Hi Shopifreaks
I know I say this a lot, but… I’ve got a really great edition for you this week! I can’t control the news, I just report it, and some weeks are inevitably more eventful than others. This week’s edition will keep you entertained.
Also, if you’re interested in learning more about agentic commerce, check out my AMA video series with ReFiBuy CEO Scot Wingo. So far we’ve covered topics including:
- Is being recommended by AI the new Buy Box?
- What does an e-commerce stack look like in 2030?
- Is Shopify or Amazon more vulnerable over the next 10 years?
- Will agentic commerce lead to a race to the bottom on price?
- A question from Commerce CEO Travis Hess for Scot Wingo on closed ecosystems.
You can follow the rest of the AMA with Scot on my LinkedIn page or YouTube channel.
In this week’s edition I cover:
- Mayor Mamdani vs Amazon
- Phoebe Gates knew the whole time
- Perplexity’s disdain for ads
- USPS inches ahead
- eBay staying the course in the U.S.
- Sezzle goes for national bank status
- Amazon says ‘no’ to class-action lawsuits
- Amazon and Flipkart’s inventory win in India
- E-commerce’s most sought-after role
- Google AI Overview mishaps
- Plus, Amazon’s secret data center
All this and more in this week’s 291st Edition of Shopifreaks. Thanks for subscribing and sharing!
Stat of the Week
Alphabet and Amazon made 70% and 65% of last quarter’s net income from stakes in other companies rather than from their own businesses. Most of Alphabet’s came from SpaceX, while Amazon’s was a $50B gain it credited primarily to Anthropic. There’s a lot of Big Tech stock value riding on each other’s success, which doesn’t feel like a very healthy or competitive market.
1. Mayor Mamdani wants Amazon and FedEx to actually employ their delivery drivers
New York City Mayor Zohran Mamdani is backing a city council bill that would require Amazon, FedEx, and other big delivery operators to employ drivers directly instead of using subcontractors, according to the Wall Street Journal. Under the Delivery Protection Act, warehouse operators would need city licenses, and getting one would mean meeting safety, training, and employment standards that put thousands of contract drivers on payroll.
Mamdani said in a statement:
“Corporations like Amazon build billion-dollar business models by insulating themselves from accountability through a system of exploitative subcontracting.”
Mamdani also said the bill could help reduce traffic accidents, citing a 2025 city comptroller report, which determined that injury-causing crashes rose in 78% of neighborhoods after last-mile facilities opened nearby. Though I’m not exactly sure how, if it’s the same drivers who are just employed differently.
Amazon said it would consider relocating its warehouses outside of New York City if the bill passed and warned that the measure could put more than 5,000 delivery jobs at risk — but I think they’re bluffing. Why would their jobs be at risk? Would Amazon not hire those same contractors as employees in order to continue serving NYC residents? In fact, the bill includes an employee retention clause requiring operators to offer jobs to the contract workers displaced by the change, so those 5,000 people would be first in line for the payroll positions Amazon says it doesn’t want to create.
Besides, what else are they going to do… move their operations to New Jersey? Last week, I reported that New Jersey Attorney General Jennifer Davenport is suing Amazon over the practice, so New Jersey might not be a safe haven for it much longer either.
Amazon also noted that its Delivery Service Partner program has paved a path to business ownership for operators (so they’re saying it’s a good thing), though I’d argue that those unnecessary operator profits could’ve gone directly to the drivers without the liability-evading middlemen. Plus, Amazon exploits the shit out of those DSP operators anyway, so no loss there if that model gets dissolved.
If the bill passes, New York would become the first city to require last-mile delivery operators to hire their drivers directly.
2. Phia founders knew they were cookie stuffing and even pushed for the practice
Remember a few weeks ago when the news broke that Phia, the AI shopping startup co-founded by Phoebe Gates and Sophia Kianni, was overriding other affiliates’ referral codes during checkout and inserting its own, claiming credit and commissions on purchases its browser extension played no part in driving? Pepperidge Farm remembers…
The company claimed at the time that a recent code release had caused the misattribution for a subset of users and that its team worked overnight to resolve the problem once notified. So just a simple mistake that could happen to anyone, right? Including Bill Gates’ daughter. No foul play happening, right? WRONG!
It turns out that Gates and Kianni actually pushed for those commission-stealing features, according to internal communications and people familiar with the matter. Bloomberg reports that the founders knew about the practice for at least seven months going back to December, despite claiming that they had just become aware of the situation “within the last 24 hours.”
I can only imagine the lawyer’s defense on this one… “No your honor. When my client said that, she was referring to the fact that she just found out they had been caught within the last 24 hours.”
In fact, what Phia called a “bug” was actually an internally developed feature called “enable coupon auto drop” that could be turned on and off remotely by the company. Gates had conversations with developers about the feature from at least Dec 18 onward after becoming concerned that the browser extension wasn’t generating enough revenue from its retail partners, asking for confirmation that the feature was turned on.
Wait, it gets even worse….
There was also a feature called “passive trigger” that was active from October 2025 to July 2026 that automatically dropped a cookie every two hours on any “top 1000 website” where the user had previously interacted with Phia!
Well, this couldn’t have inflated sales by that much, right?
Wrong again! Bloomberg discovered that the cookie stuffing practice was responsible for a significant portion of Phia’s attributed sales. After the company disabled the features in early July, average daily revenue dropped from around $80k to between $10k and $28k, according to an internal revenue chart seen by Bloomberg. And you’re telling me no-one noticed? Okay…
So in other words, Phia was committing fraud, intentionally stealing sales to inflate their sales figures to investors. The New York Post reports that Phia’s actions could be treated as federal wire fraud and carry a maximum penalty of up to 20 years in prison for the founders, plus fines and restitution. Well, maybe if she was poor. Personally, I hope they don’t get charged because I don’t want to deal with the aftermath of Gates’ dad releasing biowarfare mosquitos into the world in a vengeful rage.
Honestly, the whole scandal is flabbergasting. I can’t tell you how many times my eyes bugged out and I said “Are you serious?!” when reading that Bloomberg report. What a story!
Partner News Updates
DataFeedWatch added support for Shopify Metafield Translations and Metaobjects, which means product data that used to be stuck inside Shopify can now flow into your advertising feeds. If you’ve built translated metafields, DataFeedWatch will pull the correct value for each language automatically, so you’re no longer maintaining duplicate copies of content you already translated once. Metaobject support imports the Display Name, which you can map into titles, descriptions, custom labels, or any other attribute, so structured data like materials or size guides reaches your campaigns instead of hitting Google Merchant Center as an unusable string. Both features pair with DataFeedWatch’s existing Shopify Markets integration for market-specific pricing, currencies, and availability.
Seguno now converts customer data from your old e-mail platform into Shopify metafields when you migrate, so the preferences, quiz answers, and profile data you’ve built up over the years land in Shopify rather than staying locked in a system you’re leaving. From there Seguno can drop a customer’s data straight into an e-mail, swap in dynamic content based on it, and fuel segments and Flow automations. Metafield data written by other apps works too, so loyalty balances, birthdays, and preferences already sitting in Shopify are usable immediately. For example, a tag can tell you someone is a loyalty member, while a metafield tells you they have 847 points, information you can use to personalize your campaigns. And because the data lives in Shopify rather than alongside it, the migration makes everything else sharper too, from Sidekick results and report filtering to unified online and POS records.
3. Perplexity blocks AI-exclusive ads served by Time.com
Time.com recently began serving ads inside the markdown versions of its webpages for AI agents to see rather than human readers, and Perplexity, which seems to have a problem with anyone that can actually make a business out of advertising, began blocking those ads.
The company’s chief communications officer Jesse Dwyer called the markdown ads “deceptive advertising” and said publishers who deploy them risk a reputation downgrade in its search index, including a hit to their trust score. Oh no! You mean 0.4% of Internet searchers who use Perplexity might not see my website? How will I survive? LOL.
How do ads for AI agents work?
Time serves two versions of the same page — an HTML version for humans and vampires and a markdown version for AI, which is cleaner and easier for a model to digest. Mobian, the company that powers Time’s ads, takes a brand brief from an advertiser, generates FAQ-formatted text from it, and inserts that text into the markdown version. Now, an agent reading a Time article about retirement planning, for example, also reads Q&A markdown containing Ally Bank’s approved messaging about itself. From there, Mobian tracks how often AI search engines surface the content and how favorably they treat it.
Wait, so what’s the difference between an AI ad and every other Q&A markup a brand publishes about itself?
Apparently, where that content lives impacts how an AI model treats the claim. If it lives on the brand’s own website, the model treats it as a company claim. If it lives inside a Time article, the model gives it more editorial credibility. Or so I’ve been led to believe, because who the fuck actually knows how these AI models think? Time was transparent about the fact that this markdown content was ads and labeled it as Sponsored, but Perplexity objects to the practice either way.
What exactly is Perplexity’s problem with businesses that earn revenue from advertising?
Perplexity also recently blasted Amazon for wanting to block its Comet browser from shopping its website, claiming Amazon just cared about its $60B ad business. So what if they do? Now they’ve got their panties in a twist because Time, which has likely had its traffic from search engines eaten away by AI like every other publisher, has gotten creative with how it serves ads? Also, I fail to see what’s “deceptive” about ads that are clearly labeled as “Sponsored.” If Perplexity doesn’t know how to handle sponsored content within its models, that’s on them, not on Time.
Does Perplexity hate ads because it failed at building its own ad business? If you recall, Perplexity launched sponsored follow-up questions in November 2024 with a handful of advertisers like Indeed and Whole Foods, but only made around $20k in revenue from the endeavor, which it ended 15 months later. After abandoning the program, the company decided to take a hard stance against AI advertising, claiming that even labeled ads make users “start doubting everything” — a convenient stance to take after failing at an ad business.
My vote: If ChatGPT, Google, Amazon, and other AI companies can target human users with their own advertising, then it’s fair game for humans to create ads that target AI models — with disclosure on both sides, of course. What are your thoughts? Hit reply and let me know.
4. USPS took in $588M more revenue in the last quarter while handling 55M fewer packages
“Good news, everyone!” said Professor Farnsworth. USPS took in $588M more revenue from shipping and packages in the quarter ended June 30, a 7.7% gain, while handling 55M fewer parcels. More money for less work. Postmaster General David Steiner compared the approach to that of airlines and grocers, which price to maximize returns rather than units sold.
Steiner said:
“The results this quarter show the strong leverage that pricing can have on results, and pricing is one lever that we have to use now to grow revenue.”
Most of that shipping gain came from an 8% temporary transportation-related surcharge USPS implemented on April 26, which is set to expire January 17, 2027.
Normally I wouldn’t be celebrating USPS price hikes for small business owners, but at this point it’s ride or die for the postal service, which operates at a significant loss each year, with likely no government bailout or support headed its way anytime soon — at least while the current administration is in office. Plus, whatever price hikes you see now are nothing in comparison to what you’d see if U.S. delivery became completely privatized. Businesses would be gouged!
USPS reported:
- $19.9B in operating revenue for Q3, up 6.1% YoY
- Net loss narrowed to $2.5B from $3.1B
- Shipping and packages generated $8.25B in Q3, up $588M or 7.7% YoY
- Package volume decreased by 55M or 3.4%
- First-Class Mail revenue increased 4.3%, as volume fell 3.5%
Steiner argued for major changes to regulation during the USPS Board of Governors open session:
“As things stand, the Postal Service is expected to be self-sustaining while, at the same time, fulfilling mandates that are inherently unsustainable and do not cover their costs. We need to fix the business model that has produced the 17-year-long imbalance in costs and revenue — and that is going to require Congressional involvement.”
5. eBay says it has no current plans to push U.S. seller fees onto buyers
eBay CEO Jamie Iannone said the company has no plans to stop charging fees to U.S. sellers and bill buyers instead, when asked by analysts on the August 5 earnings call whether the European approach would work at home. He said on the call:
“As it relates to the U.S., we have no current plans to change the fees as it relates to eBay. Depop already has a buyer fee model, which is complementary to kind of that discovery-led type of experience. And so that’s kind of built into that pricing change. They made that change quite some time ago. And so overall, we feel like we’re in a very strong competitive position.”
Why in the world would you charge buyers instead of sellers?
It’s a trend I could never quite wrap my head around, but several marketplaces have made the move in recent years including eBay UK & Germany and Depop U.S. and UK. Others have tried and failed, including Mercari, which eliminated seller fees entirely in March 2024 and pushed everything to buyers, but then reversed the move in January 2025. Then there was Poshmark, which for some reason didn’t learn from Mercari’s disaster in the U.S., and attempted its own fee shift in October 2024, before reversing it less than three weeks later.
Vinted, a Lithuanian peer-to-peer marketplace for secondhand clothing, kicked off the trend when they launched in 2008, never charging sellers from day one and instead building a “Buyer Protection” fee into their business model. I can see how the fee structure works when built into the business model from the beginning, but I also see how it’s a very hard pivot when a marketplace already has an established buyer and seller base. Personally though, as a buyer, I don’t want to be charged a visible fee for the privilege of shopping on a marketplace, even if it works out the same in the end.
Either way, it seems to be working for Depop, so more power to them. And as the new owners, eBay gets the best of both models.
6. Sezzle wants to be a real bank too, like every other fintech
Sezzle plans to pursue a national bank charter instead of its prior plan to become an industrial loan company. CEO Charlie Youakim said that new BNPL regulations in states like New York, Illinois, and Oregon prompted the company to change course.
What’s the difference?
A national bank charter comes from the Office of the Comptroller of the Currency and makes you a full federal bank, subject to Federal Reserve oversight and the capital requirements that come with it. The payoff is that a national bank operates under federal rules, which means it can export its home-state rates and product terms into every state, under one regulator instead of fifty. The tradeoff is that approval is slower and the scrutiny is heavier.
An industrial loan company is a state charter, typically out of Utah or Nevada, that carries FDIC insurance without triggering Federal Reserve holding-company supervision. The charter lets a commercial parent own a bank without becoming a bank holding company. It’s also why community banks and consumer groups keep calling it a loophole and lobbying to close it, and why Congress has bills pending to do so.
For Sezzle, becoming an industrial loan company was the faster route, but the national charter is a better long-term solution that will let it avoid the kind of state BNPL laws that Youakim is worried about, and it avoids a potential political fight in the future if Congress closes the industrial loan “loophole.”
Every BNPL and fintech wants to be a bank now.
In recent months I’ve reported that Klarna, PayPal, Affirm, and others have applied for bank charters, though they all went the industrial loan route that Sezzle is now walking away from. It’ll be interesting to see if those other players change course too.
Either route they choose to take, I’m in favor of more banks. The industry needs more competition for the benefit of consumers. There are around 4,379 FDIC-insured banks and savings institutions in the U.S. today, down sharply from more than 14,000 in the mid-1980s after decades of consolidation. The industry has long been dominated by the same large incumbents, with only a handful of new banks chartered each year until a recent small uptick began in the early 2020s. It’s estimated that the top four U.S. banks, JPMorgan Chase, Bank of America, Wells Fargo, and Citi, hold roughly 45% of all U.S. bank deposits, and the top 10 banks hold a 65% share.
The space is ripe for competition and innovation, which can only come from dishing out more banking charters. The banking industry is overdue for a competitive correction. Love him or hate him, Elon Musk’s X Money is positioning itself to be one of those disruptors.
7. Amazon reinstates binding arbitration for U.S. customers
Amazon reinstated binding arbitration for its U.S. customers on Friday, barring them from bringing class-action suits against the company. The change took effect immediately, and Amazon told customers by e-mail that continued use of its services counts as agreement, where companies usually give weeks of notice.
Amazon had dropped arbitration five years ago after law firms hit it with roughly 75,000 claims over Alexa recording users without consent, and has since directed customers to Washington state court. Its new terms treat 25 or more claims over the same matter within six months as a mass arbitration, resolved in batches of at least 25. A spokesperson called it a faster, cheaper route that still leaves small claims court open. Disputes begun before Friday aren’t covered.
A nameless, faceless, dickless Amazon spokesperson said in a statement:
“We determined that reinstating the arbitration clause will offer customers a fast, cost-effective way to resolve disputes while still giving them the option of going to small claims court.”
Translation:
“We got sued too effectively last time and learned that forcing customers into one-off arbitration where nobody can pool their case is a lot cheaper than letting them sue us as a group.”
Frankly, this should be illegal. There should never exist a contract, under any conditions, that precludes a user or consumer from taking a company to court. Guns and lawsuits are our God-given rights as Americans! USA! USA! USA! But seriously, how dare they?
Remember when Noncompete Agreements were a thing for professions like hair stylists and factory workers, and then former FTC Chair Lina Khan swooped in like a fucking superhero and was like, “NOPE!”? The FTC, under her leadership, attempted to ban nearly all noncompete agreements nationwide with a single rule in April 2024, which would have freed an estimated 30M workers from contracts that mostly served to keep wages down.
Unfortunately, a federal judge in Texas killed that rule before it ever took effect, and the FTC dropped its appeal last September after Andrew Ferguson had taken over Khan’s position.
Anyway, that’s what needs to happen with mandatory arbitration clauses too, on a federal level. To be clear, I have no problem with Amazon, or any other company, offering voluntary arbitration as a pathway to resolving conflicts. My issue is with making it a requirement, and even more so, waiving consumers’ rights to pursue claims as a class, which Amazon’s batching mechanism is designed to prevent.
If arbitration really is faster and cheaper and better for consumers, companies can offer it after a dispute arises and consumers will choose it. If that’s not the case, then we have a right to sue. “Hey, hey, ho, ho, the condition that prevents us from pursuing legal recourse via collective arbitration or class-action lawsuit has got to go!” Okay, well, I’ll keep working on the chant.
8. India to let Amazon and Flipkart own inventory for the first time since 2019
India will let foreign-owned e-commerce companies like Amazon and Flipkart own the goods they sell on their marketplaces for the first time in seven years, as long as every item is exported, under a new framework that sits under the Foreign Trade Policy 2023.
Quick Backstory: A marketplace owning inventory that it sells on its own platform (thus competing against third-party sellers with an advantage) has been banned in India since a law took effect in February 2019, which is why Amazon and Flipkart exclusively operate as marketplaces in the country rather than retailers.
The law led to Amazon unwinding Cloudtail, the seller that moved Amazon’s own inventory onto the platform, in 2022, and cutting ties with Appario, which took over that role, in 2023 after regulators went after the preferred-seller arrangements that let it sell its own goods indirectly. (Cloudtail and Appario together drove about 35% of Amazon India’s total sales in 2019.) Basically, Amazon thought it was being smart by routing its inventory through sellers it partly owned instead of owning the inventory outright under its own name.
Now the government has carved out an exception, but only for goods leaving the country. Amazon and Flipkart can buy inventory from Indian manufacturers, warehouse it, and ship it abroad through their own supply chains, though they have to do it through separate legal entities that the Directorate General of Foreign Trade will monitor. Amazon and Flipkart still can’t sell the inventory direct-to-consumer on their own Indian marketplaces, so the protections that were originally put in place to protect smaller retailers remain intact.
The reason for the change is that India wants $200B to $300B in e-commerce exports by 2030 and isn’t anywhere near making that happen. The obstacle isn’t necessarily lack of demand for Indian goods, but rather that small manufacturers have difficulty handling customs clearance, foreign warehousing, and international returns on their own. Whereas, letting Amazon own the inventory moves that burden onto Amazon.
CAIT, the trade body representing India’s small retailers, objects to the new rule on the grounds that nobody has explained how the government will verify that export-bound inventory doesn’t secretly end up in a domestic warehouse. I’d also argue that a better solution would be to create a government-assisted entity that actually helps India’s small manufacturers handle those types of export logistics instead of passing the buck to Amazon and Flipkart, giving them the same international advantage that those mega-retailers have in place — but once again, nobody asked me first.
9. Other e-commerce news of interest
Wix launched Symphony, a standalone AI agent platform to help SMBs grow with a team of agents built around their goals and workflows. The platform’s central AI agent, Maestro, understands what is happening across the business, identifies what needs attention next, and coordinates the right specialized agent for each job. Once the team of agents is established, they can execute tasks like replying to clients, chasing unpaid invoices, running marketing campaigns, providing updates and analytics, and flagging key actions for approval. Symphony is platform-agnostic and doesn’t require that a business use Wix. I don’t know about you, but I don’t want any AI-powered agents e-mailing my clients or running marketing campaigns, as I’ve yet to meet one that isn’t absolute hot garbage. Though I do think it’s smart for Wix to expand its product portfolio beyond website builders.
TikTok Shop is becoming a standalone job at consumer brands rather than one more channel on a social manager’s plate or work farmed out to an agency, according to Julia Waldow at Modern Retail. Brands like Mammoth Brands, Stila Cosmetics, BaubleBar, and Jool Baby are all recruiting for the position with postings asking for P&L ownership, forecasting, and inventory coordination on top of creator partnerships and livestreams. Filling the role is proving to be difficult for many brands because people with real experience are working at agencies that pay big commissions and get first dibs at TikTok’s own programs and tools. TikTok Shop only launched in the U.S. three years ago, but already accounts for 2% of U.S. online retail spending as of July, per Consumer Edge, up from 1.2% a year earlier.
Google added Venmo as a payment method on Google Play in the U.S., covering games, apps, add-ons, subscriptions, and tips sent to creators inside content apps. Buyers link a Venmo account through the payment settings in their Google account, then pay from the Venmo balance or from a bank account or card attached to it. Google Play already took PayPal, which owns Venmo, along with Cash App and cards from American Express, Visa, Mastercard, Discover, and JCB. Outside the U.S., Google Play also lets users pay for digital goods with cash at a nearby store. I worked on an e-commerce store in Colombia many years ago, in which more than half of their customers paid that way, which came as a surprise to me at the time.
Google’s AI Overviews are spreading misinformation about some small businesses, with little that the owners can do to correct it. Business Insider shared the story of a UK plastics supplier that AI Overviews said had “overwhelmingly negative” feedback, which it backed up with complaints connected to a different company altogether. Even worse, the business owner was paying Google £700 a month for ads, while the summaries were warning customers to go elsewhere. Other business owners shared stories about their AI Overview company descriptions being conflated with other similarly named businesses or declaring that they service a different city or state. The bad news: there’s nothing you can do about it and no-one to contact at Google. The good news: Google made record profits in its most recent quarter. Ah sorry, I meant “good news for Google.” You’re still fucked.
Shopify is planning to promote its Shop app through a longer, multiphase holiday campaign running from now to December, the first time it has pushed the app this hard into the season, according to Jess Jacobs, VP of consumer marketing and head of growth for Shop. Jacobs told Retail Brew that the company doesn’t want to “inundate” shoppers with the same message over and over again throughout the entire holiday season, which is much longer nowadays given that Pre Black Friday Sales begin in October. Shop app started in 2020 as a package tracker and has since evolved into a customer acquisition channel with AI search and a shoppable feed. Ultimately the marketplace that Shopify won’t admit is a marketplace.
OpenAI refused to refund a UK agency charged for ChatGPT ads that kept delivering after it paused its campaigns, according to The Register. Excel4Business managing director Ed Bolton said ads ran overnight on paused U.S. and Canadian campaigns, and an OpenAI support specialist first agreed the campaigns had kept serving because a separate ad-level status hadn’t refreshed, calling it a defect its engineering team had escalated and promising to review the charge with the billing department. However, days later, OpenAI reversed itself and denied any credit under Section 11.1 of its advertising terms, which allows delivery to continue for a full business day once a campaign is paused or canceled, with the advertiser still liable. Luckily, the charges were under £70, so the company isn’t about to go broke over it, but OpenAI should treat situations like this with white gloves if they want companies to increase their daily budgets to five, six, and seven figure spending at some point.
Google is extending Ask Advisor, its AI assistant for marketers, across Google Ads and Analytics to help users obtain faster insights and automated analysis to make decisions. Analytics is getting AI Overviews on the homepage that recap what’s changed since a user last signed in, flagging things like traffic swings or seasonal sales spikes, and proposing next steps, with an option to have those recaps pushed by e-mail or mobile notification. The Ads homepage is being rebuilt around personalized insight cards, and advertisers can put questions in plain language, such as how rivals are eating into their impression share. Dashboards is coming to Ads too, turning text prompts into visual reports with AI-written summaries explaining the trends behind each chart, with Analytics support coming later.
Square opened Bill Pay to all sellers and now lets them fund vendor payments like rent, insurance, and marketing expenses with the Square Credit Card even when the vendor won’t take cards, routing the money out as an ACH deposit or a mailed check. The card pays 3% back on Bill Pay charges, which Square says always covers the Bill Pay fee without saying what that fee is, so the reward and the cost roughly cancel. Square issues the card and stands in as the merchant on those payments, so most of the interchange it pays is interchange it collects. The card carries no annual fee, no late fees, no foreign transaction fees, and no personal guarantee, and the credit line grows with a seller’s sales. I’d be careful putting some of those expenses, especially rent, on a credit card, as that can likely spiral quickly with interest accruing.
Google was given one week to remove what Judge James Donato called “anticompetitive friction” from the way rival app stores are found and installed in the Play Store. An Epic lawyer at Wednesday’s compliance hearing typed “store for apps” into Play and got Walmart and other physical retailers back rather than any app store. He then demonstrated that installing Aptoide Games, the first rival storefront distributed inside Play, requires tapping through a “Looking for Aptoide Games?” banner and a View button before an Install button appears. Donato ordered Google to return a list of stores for that search and any similar wording, to swap the View button for Install, and to drop the banner entirely, all of which Google agreed to do.
Amazon has stopped naming purchased items in its order confirmation e-mails, which now show only a category like Beauty or Wireless Accessories next to clip art, according to The Verge. The vague version started showing up around July, requiring the customer to exit their e-mail and visit Amazon’s website or app for more information about the purchase. A spokesperson said Amazon simplified the e-mails to point customers toward its own order pages, and that doing so reduces the customer information leaving its properties, but they did not address whether the format specifically shields purchase data from Google, whose Gemini mines a user’s inbox for details it can act on. Amazon pulled the same trick in 2020, stripping item details down to an order total and a date, and observers at the time read it the same way, as a move to stop third parties from scraping inboxes.
MacKenzie Scott, the billionaire ex-wife of Jeff Bezos, is publishing her novel Last Days in Two Nations (A History in Eighteen Minds) on Substack, releasing it in pieces rather than working with a publisher or a bookseller. Ironically, she’s not publishing the book at all on Amazon, despite having been made a wealthy woman from the company. So far, Scott has published the prologue and three chapters of the book, which appears to be about a fictional religious community called the Church of Long Tomorrow, whose members are “trapped by beliefs that were not their own.” Scott studied under Toni Morrison at Princeton and won an American Book Award for her first novel in 2006, so she’s the real deal, not just a hobbyist writer. Jeff Bezos seems to have a thing for authors. His new wife, Lauren Sanchez Bezos, wrote two children’s books, The Fly Who Flew to Space (2024) and The Fly Who Flew Under the Sea (2026), which are 32 and 48 pages long, respectively.
Creators earning more than $250k a year pick Instagram over TikTok for branded content by 60% to 30%, according to a CreatorIQ survey of 5,000 mostly U.S. creators. TikTok ranked first overall at 52% as the leading platform for branded content, compared to Instagram at 43%. However, once creators start earning the big bucks, they lean toward Instagram, which they judge as having better longer-term business potential. Talent firm Illuminate Social said 64% of the brand partnerships it handled this year involved Instagram against 14% for TikTok. Business Insider notes that TikTok is where smaller creators get paid because Spark Ads let brands cheaply boost micro-influencer posts and TikTok Shop opens affiliate revenue at 1,000 followers, while larger budgets follow creators to Instagram as they scale.
Sam Altman is not dead, for the record. Last week, Google showed a Knowledge Panel that claimed the OpenAI CEO had died, after posts had appeared on X and Reddit. Google responded hours later on X that the result was gone and wasn’t a manual change, and that vandalism of public information sources can affect what appears in Search. Altman’s Wikipedia page was edited 22 times Wednesday, against five edits over the rest of the month, including one claiming he’d been assassinated and later ones repeating the claim and altering his surname from “Altman” to “Gaiman.” The Wikimedia Foundation told Business Insider that volunteers reverted the edits in less than an hour and semi-protected the page, which limits editing to established accounts. Anyway, he’s still ticking.
In lawsuits this week…
- Meta went to trial in federal court against 29 state attorneys general who accuse it of designing Facebook and Instagram to addict minors and of unlawfully harvesting their data, with CEO Mark Zuckerberg and Instagram head Adam Mosseri due to testify over a seven-week period. Meta estimates the damages sought could reach $1.4T against its own $1.5T market cap, and four states want the judge to impose age restrictions, kill infinite scroll, and force deletion of algorithms trained on children’s data.
- The SEC sued Andrew Spaventa, founder of investment firm The Spaventa Group, and three entities he controls, alleging they ran a boiler room that used over 100 cold callers to sell retail investors pre-IPO stakes in SpaceX, Anduril, Anthropic, and Perplexity while promising no hidden fees. However, surprise, surprise… there were hidden fees! Companies Spaventa owned bought the shares first and flipped them to his own funds at premiums averaging 46% and running as high as 91%, with more than 800 investors putting $74M into 11 funds while paying $23M in undisclosed fees.
- Shein lost its UK copyright case against Temu, in which it alleged that Temu had copied its product images “on an industrial scale” to advertise knockoffs of its own-brand clothing. The court ruled that Temu wasn’t liable for what its third-party sellers uploaded, since its terms prohibit infringing content and it removed the listings once notified, and it ordered Shein to compensate Temu for listings pulled over images Shein didn’t actually own.
In corporate shakeups this week…
- Kroger hired Nate Faust, the co-founder of Jet.com and former Walmart U.S. executive, as its new executive VP and chief eCommerce officer.
- Target named Chandhu Nair as its first chief AI officer, hiring him out of Lowe’s, where he ran stores, data, AI, and innovation as senior VP. Target also promoted Purvi Shah to senior VP from her role running user experience design, a group which helped launch the company’s conversational shopping experience inside ChatGPT eight months ago.
- OpenAI’s ethics lead Chloe Bakalar left in July with no announcement and no replacement named, according to the Financial Times. She was reportedly the company’s only dedicated ethicist, and the position remains vacant. Brad Lightcap, an eight-year veteran who previously served as CFO and COO, announced his departure to build something new, which he hasn’t yet described. Lastly, the company is recruiting a power-trading lead to hedge the electricity and natural gas its data centers burn, a year after Meta made the same move.
- Native Commerce named Michael Erhard as its new COO, hiring him out of Amazon after ten years with the company, where he ran operations engineering across robotics, warehouse automation, and fulfillment infrastructure.
- Lululemon’s chief AI and technology officer Ranju Das and chief strategy officer Rachel Acheson have left the company, weeks before former Nike executive Heidi O’Neill takes over as CEO on September 8.
- eBay hired Andres Gonzalez as VP and GM of global shipping and authentication, bringing in a Walmart supply chain and last-mile veteran who also spent four years on parcel operations at Amazon Global Cross Border.
Flipkart started selling gourmet groceries on its quick-commerce app, Minutes, and launched a private label called Pykd to sit alongside the outside brands it sells. Available gourmet products include imported cheese, specialty coffee, ghee, wood-pressed oils, ramen (that’s gourmet?), chocolate, kombucha, and boba, while the Pykd brand is limited so far to namkeen and chips. Flipkart’s biggest rivals are making the same move into gourmet groceries, including Blinkit, Zepto, FirstClub, and Swiggy.
🏆 This week’s most ridiculous story… Amazon began building a 438,500-square-foot data center on 56 acres in Gilroy, California without ever actually going before the city council or the planning commission for permission. The company purchased the parcel in 2020 before anyone was thinking about data centers, and since the city’s industrial zoning already allows them, clearance took only a single signature from the city’s community development director in July 2023, who I imagine can now be seen around town driving a 2024 Mercedes G-Wagon. Most residents learned of the data center when construction started last month, with one resident who tried to raise drought concerns told that the comment window had closed in September 2024. Amazon said the company followed the standard process for industrially zoned land, but recognizes that it should have engaged residents sooner. The city is now drafting rules requiring commission review and neighbor notice before the next one. Wait, Gilroy California, a city with a population of 60k residents, needs more than one mega data center?
10. Seed rounds, IPOs, & acquisitions
Lovable, the AI vibe-coding platform that turns plain language prompts into web applications, raised $400M at a $13.3B valuation in a Series C round co-led by Menlo Ventures and the EQT-managed Scaleup Europe Fund. The company plans to use the funds to make its platform more capable of building and running businesses (as opposed to just software), improve its systems to detect which software creations produce successful outcomes, and expand its workforce. Lovable says that users have created more than 60M projects on its platform since launch in Nov 2024, and that Lovable-powered apps now receive more than 900M visits per month.
Anthropic is in talks to acquire Decart, an Israeli AI lab that builds chip optimization software and real-time world models, including a virtual try-on model used by eBay, for around $6B, marking a 50% premium over Decart’s valuation in its May funding round, according to Bloomberg sources. The potential deal, which would be Anthropic’s largest known acquisition, aims to help Anthropic’s existing computing infrastructure handle more tasks without having to expand its GPU footprint. Anthropic has publicly acquired only six companies, and the reported $6B purchase of Decart would be roughly 10x the disclosed value of all previous deals combined.
Thrive Holdings, a Thrive Capital spin-off that buys traditional service businesses like accounting and IT firms and rebuilds their workflows around AI, raised over $2B at a $12B valuation from investors including SoftBank, D1 Capital Partners, and Altimeter Capital. OpenAI took an ownership stake in Thrive Holdings in December 2025, and as part of that arrangement it embedded research, product, and engineering staff inside Thrive’s companies to accelerate AI adoption. Part of the funds will go toward a new platform focused on the technical and regulatory work of getting infrastructure permitted, built, certified, and kept in operation.
Yuno, a Colombia-founded payments orchestrator that connects merchants to 1,000+ payment methods through one API, raised $45M in a Series B round led by Global PayTech Ventures. The company plans to use the funds for R&D focused on expanding its AI-native payments infrastructure as it targets profitability next year. Yuno sells one-click checkout, intelligent routing, AI fraud screening, and KYC/KYB across more than 190 countries and says it has recovered over $5B in transaction volume that would otherwise have failed last year.
Yulu, a Bengaluru company that rents electric scooters to gig delivery workers on weekly subscriptions, raised $93M at a reported $170M valuation in a Series C round, split between $63M in equity led by GEF Capital Partners and $30M in debt. The company plans to use the funds to take its fleet from about 50k vehicles to 200k within two years, with a third of them being a new faster scooter built for longer e-commerce runs, bike taxis, and parcel work its slower bikes can’t handle. Yulu CEO Amit Gupta said this should be the company’s last equity raise before a public listing, with further expansion funded by debt and leases.
Anthropic is planning to go public in October at a valuation of roughly $2T, which would top SpaceX and make it the largest stock offering ever, according to the Financial Times. Investors told the FT they expect the company’s annual revenue run rate to reach $100B to $120B by the end of the year, more than 10x its level at the start of 2026 and up from $47B in May. Anthropic hasn’t set a valuation target itself yet, and the $2T figure comes from investors running their own models.
SpaceX completed its acquisition of Cursor, an AI code editor that competes with OpenAI’s Codex and Anthropic’s Claude Code, for $60B in an all-stock deal. SpaceX agreed in June to acquire the company, which operates as Anysphere, Inc., and the two had already been working together since April on a model training partnership. Cursor said in a blog post that it will get access to what it described as the world’s largest GPU fleet, enough compute to train stronger models that are cheaper to run, and that customers should see those savings, citing Grok 4.6 as an early look at what the combined operation can produce.
OpenAI completed a $7B tender offer that let employees sell shares of the company ahead of a potential IPO, according to Bloomberg sources. The company used its own cash-on-hand to buy back shares from current and former employees rather than tapping outside investors for the purchase like in the past. The sale priced the company at $852B, flat against its March round, marking the first tender offer that didn’t come with an increased valuation.
Jumia, an African online marketplace that operates its own logistics and payments network, raised $50M from the World Bank’s International Finance Corp, Axian Telecom, and other investors. The company, which launched 15 years ago to build Africa’s equivalent of Amazon, is listed on the NYSE and currently sits at a $743M valuation, with shares at $6 after peaking above $60 in early 2021. Jumia now operates in eight markets and expects to be profitable in Q4 for the first time ever.
ClearJet, a parcel network that ships e-commerce orders in unused cargo space on commercial flights, raised $25M in a Series B round led by Edison Partners, bringing its total amount raised to $40M. The company owns no planes or trucks, and instead sells retailers space on passenger flights already running between US cities, and then hands the cargo off to last-mile carriers like FedEx, USPS, DoorDash, Uber, OnTrac, and Veho. ClearJet said the model takes as much as 35% out of shipping costs and one to three days off delivery, and that the company is profitable, with revenue more than tripled from a year ago and nearing nine figures.
Graas, a retail AI platform that unifies brands’ sales, inventory, and marketing data into one system, raised $17M in a Series B round led by LemmaTree, bringing its total amount raised to roughly $66M. As part of the deal, Graas acquired Trustana, a product data platform that turns messy supplier catalogs into structured, searchable product listings, for an undisclosed amount. LemmaTree was already a Trustana investor and becomes Graas’s largest shareholder following the round.
Poshmark acquired the operational facilities of Linda’s Stuff, a luxury consignment reseller that spent 25 years on eBay and became its exclusive handbag consignment partner in 2023, for an undisclosed amount. The company’s eBay store is winding down its remaining inventory, with over 100k items still listed, and the luxury livestreams it ran on eBay Live are moving to Poshmark. eBay has deleted its US and UK consignment pages without announcing anything, which leaves the program with no partner in either market after Luxe Collective closed in Britain following a high-value theft.
Thanks for being a Shopifreak!
If you found this newsletter valuable, please leave a review on Google and share the newsletter with your friends and colleagues to help us grow.
See you next Monday,
PAUL
Paul E. Drecksler
🌐 Shopifreaks.com
🧑💼 Add me on LinkedIn
📧 [email protected]
📱 +1-828-273-3031
⭐ Leave A Review
PS: Every time I think a negative thought, I put $1 in my Pessimism Jar. It’s currently half empty.