#259 – Meta’s Government Playbook, OpenAI’s Secret Pen, & The Booming Returns Business

by Paul Drecksler | Jan 5, 2026 | Recent Newsletters

Hi Shopifreaks

Happy New Year! Thank you for continuing your readership of my newsletter into 2026.

I started Shopifreaks five years ago in January 2021 as a way to “publish my homework” that I was doing each week to keep up with the industry and best serve my clients. It started out with a few hundred readers that I invited to subscribe through my personal network and social media and has since grown to over 20,000 readers across all walks of the e-commerce industry — from merchants and agency owners to CEOs and executives of the biggest e-commerce platforms and marketplaces in the world.

I'm grateful for your readership and will do my best to continue providing you with unique coverage and perspective on the industry in the years ahead.

If you love this newsletter, please write a review on Google and share the website with your network and colleagues so that they too can benefit from my weekly content.

Wishing you a successful year ahead!

In this week's edition I cover:

Meta's playbook to deceive government regulators
OpenAI wants to sell you a pen
A Redditor blows the whistle on food delivery apps
Meta Advantage+ is creating unwanted AI ads
xAI introduces Grok Business & Enterprise plans
2026 called, and it's not hiring anyone
Saks Global is facing bankruptcy
The returns business is booming
Etsy warns sellers to lower shipping rates
TikTok Micro Dramas are growing in popularity
Reddit overtook TikTok in the UK
Free nicotine for all tech workers!

All this and more in this week's 259th Edition of Shopifreaks. Thanks for subscribing and sharing!

Stat of the Week

Annual returns in the U.S. is larger than the GDP of roughly 170 countries. National Retail Federation predicts that 17% of holiday purchases will be sent back and that total returns for the retail industry will reach $850B this year.

1. Meta created a playbook to fend off government scrutiny

Meta developed an internal “playbook” aimed at deceiving government regulators who aimed to crack down on scam advertising on Facebook and Instagram, according to a new report by Reuters. The report shows that Meta adopted tactics that made it more difficult for regulators to find ads for scams on its platforms, instead of working to reduce the actual number of scam ads, so that it could continue to bring in billions of revenue.

Here's a brief timeline of what went down:

Last year Japanese regulators were upset over the amount of ads for obvious scams that appeared on Facebook and Instagram.
Meta feared that Japan would start requiring it to verify the identify of all it advertisers, which would reduce fraud but also cost the company revenue.
Rather than make effort to reduce the number of actual scam ads, Meta instead took actions to make the problematic ads less discoverable for Japanese regulators.
For example, when Meta realized that the regulators were using its Ad Library to test its effectiveness at tackling scams, the company identified the top keywords and celebrity names they used to find the fraud ads, and deleted the number of ads they'd find.
Internal documents revealed that Meta directed staffers to make problematic content “not findable” for “regulators, investigators and journalists.”
Their tricks worked! At one point, Japanese regulators publicly praised Meta for its reduction in fraudulent ads.
Later, Japan decided not to mandate the verification and transparency rules that Meta feared, meaning Meta's bullshit prevailed.
The search-cleanup in Japan was so successful that Meta added the tactic to their “general global playbook” which it deployed against regulatory scrutiny in other markets including the U.S., Europe, India, Australia, Brazil, and Thailand.

All I can really say is — wow! Just when you thought Meta couldn't get any worse… surprise! Well, it's not actually a surprise at all, especially after reading Careless People last year and learning about the extreme lengths the company took to bring Facebook to China, as well as its general disregard towards adhering to government rules.

Meta spokesperson Andy Stone told Reuters that there is nothing misleading about removing scam ads from the library because that means that Meta also removed the ads from its systems overall. He also noted that the company has seen a 50% decline in user reports of scams — which immediately makes me question whether Meta simultaneously took effort to make reporting scams more difficult and/or began classifying user reports differently.

The Reuters report goes on to describe Meta's aversion to advertiser verification and the efforts that the company took to stall and dismiss regulation for it in various jurisdictions, purely for financial reasons. Meta staffers estimated that the cost of implementing a verification system for all advertisers globally would cost the company around $2B (or roughly 1.2% of its $164.5B revenue last year), and that blocking unverified advertisers could cause the company to lose up to 4.8% of its total revenue.

At the same time, Meta has internally cited estimates that scam ads are responsible for $63B in annual damage to consumers worldwide.

So let's do the math…

If Meta spent $2B in implementation one time, and potentially lost $7.9B in revenue each year by blocking scam ads from its platforms, it could save consumers $63B in annual damages. That's an 800% return on ‘no scammer' ad spend!

At this point it's proven without a doubt that Meta doesn't give a fuck about its users or government regulation. Drastic measures — such as blocking Meta's services in countries or suing / fining Meta to the point that it becomes too expensive for them to continue serving scam ads — is the only thing that's going to make a dent in making its platforms less damaging to the world.

2. OpenAI's secret project with Jony Ive could be an AI-powered pen

“Sell me this pen.”

Sam Altman apparently took those words to heart when watching Wolf of Wall Street back in the day because now he wants to sell you his pen. (No, not his pen fifteen.)

A post on X from industry tipster Smart Pikachu revealed that OpenAI's hardware device that's being created in collaboration with former Apple chief designer Jony Ive could take the shape of a pen.

Mashable wrote:

“The form factor hints at a focus on handwriting and creativity, potentially blending digital note-taking with generative AI. The pen could allow handwritten notes to be instantly transcribed, sent to ChatGPT, or even function as a stylus with direct AI integration on tablets and other devices.”

The leak also suggests that OpenAI could be working on a standalone audio-based AI assistant designed to reduce reliance on smartphones and laptops to access artificial intelligence, offering hands-free access to ChatGPT in an always-available format. Like a portable Echo device?

That speculation aligns with an article from The Information, which reports that OpenAI is working to upgrade its audio AI models, aiming to develop an architecture that handles interruptions and produces more natural, emotive responses.

Smart Pikachu also said that OpenAI does not plan to produce its hardware in China, and instead is partnering with Foxconn, which has manufacturing facilities in Vietnam and the U.S.

Personally I'm ready for an audio-only “third device” to complement my life and workflows. Humane was ahead of its time and lacked the backing of a powerful enough audio-focused LLM. The next generation of devices could have more success. You'll know we're there when Apple makes a version of it.

🔥 Partner News

Napster launched Napster Station, the first AI concierge kiosk designed to perform in noisy, high-traffic environments like airports, hotels, hospitals, and retail spaces where traditional voice assistants fail. The system combines purpose-built hardware with Microsoft Azure OpenAI to deliver conversational video AI that can isolate a single user’s voice, understand context, and provide multilingual assistance in real time. Napster says Station costs about $1 per hour to operate and will be available for enterprise deployments starting in Q1 2026, with live demos running throughout CES.

3. Developer for a major food app whistle blows on Reddit

An anonymous Reddit post claiming to be written by a backend engineer at a major food delivery platform blew the whistle on alleged internal practices at the company tied to pricing, dispatch, and driver pay. The post describes how certain fees, upsells, and algorithms are supposedly designed to exploit drivers and scam customers with psychological value adds.

Here are highlights from the post:

Managers internally refer to drivers as “human assets” and talk about them as if “they are resource nodes in a video game, not fathers and mothers trying to pay rent.”
The “Priority Delivery” fee, which costs customers $2.99, is just a “psychological value add” that does nothing to actually speed up your delivery.
At one point last year, the company tested purposefully delaying non-priority orders by 5-10 minutes to make the priority ones feel faster by comparison.
The company internally has a “Desperation Score,” which is a hidden metric for drivers that tracks how much they need the cash based on their acceptance behavior. If a driver logs on late and accepts every low paying order without hesitation, the system tags them as “High Desperation” and deliberately stops showing them high-paying orders because they know they will take the low paying ones.
Conversely, the app saved the good tips for the “casual” drivers to initially hook them in and gamify their experience.
The “Benefit Fee” also known as a “Regulatory Response Fee” or “Drive Benefits Fee” is worded to make you think you're helping the worker, but in reality, the money goes “straight to a corporate slush fund used to lobby against driver unions.”
The app uses predictive modeling to dynamically lower driver base pay if the algorithm predicts that the customer is a high tipper. The whistle blower wrote, “The result is that your generosity isn't rewarding the driver; it’s subsidizing us. You’re paying their wage so we don't have to.”

Is any of it true? Does the post author even work for a major delivery app or is this just a Reddit role-playing fantasy post?

Several comments pointed to the post being AI generated and poked holes in the plot, while others substantiated OP's claims.

One commentor claimed to actually work for DoorDash and verified several of the claims, but “corrected” others. For example, regarding lowering base pay for higher tippers, he wrote:

“The part that's misleading and/or incorrect: – while it's true that we modify the base pay, typically it's to increase the base pay when the customer doesn't tip in order to guarantee a minimum amount of pay to the drivers.”

However as another commentor replied:

“This is the same as lowering the base pay when the tip is large, just with positive wording that makes it sound nicer.”

What do you think? Is the post fact or fiction? Hit reply and share your thoughts or join the conversation on LinkedIn.

4. Meta Advantage+ is forcing unwanted AI creative on brands

Meta advertisers are reporting that the platform's “Advantage+” and “automatic adjustments” settings are auto-enabling to generate bizarre, unauthorized creatives in their campaigns, despite the brands explicitly opting-out from using the tools.

Business Insider offered a few examples:

Bryan Cano, head of marketing at the clothing brand True Classic, said that Meta switched out his top-performing ad featuring an attractive millennial man in a matching fleece set with an AI-generated photo of a cheerful grandma sitting in an armchair smiling. The company typically targets its ads to men ages 30 to 45. The ad ran for several days before customers alerted the brand to what they were seeing.
European footwear brand Kirruna said that Meta created an ad featuring a model whose leg had twisted around completely the wrong way.
E-bike company Lectric shared a story of Meta producing a creative for the company featuring a car flying through clouds, but they were able to catch the ad before it ran.

A Meta spokesperson gave a statement:

“Advertisers who use our full image generation feature have the opportunity to review the generated images before running their ad. We are continuously improving our products and features based on advertiser feedback.”

Maybe. But Cano at True Classic denies being able to preview the grandma ad before it went live. Other advertisers told Business Insider that they'd encountered issues where Meta would automatically switch on its “automatic adjustments” settings even after they'd explicitly turned it off.

Rok Hladnik, CEO of the marketing agency Flat Circle, which manages around $100M in annual Meta ad spending for D2C brands, confirmed that the setting will randomly turn on, and said that his company is now setting aside several mornings a week to manually check that AI enhancements are still switched off.

Meta's not concerned with the issues or the backlash, as they've likely got a playbook to deal with it. LOL.

5. xAI introduces Grok Business & Enterprise plans

xAI launched Grok Business and Grok Enterprise to provide organizations with secure, private access to its AI models without using their data for training. The new tiers feature high rate limits, Google Drive integration, and administrative tools like single sign-on and directory sync for larger teams.

Here's a breakdown of what the Business & Enterprise plans offer:

Grok promises to never train its models on your data. “Your data stays yours: no training on it, ever.”
Access to xAI’s strongest Grok models — Grok 3, Grok 4, and Grok 4 Heavy — with high rate limits, as opposed to its stupid models that it makes public.
Shared team workspace with conversations and project collaboration.
Grok respects your existing Google Drive permissions. If a file isn't shared with you in Drive, you won't see it in Grok.
Every answer includes citations linking directly to source documents, with quote previews and highlighted relevant sections.
Agentic search across large internal document collections, such as a data room for analyzing legal documentation or building financial models.
Centralized admin console with user management, usage analytics, and unified billing.
Self-serve onboarding for small and mid-sized teams.
The Enterprise edition adds custom SSO, directory sync via SCIM, advanced security and audit controls, and Enterprise Vault, which is a new offering that provides an isolated data plane with customer managed encryption keys, keeping enterprise data fully segregated from xAI’s shared consumer infrastructure. This has become a standard offering among Enterprise-grade AI offerings.

This isn't xAI's first premium AI subscription offering — it previously monetized Grok through consumer subscriptions, initially bundled with X Premium Plus, and later offered as a standalone subscription outside of the X social platform. However the launch of Grok Business & Enterprise is the company's first time entering the competitive enterprise space currently led by OpenAI's ChatGPT Team, Anthropic's Claude Team, and Google Gemini.

The release of Grok Business & Enterprise comes the same week that xAI took heat for Grok removing clothing from pictures of people without their consent (including minors). The company apologized for the issue and called it a “failure in safeguards.”

6. Larger employers don't plan on hiring in 2026

Companies are looking to stay lean in 2026 and rely on technology to take on more tasks, according to a Wall Street Journal report, which said the corporate playbook for this year is: “Don't hire.”

The Journal reported:

66% of CEOs surveyed last month by the Yale School of Management said they planned to either fire workers or maintain the size of their existing teams next year. Only one-third planned to hire.
The unemployment rate in the U.S. rose to 4.6% in November, its highest in four years.
Economists at Indeed predict the unemployment rate to hover around that 4.6% throughout 2026.
Prominent employers such as Amazon, Verizon, Target and UPS have cut white-collar roles in recent months, which has added to the unease of the companies' remaining staff.
Federal Reserve governor Christopher Waller said, “We’re close to zero job growth. That’s not a healthy labor market. When I go around and talk to CEOs around the country, everybody's telling me, ‘Look, we're not hiring because we're waiting to try to figure out what happens with AI. What jobs can we replace? What jobs do we don't?'”
He later said, “Everybody’s afraid for their jobs. I’m dead serious.”
The hiring environment is causing many employees to cling to their jobs. For example, the voluntary attrition at IBM is now under 2%, a decline from the typical 7%.
Shopify CFO Jeff Hoffmeister recently said at a conference, “I don't see us next year needing to increase head count in any way. It has been over two years we've been at this head count. As I look to next year, I think we can continue to be disciplined on head count.”

How confident do you feel at your current employment? Is it as shaky ground in corporate America as WSJ describes? Hit reply and let me know. As always, your replies remain 100% confidential.

7. Saks Global is facing bankruptcy

Saks Global, the parent company of Saks Fifth Avenue, Neiman Marcus, and Bergdorf Goodman, is racing to land more than $1B in rescue financing from new and existing investors to pay off debts, or the company may face bankruptcy.

Here's what we know about the developing situation:

The company's CEO Marc Metrick stepped down and has been replaced by Richard Baker, the execute chairman of Saks Global.
Saks missed an interest payment on its debt last week, of which it has a lot.
The company accumulated a ton of debt following its $2.7B acquisition of Neiman Marcus Group in 2024 and has since struggled to revive sales since the deal closed. The deal was financed by $2.2B worth of junk bonds, which S&P Global called “unsustainable.”
Debts include a $100M interest payment to bondholders that came due last week. Plus millions to vendors, many of whom have not been paid in full for more than a year.
In August, Saks completed a debt structuring to take some pressure off its finances and strengthen its balance sheet, but that might not prove to be enough.
Last month, Saks sold its Neiman Marcus property in Beverly Hills to an investment firm for an undisclosed sum to free up more cash. Moves like this are eventually what led to Red Lobster's demise.
In early December, Hilldun Corp., which acts as a guarantor or insurer for brands and represents about 140 brands that sell to Saks, said it would no longer back shipments to retailer.

All this happening while consumers simultaneously decreased their spending on luxury goods over economic uncertainty led to the perfect storm for Saks Global.

If talks with investors fail, and Saks is unable to raise the funds to cover its debts, the next logical steps for the company would be Chapter 11 Bankruptcy to restructure. A source told The Post that discussions with investors will likely wrap up within a couple of weeks.

As you might recall, Amazon and Salesforce are investors in Saks Global and helped fund the Neiman Marcus deal, which in turn created the Saks Global holdings company. I'm curious to see if / how they play a role in the potential rescue financing and future of the company.

8. Business is booming for gig worker return services

The Wall Street Journal reports that Taskrabbit has seen a 62% YoY increase in return-related bookings during November and December (and that's likely about to blow up even more in January), with more people hiring its taskers to do the return leg work.

The article also features a six-year-old startup called ReturnQueen, which says it expects up to a 20% increase in January and February, its biggest months for returns each year, as well as College Hunks Hauling Junk & Moving, which told WSJ that it has been getting an increasing number of inquiries from customers during the past two years for returning heavy purchases like treadmills and other fitness equipment.

The premise of a gig worker return service is simple: A driver appears at your door and collects your returns. Then they either take them to local retail locations and drop-off stations, wait in line for you, and process the return on your behalf, or they take the items back to their sorting facility to ship them back. It's a pretty brilliant idea for folks who simply don't want to deal with returns.

Returns are getting out of hand in the U.S.

The National Retail Federation predicts that 17% of holiday purchases will be sent back and that total returns for the retail industry will reach $850B this year. In other words, our annual returns in the U.S. is larger than the GDP of roughly 170 countries.

Holy shit we are wasteful! The WSJ article mentions an online shopping addict named Gila Nussen who intentionally over-orders during the holiday season, particularly clothing, and then makes “keep” and “return” piles with her daughter after opening gifts. It's simply part of their holiday traditions now.

I recently saw a TikTok video of a line of people returning their used Christmas trees to Costco. I also regularly read comments on videos and posts online from people saying things like, “Just buy a new one and return that one in the box” about items they've broken. Do folks have no shame anymore?

Retailers are noticing and are tightening up their return policies, particularly around the holiday season. For example, Best Buy, Macy's, Kohl's, and other retailers introduced return fees and stricter documentation rules this year, while other major retailers shortened their return windows.

Honestly the American consumer's relationship with returns has gotten out of hand, and I support these retailers tightening up their policies. It's happening all over the world too, not just in the U.S.

It's unfortunate that ~~a few people~~ tens of millions of entitled American consumers can ruin good return policies for the rest of us.

9. Other e-commerce news of interest

The U.S. government issued visa bans for five European Union citizens, including Digital Services Act architect Thierry Breton, to challenge Europe's regulatory stance on American tech companies. Nothing says cooperation like an outright ban! Quite the olive branch. Bloomberg Economics reported that “the escalation could have wide-ranging affects on trade negotiations and more than $400 billion in transatlantic trade in digital services.” The Trump administration characterized the EU's digital sovereignty policies as discriminatory against U.S. firms while European officials defended their right to enforce consumer protection laws.

Etsy is advising sellers to “optimize” their shipping rates, which many are reading as instructions to “lower their rates or else.” An e-mail sent Jan 1st to sellers revealed that the marketplace is now prioritizing listings with shipping prices lower than $6, despite rising shipping costs in the U.S. The letter explained, “Why under $6? Our data tells us that once shipping prices are $6 or above, buyers are less likely to make a purchase. Lowering your shipping prices could help improve the visibility of your items in search, which could lead to more sales for your shop!” Unfortunate news, especially for anyone selling bulky or oversized items.

TikTok hosted over 6,000 AI-generated deepfake videos of UK Prime Minister Keir Starmer between May and December last year, many of which had him making announcements about fake curfews and other misinformation, according to a study by NewsGuard. The incentive for producing these videos? Money! NewsGuard says the video do not appear to be created by foreign adversaries like Russia or the prime minister's domestic political opponents, but rather, by content creators whose primary goal is to make money from TikTok's Creator Fund by making the videos go viral. It's estimated that the curfew video alone generated as much as $3,500 for the creator. TikTok has since removed the videos and reinforced that the platform prohibits “fake authoritative sources or crisis events, or falsely shows public figures in certain contexts.”

Instagram CEO Adam Mosseri said in a recent Threads post that social media platforms like Instagram will be under mounting pressure to help users tell the difference between AI and real content, and that the job will become more difficult the better AI gets. He wrote, “There is already a growing number of people who believe, as I do, that it will be more practical to fingerprint real media than fake media,” noting that the fingerprint could be created within cameras themselves if their manufacturers “cryptographically sign images at capture, creating a chain of custody.” He added, “We need to label AI-generated content clearly, and work with manufacturers to verify authenticity at capture — fingerprinting real media, not just chasing fake.” Meanwhile his own parent company is one of the major players leading the AI slopification of social media platforms globally.

TikTok introduced a “TikTok Minis” section dedicated to “micro dramas,” which are low-budget, vertical soap operas chopped into short episodes that feature fast-paced plots and terrible acting. The micro dramas, which were popularized in China and are now making their way to the U.S., mostly operate on a freemium model, with users exposed to about eight or ten episodes (which is like 4 minutes of content, LOL) before being asked to pay around $10 per movie or purchase a subscription for $40 to $80 per month for unlimited viewing. TikTok also offers a 10% discount to incentivize viewers to stay within the TikTok app rather than click away to the short drama app itself. Two micro drama producers told Business Insider that TikTok pitched them on sharing full, original episodes on the platform for free (ie: no paywall), in exchange for a share of ad revenue and a licensing fee of up to $10k for an original series. Who wants to start creating a micro drama with me?!

The United States Postal Service is collecting data so it can perform its annual peak-season performance review. Last year USPS determined that it didn't meet five of its six service targets during peak season, according to the report published in July. The review will assess on-time delivery across major mail and package categories, the effectiveness of peak-season preparedness plans, and how weather and staffing levels affected performance. The Postal Service is also examining post-peak volumes, including returns, which have historically added strain to the network after Christmas.

OpenAI is facing a class action lawsuit filed by a Virginia consumer alleging the company bears liability for illegal robocalls/texts generated through its platform. The complaint argues that OpenAI and Twilio, a communications API provider, facilitated violations of the Telephone Consumer Protection Act by enabling third-party Fresh Start Group to send non-consensual marketing messages, and seeks to represent all U.S. recipients of unauthorized AI-generated outreach. The plaintiffs claim the companies knew or should have known their tools were being used for harassment. My guess is that if the outcome is anything like recent copyright infringement cases, it'll merely be a slap on the wrist.

Triller, a Los Angeles-based creator platform and TikTok competitor, was delisted from the Nasdaq exchange after failing to file its 2024 annual report and subsequent quarterly statements on time. The suspension removed the “ILLR” ticker from public trading following multiple warnings regarding the missed deadlines. Management attributed the accounting delays to a technical consolidation issue and pledged to regain compliance within weeks despite reporting a 57% revenue decline earlier in the year. Yikes, that company sounds like it's in trouble, and the U.S. TikTok deal going through may have been the nail in the coffin after a difficult few years.

OpenAI filed a request to dismiss a trade secret lawsuit from xAI for the second time after characterizing the amended complaint as fundamentally flawed. In a lawsuit first filed in September, xAI accused the company of hiring eight of its employees, including engineers and a senior finance executive, as part of a “deliberate scheme” to steal confidential information. From there, OpenAI filed to dismiss the lawsuit in October, before xAI filed an amended complaint a few weeks later, which OpenAI says is “even more flawed” and fails to provide evidence that the eight poached employees actually disclosed confidential information to their new employer. Attorneys for OpenAI described the litigation as another attempt by Musk to harass a competitor rather than a legitimate claim.

In Meta lawsuits this week (always something)… The U.S. Virgin Islands Attorney General is suing Meta over allegedly generating revenue from fraudulent advertisements to boost engagement. The lawsuit cites an article by Reuters last month that revealed how Meta internally projected that 10% of its 2024 revenue ($16B) would come from ads for scams, illegal gambling, and banned products. Meta also received a formal request from 29 state attorneys general last week to consolidate their social media addiction lawsuits into a single joint trial to avoid years of repetitive litigation, arguing that separating the cases would allow the company to weaponize procedural delays against claims seeking injunctive relief and profit disgorgement, which is when a company is legally required to give up profits it earned through unlawful or improper conduct. Lastly, Meta is urging a federal judge to authorize an immediate appeal of a ruling that allows users to sue the platform for financial losses caused by third-party scam ads. The dispute hinges on whether the company's Terms of Service and Community Standards, specifically promises to “take appropriate action” against harmful content, constitute an enforceable contract or merely “aspirational” goals.

Last but not least in lawsuit news… Amazon cannot pursue its constitutional challenge to the National Labor Relations Board in federal court, after the 9th Circuit ruled that labor law bars courts from intervening in cases tied to labor disputes. Circuit Judge Danielle Forrest cited the Norris-LaGuardia Act, which strips federal courts of power to issue any injunction in a case involving a labor dispute, saying, “Amazon’s requested injunction would impede union activities, the very outcome the Act was enacted to prevent.” The lawsuit stemmed from an NLRB case alleging Amazon was a joint employer of delivery drivers working for a contractor and should have bargained with their union, which Amazon denies is true. The decision conflicts with a recent 5th Circuit ruling involving a case with SpaceX that allowed similar challenges to proceed, increasing the odds the Supreme Court will eventually step in.

Walmart experienced a service disruption on Tuesday morning that prevented users from accessing its mobile app and website for approximately one hour. Downdetector logged over 6,000 complaints during the peak, with 70% of issues stemming from the app. A Shopify spokesperson said, “One hour on a Tuesday? That's rookie numbers! Try a system-wide admin outage on Cyber Monday!”

The Save Mart Companies has expanded Amazon return kiosks to 140 Save Mart, Lucky, and FoodMaxx stores across California and Western Nevada after a pilot at 15 locations. The kiosks allow customers to return eligible Amazon items without boxes, labels, or tape during grocery trips, with full rollout expected by February 2026. To drive usage, Save Mart will offer a $5 discount on same-day purchases of $25 or more for customers who complete an Amazon return in store. Uh oh, are they sure they want to do that? Did Save Mart see how that ended for UPS Stores — which last year said that Amazon returns only accounted for 10% of their revenue, but took up 90% of their working day?

eBay is losing Dan Leiva, VP of Customer Service and Marketing Technology, as he moves into semi-retirement and launches a private consulting business after nearly nine years at the company. Leiva oversaw large-scale AI-driven customer support initiatives across buyer and seller services, payments, CRM, and marketing operations, leading teams of more than 900 employees. His departure comes as eBay continues to push deeper into AI-based support tools despite ongoing seller criticism that automation has replaced human assistance rather than improved service quality.

Yann LeCun, a renowned computer scientist who recently served as Chief AI Scientist at Meta before leaving in November to form his own startup, said that Alexandr Wang, the 28-year-old Scale AI cofounder recruited to lead Meta's Superintelligence Labs via a $14B acqui-hire, was “inexperienced” and didn't fully understand AI researchers. LeCun said, “He learns fast, he knows what he doesn't know… There's no experience with research or how you practice research, how you do it. Or what would be attractive or repulsive to a researcher.” LeCun predicts that more AI researchers will soon leave Meta as a result.

eBay UK temporarily removed DHL from its Simple Delivery, leaving some private sellers no options to ship large or bulky items that are mandated to use the shipping program. Sellers have reported significant issues with DHL service for heavy items in Simple Delivery since it was introduced in August, such as the courier failing to collect items. The move has left sellers unable to ship items over 61cm through Simple Delivery and no option to use custom postage instead.

Amazon cancelled its drone delivery plans in Italy, saying that it had made progress with aerospace regulators, but that broader business regulatory framework in the country did not support the project. The Italian civil aviation authority ENAC said the withdrawal was “unexpected,” and claimed the decision was linked to “recent financial events” rather than operational hurdles. The move follows Amazon's successful completion of initial tests of delivery drones in San Salvo in December 2024.

Reddit has overtaken TikTok as Britain's fourth most visited social media service behind Facebook, YouTube, and Instagram. The platform has undergone huge growth over the last two years within the country, with three in five Brits now using the website or app, up from a third in 2023. It is now the sixth most visited platform of any kind by UK users aged 18 to 24, up from tenth a year earlier. Ah, that explains why so many Redditors are saying “lift” instead of “elevator” and “flat” instead of “apartment.” It all makes sense now.

Coupang, the South Korean e-commerce giant that recently suffered a data breach that exposed personal details of 34M users, announced a $1.17B compensation package for users affected by the breach, marking the largest payout for a single security incident in the country's history. The company will distribute roughly $35 per person affected in the form of non-cash vouchers including $3.50 for general e-commerce, $3.50 for Coupang Eats, and $28 for its travel and luxury platforms. Wow thanks for the gift cards to your insecure platform! In my humble opinion, impacted customers should have the option to receive their measly $35 compensation as cash to use anywhere they please — just like their data is currently being used.

🏆 This week's most ridiculous story… Palantir, Hello Patient, and other tech companies have begun stocking offices with free nicotine pouches to stimulate employee focus and productivity. Sesh, an Austin-based nicotine startup backed by 8VC, and Lucy, a Y Combinator-backed brand, installed branded vending machines and fridges at various headquarters to distribute the flavored stimulants for free to employees and guests over the age of 21. The companies are claiming that the products provide a focus boost similar to caffeine. So what's next? Adderall dispensers? Basically companies are admitting — one nicotine vending machine at a time — that they don't care the slightest about your health. Just suck on your pouch and do your work while we figure out how AI can replace you! Just curious, but do those same offices provide free menstrual products in their women's restrooms? Sorry ladies, you're on your own… but enjoy a free nicotine pouch on your way to the convenience store.

10. Seed rounds, IPOs, & acquisitions

Meta is acquiring Manus, a Singapore-based AI platform that can conduct deep research and execute multi step tasks across tools, files and web workflows on a user's behalf, for $2B. Manus, which debuted in mid-2024 with a video demo that showed an AI agent doing things like screening job candidates, planning vacations, and analyzing stock portfolios, has since raised over $85M, signed up millions of users, and reached annual recurring revenue of more than $100M. Its parent company, Butterfly Effect, was founded in China in 2022, but later moved to Singapore that same year, and Meta says the company won't have any ties to Chinese investors or operate in China anymore after the acquisition. Meta plans to keep Manus running independently and weave its agents into its apps and chatbot.

SoftBank is acquiring DigitalBridge, a global investment firm focused on digital infrastructure like data centers, cell towers, fiber networks, and small cells, for $4B, or $16 per share, marking a 15% premium to last Friday's closing price. The deal secures SoftBank's control over several data center subsidiaries, including developers involved in the Stargate project for OpenAI and Oracle. DigitalBridge will continue to operate as a separate unit under CEO Marc Ganzi once the deal closes in the second half of 2026. SoftBank CEO Masayoshi Son said the acquisition “will strengthen the foundation for next-generation AI data centers” and advance the company's vision to become a leading “Artificial Super Intelligence” platform provider.

In other SoftBank news… The firm finalized its $41B investment in OpenAI to secure an approximate 11% equity stake in the company at a $300B valuation. The total funding package combines a fresh $22.5B capital injection with $11B in syndicated co-investments and a prior $7.5B tranche.

Nvidia finalized its purchase of $5B worth of Intel shares, carrying out the transaction that was announced in September. Nvidia paid $23.28 per share for the stock, which certainly feels like a deal now that Intel is trading around $40 per share, but then again, the news of Nvidia's investment was one of several catalysts that bumped the stock price up. The transaction was cleared by the FTC on December 18th and closed on the 26th. Under the terms of the deal, Nvidia and Intel will jointly develop “multiple generations” of chips for datacenter and PC.

Zhipu AI, a Beijing-based startup that develops large language models for consumer and enterprise applications in China, kicked off a share sale last Tuesday to raise HK$4.35B (US$560M), aiming to become the first LLM developer to list in Hong Kong. The offering prices 37M shares at HK$116.20 each, hoping to secure a post-listing valuation of HK$51.16B ($6.6B), with 10% allocated to retail investors.

In other Chinese LLM news… MiniMax, a Shanghai-based generative AI developer backed by Alibaba and Tencent, initiated its IPO in Hong Kong, seeking to raise up to HK$4.19B (US$538M) just one day after Zhipu AI's announcement. The company plans to offer 25.39M shares globally, priced at HK$165 each, with about 5% allocated to Hong Kong retail investors and the rest to international investors. Trading is expected to begin on January 9 under stock code 0100.

Warner Bros Discovery will likely reject Paramount Skydance's $108.4B hostile bid, despite a personal guarantee from Larry Ellison backing the offer, according to Reuter's sources. The company did not increase its $30 per share all-cash offer, but instead raised its regulatory reverse termination fee to match Netflix and extended its tender offer deadline, however, analysts say that Netflix's $82.7B bid offers a clearer financing structure and fewer execution risks. Major shareholder Harris Oakmark criticized the revised offer as “insufficient” because it failed to cover the $2.8 billion breakup fee required to terminate the Netflix pact. The board has yet to make a decision, but is expected to meet this week.

Octopus Energy, a UK-based energy company that supplies electricity and gas while also licensing its Kraken software platform to utilities for billing, customer management, and grid optimization, raised $1B in its first standalone funding round for Kraken, valuing the division at $8.65B. The round paves the way for a spinoff of the SaaS unit, with Octopus targeting a separation by mid-2026, and sets the stage for an IPO. Octopus will retain a 13.7% stake and Origin Energy, which holds a major stake in Octopus, will hold 22.7%.

Elliott Investment Management, an investment firm known for its activist strategies across public equities, debt, and distressed assets, is consulting with banks regarding a potential dual listing for the American book retail chain Barnes & Noble and the U.K. book retailer Waterstones as early as the second half of 2026. The company, which acquired Waterstones in 2018 and Barnes & Noble in 2019 for a combined $962M, has overseen a significant turnaround led by CEO James Daunt by decentralizing inventory decisions and empowering local store managers to choose titles, as well as expanding Barnes & Noble’s footprint from 627 to 702 locations. The proposed offering follows a record 2025 for both of the retailers, who now share back-end infrastructure, customer service software, and staff training protocols.

Zepto, an India-based quick commerce grocery delivery service that reached unicorn status in August 2023, confidentially filed for a ₹110B ($1.22B) IPO, targeting a 2026 listing, following a recent funding round in October that valued the company at over $7B. Despite reporting a 129% revenue surge in FY25, the startup's net losses widened by 177% to ₹33.67B, fueling industry warnings, including from rival Blinkit's CEO, that “relentless fundraising” to subsidize burn rates is creating a sector-wide bubble. The filing comes as India’s quick commerce sector sees rising competition from players including Amazon, Flipkart, and Swiggy.

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: What do you call a witch on the beach? A sand witch.