Walmart Turns ChatGPT Into a Shopping Mall
4IR - Daily AI News
Welcome back to 4IR. Here’s today’s lineup:
Walmart turns ChatGPT into a shopping mall - Nation’s largest retailer bets on “agentic commerce” as 270 million weekly shoppers get AI-powered checkout, Sam Altman calls it making “everyday purchases a little simpler”
AMD lands 50,000-chip Oracle deal in Nvidia’s backyard - MI450 processors hitting Oracle Cloud by Q3 2026 as chipmaker’s stock surges 3%, marking second massive AI deal in eight days following OpenAI partnership
Marvel and DC declare war on AI art at Comic Con - Editor-in-Chief CB Cebulski says “we’ve never used AI” as Big Two publishers draw hard line on ownership, scan submissions for telltale “six fingers”
Walmart turns ChatGPT into a shopping mall
The story: Walmart dropped a bombshell partnership with OpenAI today that transforms ChatGPT into a full shopping platform. Using OpenAI’s new “Instant Checkout” feature, Walmart and Sam’s Club customers can now describe what they need conversationally—”plan a weeknight taco dinner” or “restock my staples”—and complete purchases without leaving the chat. Doug McMillon says the “search bar and long list of item responses” era is over, replaced by what he calls “multi-media, personalized and contextual” AI shopping. For OpenAI, this is their biggest retail partner yet and validates CEO Sam Altman’s push into e-commerce.
What we know:
Customers link Walmart accounts to ChatGPT, then buy via “buy” button in conversational interface
Catalog includes apparel, entertainment, packaged food, household essentials from Walmart and Sam’s Club
Third-party sellers supported when feature rolls out “later this fall”
OpenAI charges transaction fees, creating new revenue stream beyond subscriptions
Walmart already using ChatGPT Enterprise internally, among first to adopt OpenAI Certifications
Partnership follows OpenAI’s September announcement of Instant Checkout with Etsy and Shopify merchants
Why it matters: This is the moment AI shopping graduates from novelty to serious distribution channel. Walmart processes 270 million customer visits weekly—if even 1% try ChatGPT checkout, that’s 2.7 million people shopping by conversation. The “agentic commerce” framing is key: this isn’t search, it’s AI anticipating needs and closing transactions without forcing customers to click through product pages. For retailers watching Amazon’s Rufus assistant gain traction, this partnership proves conversational commerce isn’t speculative anymore. And for OpenAI, transaction fees from Walmart volume could become material revenue while they figure out the whole “profitability” thing.
Walmart launching this during the holiday shopping ramp-up is strategic timing. If ChatGPT checkout converts well on grocery reorders and gift shopping, they’ve just created a moat against Amazon’s AI assistant in the one category where Walmart competes effectively—everyday low prices on staples. The real test is whether people want to shop by conversation or if this becomes another voice commerce gimmick like Alexa shopping. The fact that OpenAI is charging per-transaction fees instead of flat licensing tells you they’re confident in volume. Watch for Amazon to announce something similar within weeks—they can’t let OpenAI own this interface.
AMD lands 50,000-chip Oracle deal in Nvidia’s backyard
The story: AMD scored another massive AI chip win today, announcing Oracle will deploy 50,000 of its upcoming MI450 processors into Oracle’s cloud computing infrastructure starting in late 2026. This is AMD’s second huge AI deal in just eight days—they announced a multibillion-dollar OpenAI partnership last week. Oracle will use AMD’s new “Helios” system, which packages the AI chips together with regular processors and networking gear into ready-to-go racks that cloud customers can immediately use for training AI models. AMD’s stock jumped 3% on the news (while the broader market was down), and the company is now up 79% for the year as Wall Street bets that Nvidia’s near-monopoly on AI chips is finally breaking.
What we know:
50,000 MI450 chips deploying in Q3 2026, with expansion planned through 2027 and beyond
MI450 chips have massive memory capacity (432GB per chip) designed to handle much bigger AI models
Oracle using AMD’s “Helios” pre-built rack systems—plug-and-play AI infrastructure
Oracle already runs AMD’s current-generation MI300X chips, so this expands existing relationship
Deal follows AMD’s October 6th announcement of 6-gigawatt partnership with OpenAI
AMD stock up 79% year-to-date as company becomes credible Nvidia alternative
Why it matters: For years, if you wanted to build serious AI, you bought Nvidia chips or you bought nothing. AMD was the distant second place that nobody took seriously for cutting-edge work. That just changed. Oracle committing to 50,000 chips isn’t a test—it’s betting real money that AMD can deliver what enterprises need. The timing matters too: this comes right after AMD announced the OpenAI deal, which means AMD finally has the one-two punch of credibility (OpenAI uses us) and scale (Oracle deploys us). For cloud companies and AI startups tired of Nvidia’s pricing and chip shortages, AMD suddenly looks like a viable escape route. And for Nvidia shareholders, this is the first real competitive threat in years.
AMD has been trying to break into AI chips forever and kept getting shut out because Nvidia’s software ecosystem (CUDA) made switching prohibitively expensive. Something clearly changed. The fact that both OpenAI and Oracle committed within a week suggests AMD cracked the software problem—you can’t just have better hardware, the code needs to actually run. Oracle’s role is clever: they’re the middleman making both deals less risky. OpenAI gets AMD chips through Oracle’s cloud, Oracle gets to offer customers choice beyond Nvidia, AMD gets massive orders. The 2026 timeline means manufacturing is probably already happening. If AMD actually ships 50,000+ chips on schedule and they work as advertised, Nvidia’s 90%+ market share becomes very vulnerable very quickly.
Marvel and DC declare war on AI art at Comic Con
The story: The Big Two comic publishers drew a hard line against AI-generated artwork at New York Comic Con today, with Marvel Editor-in-Chief CB Cebulski declaring “we’ve never used AI” and “we don’t condone it in the Marvel Comics division” during a Spider-Man panel. The statement came one day after DC Publisher Jim Lee told retailers “DC Comics will not support AI-generated storytelling or artwork. Not now. Not ever—as long as Anne DePies and I are in charge.” Cebulski revealed Marvel employs staff who scan submissions for AI tells like “six fingers” and the company severs ties with artists caught using the technology. The announcements drew applause from creators worried about their livelihoods as generative AI floods the market with cheap, algorithm-generated art.
What we know:
Marvel “never used” AI and “will not be using it” per Cebulski, specific to Marvel Comics division
Marvel staff actively scan art for AI discrepancies, terminate relationships with artists using AI
DC’s Jim Lee: AI-generated work rejected because “what we do is rooted in our humanity”
Policy driven by ownership concerns—AI-generated work may not be copyrightable, becomes public domain
Statements clarify Comics division only; Marvel Studios’ Wonder Man show reportedly used AI for props
Both publishers facing creator pressure as AI art tools proliferate
Why it matters: This isn’t a moral stance—it’s a legal one. If Marvel publishes AI-generated Spider-Man art, they might not own it. Copyright law remains murky on AI-generated works, and publishers sitting on billion-dollar IP portfolios can’t risk characters entering public domain because an artist used Midjourney. The “six fingers” detail is telling: Marvel’s not asking artists to pinky-swear, they’re forensically analyzing submissions. For freelance comic artists watching AI tools improve weekly, these statements offer temporary relief but zero long-term protection. The moment AI art becomes legally copyrightable, these policies flip overnight. The Marvel Studios carve-out—props in Wonder Man allegedly “touched up by AI”—shows exactly where this is headed.
Marvel and DC are protecting IP, not artists. The second the Copyright Office clarifies AI-generated work is ownable, these policies vanish. The “humanity” framing from Jim Lee sounds noble, but the real driver is the legal department saying “we can’t guarantee ownership.” What’s more interesting is the enforcement—Marvel paying people to spot AI artifacts means they’re seeing enough AI submissions to warrant dedicated staff. The fact that Cebulski had to specify “Marvel Comics division” while Marvel Studios allegedly used AI on Wonder Man tells you where corporate priorities actually lie. Comic artists have maybe 18-24 months before “AI-assisted” becomes industry standard and these pledges get quietly retired.
Note: Commentary sections are editorial interpretation, not factual claims
Excellent analysis! Thanks for this insight. The move towards agentic commerce with Walmart leveraging ChatGPT is a pivotal moment for consumer AI adoption. It definetly highlights the practical application of large language models, making everyday purchases alot more efficient.