internet faster and protects it from bad guys” into infrastructure for a web increasingly used by machines rather than humans. He describes the company more ambitiously as trying to “rebuild the internet the way it should have been built from the beginning,” using a network that spans more than 350 cities and thousands of data centers. That network already handles security and connectivity, but Cloudflare’s fastest-growing activity is increasingly the developer platform that lets customers build applications directly on its infrastructure. AI makes that physical and network footprint more strategically important because automated traffic has exploded far faster than Cloudflare expected. Prince initially thought bots and agents would surpass human internet traffic in 2027; by May 2026, Cloudflare’s data showed the crossover had already happened. His intentionally aggressive extrapolation is that machine traffic could eventually become 1,000 times human traffic—not because humans browse less, but because agents can make vastly more requests than a person ever would. That creates an obvious infrastructure opportunity for Cloudflare, but also a basic economic problem: “Bots don’t click on ads,” so the advertising model that funded much of the open web does not naturally pay for the bandwidth and servers consumed by AI agents. Prince therefore thinks the internet may need a new economic layer in which humans can access information freely while bots pay for it. Cloudflare is working with companies including Stripe and Coinbase around HTTP 402, the old “payment required” protocol, so a website could charge an agent a fraction of a cent each time it retrieves information. The analogy Prince repeatedly reaches for is Spotify: users pay for access to a large pool of content, and some portion of those economics ultimately flows back to the people who created what was consumed. His argument is less that every page should become expensive than that even a tiny price introduces scarcity into an environment where an agent can otherwise query millions of resources at effectively zero marginal cost. Cloudflare believes it occupies an unusually powerful position to make such a market work because it sits in front of more than 20% of the web. That gives it both visibility into bot behavior and the ability to let millions of websites block, admit, or eventually charge specific crawlers without individually negotiating with every AI company. Prince thinks the missing ingredient is not necessarily willingness: large AI buyers appear prepared to pay for useful information and large content owners want to charge them; what is missing is infrastructure connecting the two sides cheaply enough to create a market. This also changes which information becomes valuable. Prince argues that frontier models will gradually converge in technical capability, making proprietary access to “true knowledge” one of the more durable ways for AI services to differentiate themselves. Generic coverage of heavily reported events has lots of substitutes, whereas local reporting, niche expertise, Reddit discussions and genuinely new knowledge are difficult to reproduce if a model cannot access the original source. He gives the deliberately provocative comparison that Reddit reportedly commands several times more value per token in licensing deals than major newspapers because “if you don’t have Reddit, there’s no substitute for it.” In his view, AI could therefore reverse some of the economics that destroyed local and specialist media by assigning higher value to information that is unique rather than information engineered merely to win clicks. Cloudflare is already willing to use its control point aggressively to force the issue. Prince said Cloudflare would default its free customers to blocking Google’s crawler for AI training and, if Google would not distinguish AI crawling from search crawling, potentially block the crawler entirely unless site owners opted out of that policy. His leverage argument is straightforward: because Google’s ranking system depends on mapping relationships across the web, removing access to a material portion of the internet creates a much bigger problem than simply hiding a handful of pages. At the same time, Prince said Google has become more constructive, including discussing ways for its crawler to identify why it is requesting a page so publishers can permit search indexing while rejecting other uses. Subscribe now Read more

AMZNAmazonNeutralAmazon

2w ago

[scuttleslops] Chatbot advertising, AMZN, ROP, China EV’s and Humanoid Robots, APH

Upcoming posts (in no particular order): CoStar, Gartner, Tractor Supply, Intuitive Surgical, Fortive, and Veralto What are scuttleslops? Mobile Dev Memo Podcast; Season 7, Ep. 31: Understanding the privileged position of chatbot advertising ; September 2, 2026 Chatbot advertising could become unusually powerful because a chatbot can combine the strongest feature of search advertising—explicit commercial intent—with the strongest feature of social advertising—behavioral targeting based on the user. Search usually sees a short query, while a chatbot may see a long conversation in which the user gradually explains what they are trying to accomplish, what constraints they have, and what trade-offs they care about. The author calls this a “privileged position” because the ad system can target either the conversational task itself or the user more broadly, depending on whether the current session has obvious commercial intent. In other words, if somebody is actively planning a Disney trip, the system can use that context; if somebody is discussing a noncommercial topic, it can instead rely on first-party behavior, customer lists, conversion data, or other off-platform signals much like Meta does. That makes chatbot advertising structurally different from traditional search. A search engine historically tried to answer a query well enough to send the user somewhere else, so the successful interaction often ended with one click. A chatbot tries to keep the user inside the product until the problem is resolved, meaning the economically relevant unit can become an entire multi-turn session rather than a single query. The author’s example is useful: “Hotels on Monorail Disney World August 2025” contains obvious commercial intent, while “How can I get back and forth to Disney World without a car?” initially looks informational but may develop into hotel, transportation, restaurant, and itinerary decisions after a few follow-up questions. That means an initially unmonetizable prompt can become commercially valuable as the chatbot learns more about what the user actually wants. The author thinks Google and OpenAI are converging on the same end-state from opposite directions. Google is turning search from what he calls a “distribution mechanism” into an “engagement sink” through AI Overviews and AI Mode, keeping users inside Google longer and creating more opportunities to monetize the subsequent steps of a search journey. OpenAI started with the engagement sink and is now layering on the infrastructure of a performance-advertising business. This is already visible in Google’s rising query volumes and strong search revenue growth, even as AI-generated answers were supposed to cannibalize clicks and therefore hurt monetization. The simple “AI destroys search ads” thesis is too crude because conversational search can actually create more queries, more ad exposures per session, and more opportunities to pull commerce directly into the interface. Subscribe now Read more

WMTWalmartNeutralWalmart

3w ago

[scuttleslops] Amazon grocery, John Deere and AI, picks and pans in P&C Insurance, Copart earnings and the ACV acquisition

Upcoming posts (in no particular order): CoStar, Gartner, Tractor Supply, Intuitive Surgical, Fortive, and Veralto What are scuttleslops? The Jason & Scott Show: The Amazon Briefing, Pt. 1 ; September 10, 2026 Amazon may finally have cracked the hardest part of grocery: getting perishables into the same fast-delivery network that already made it dominant in everyday essentials. Jason Goldberg’s framing is that grocery matters less because of its margins than because of frequency: people buy TVs every few years, but they buy food constantly, and “frequency drives familiarity and loyalty.” That was Walmart’s playbook decades ago when it added fresh food to its discount stores; grocery turned Walmart from a place consumers visited occasionally into a weekly destination, and that frequency then lifted the rest of the business. The hosts argue Amazon is now attempting the same trick digitally by using grocery to make itself a more frequent part of consumers’ lives. Historically, Walmart and Amazon were both huge online grocers but were barely competing head-to-head because Walmart was strongest in perishables while Amazon dominated shelf-stable consumables such as paper towels, pet food and household goods. That changed when Amazon added refrigerated and frozen capacity to its sub-same-day fulfillment centers, allowing customers to put milk, meat and produce in the same order as the rest of their Amazon basket. Amazon effectively rebuilt a large portion of an already enormous distribution system to handle multiple temperature zones, something Goldberg argues most retailers could never contemplate economically. The early consumer response appears strong enough that the hosts think Amazon’s grocery trajectory has genuinely changed. They cite Amazon’s public disclosures that monthly active perishable customers were up roughly 50% from the start of 2026, that same-day orders containing perishables carry roughly three times as many items, and that fresh groceries now occupy six of Amazon’s top 20 best sellers. Amazon is not merely adding another category; perishables appear to be pulling occasional customers into much more frequent use and even attracting some customers whose first meaningful Amazon engagement begins with grocery. The second major theme is that faster delivery is changing the structure of grocery demand itself. Traditionally, households made one large weekly trip because shopping was inconvenient, but if an item can show up in three hours, one hour or 30 minutes, consumers no longer need to plan a 60- to 100-item basket a week in advance. Goldberg argues that “that big stock-up trip is largely going away” for many families and being replaced by smaller, more frequent orders made meal by meal or item by item. Subscribe now Read more

NeutralAVGOBroadcomBroadcomNVDANVIDIANVIDIA

3w ago

[scuttleslops] UNP-NSC merger, HEI, NVDA, AVGO, DG, DLTR

I will publish a long overdue update on TransDigm tomorrow. Other upcoming posts (in no particular order) include: CoStar, Gartner, Tractor Supply, Intuitive Surgical, Fortive, and Veralto What are scuttleslops? Union Pacific; Virtual Fireside Chat hosted by Bernstein Research; September 1, 2026 source: koyfin The core of Union Pacific’s pitch is that the proposed $85bn combination with Norfolk Southern is not primarily a cost-cutting merger; management sees it as a way to turn two separately optimized railroads into a single coast-to-coast network that can take meaningful freight share from trucks. Eliminating railroad-to-railroad handoffs should make rail faster and more predictable over long distances, which allows Union Pacific to compete for freight that currently moves by truck or does not move by rail because the service is too cumbersome. An internal crew change can take less than ten minutes, while an intermodal train handed to another railroad can sit for hours because the receiving railroad controls its own crews, slots and priorities. On carload freight, the gains can be larger still because the combined company could pre-block cars for their final destination rather than repeatedly sorting them at interchange points, potentially removing “24 to 36 to 48 hours” from a shipment. Contractual cooperation cannot fully replicate a merger: two railroads may cooperate, but they still optimize for themselves. Union Pacific CEO Vena pointed to fights over train length, locomotive availability and capital allocation as examples of seemingly mundane issues that create real friction when two management teams control different pieces of the same shipment. His argument is essentially that ownership eliminates the coordination problem: “you’re better off having one team that leads the entire railroad versus multiple partners that have sometimes their own view.” This is also why Union Pacific thinks the merger’s advantage would extend beyond intermodal; even a relatively smooth intermodal interchange is materially slower and less predictable than moving the train across one network. Strategically, management is betting that reliability itself can expand rail’s addressable market, particularly for more time-sensitive freight such as automotive parts, where poor service historically pushed customers toward trucking. The secular opportunity is therefore modal conversion: Union Pacific says the merger could remove 2.1 million annual truckloads from highways and generate $3.5 billion of shipper savings, with the latter calculated primarily from the lower cost of rail versus truck. Faster service reduces customers’ railcar inventories and improves asset turns, while lower railroad costs create the option to price aggressively where truck competition requires it. Management also thinks a coast-to-coast system opens markets that are effectively uneconomic today — for example, a shipper in the middle of the country that naturally uses West Coast ports because reaching the East Coast requires an interchange could suddenly have both options. In that sense, the growth opportunity is not simply taking existing traffic from BNSF or CSX; it is trying to make rail viable in lanes where the inconvenience of an interchange currently gives trucks the advantage. The same efficiencies Union Pacific sees as the merger’s virtue are exactly what its railroad competitors fear. Vena believes competitors oppose the deal because a lower-cost single-line Union Pacific would pressure their pricing, and he rejects the idea that regulators should preserve less-efficient routings merely to protect another railroad’s share. The more substantive regulatory question is whether shippers that are captive to rail — chemicals are the example raised in the discussion — actually benefit, or whether they simply bear the congestion and competitive risks created by a much larger network while truck-competitive shippers capture most of the upside. That is, if UP fills the network with millions of new intermodal loads converted from trucks, the chemical shipper worries that its existing rail service could actually become more congested. The second concern is losing effective rail competition. Suppose the chemical customer currently has three viable Class I railroad combinations serving it, and the merger reduces that to two — what the transcript calls a “3:2” situation. Subscribe now Read more

NeutralCRMSalesforceSalesforceNVDANVIDIANVIDIA

9/3/2026

[scuttleslops] VEEV, CRM, WDAY, SNPS, NVDA

What are scuttleslops? Veeva Systems, Q2 ‘27 Earnings Call, August 26, 2026 source: Koyfin Commercial subscription revenue grew about 13% year over year, and even excluding Crossix it was still up double digits. Most importantly, CRM itself is growing when investors had worried the Salesforce split would turn it into a shrinking asset. Management now sounds considerably more confident about taking back some large customers that previously chose Salesforce, arguing that those implementations are running into delays and complexity because “the product is very deep, and you need that.” Veeva has deliberately positioned itself as these customers’ “plan B,” and because legacy Veeva CRM remains available through the end of 2029, Gassner thinks the real win-back window opens in 2027–2028 as failed or delayed migrations become harder to ignore. Veeva says it already has more than 180 customers live on Vault CRM while Salesforce has yet to get customers live on some of its biggest projects, and Veeva is now targeting more than 70% long-term CRM market share. I think Peter alluded to earlier how Salesforce has been struggling with some of the larger projects that they have, and they don’t really have any customers live. And when you look at what Veeva is doing, it contrasts pretty significantly. We have over 180 customers live. We have customers that are turning AI on. We had a big milestone in CRM this quarter where one of our top 20 turned AI on in CRM for their entire field force. So really significant milestone. So the contrast between how Veeva is executing and what we’re seeing on the other side is very significant. As Peter mentioned, we’re very close with these customers. They buy a lot of products from Veeva. We have deep relationships. We stay very close with them. Falcon (Veeva’s AI “agentic labor” product for regulated life-sciences work) is potentially more consequential because Veeva is no longer merely selling software that helps employees do work; it wants to sell the work itself through what Gassner calls “agentic labor.” That changes the revenue opportunity from competing for an information technology budget to potentially competing against salaries and outsourced labor budgets, particularly in safety, clinical and regulatory functions. Management argues Falcon should be easier to deploy than conventional enterprise software because there is no system cutover, extract-transform-load process or large-scale data migration; if the product works, customers should reach value much faster. Gassner went so far as to say that if the first customers were already live and successful, “Falcon would be flying off the shelf.” Subscribe now Read more