Sunday Signals: 30,000 jobs gone by email, Big Tech's tobacco moment, and the standoff nobody wants to name
Oracle's AI trade-off laid bare, a jury calls social media a defective product, and the emerging tension of design/PM/engineering overlap.
Before we get into the heavy stuff, I want to start somewhere better. On Tuesday, four astronauts launched aboard Orion and left Earth orbit for the first time since Apollo 17 in 1972. As I’m writing this, they’re cruising toward the moon, about to break the distance record set by Apollo 13. Fifty-four years between crewed flights beyond low Earth orbit, and we’re finally doing it again. Reid Wiseman, Victor Glover, Christina Koch, and Jeremy Hansen are farther from the planet right now than any human being has been in half a century, and there’s something genuinely exciting about that, even when the rest of the week’s news makes you want to close your laptop.
Also, I love this: during a livestream at 2 a.m. on Thursday, Wiseman radioed Houston because he had two Microsoft Outlooks open on his onboard computer and neither one worked. NASA had to remote into his machine from 90,000 miles away. We can send humans around the moon but can’t make Outlook behave. Some things are truly universal.
Last thing on Artemis: this official NASA poster is sick.
You can grab the full, high-resolution version from Wikimedia for background wallpaper purposes.
One more thing before we get into the week's news: a review paper dropped this week in Brain Sciences that I think deserves a minute, especially if you’ve ever fallen down the “is consciousness quantum?” rabbit hole like me (I mean, who hasn’t…?) Oscar Arias-Carrión, Emmanuel Ortega-Robles, and Elías Manjarrez surveyed the full landscape of quantum-inspired theories of consciousness — microtubule vibrations (shoutout to an OG, Roger Penrose), nuclear spin networks, zero-point field resonance, entanglement-structured learning, all of it — and tried to assess where the actual evidence stands as of right now.
What I find interesting about this topic is how consciousness continues to be so resistant to explanation that serious researchers keep reaching for the most exotic physics available, and even then the best honest answer is “we don’t know yet.” Not every hard problem has a breakthrough around the corner, even one as fundamental as “what is consciousness even?”
Okay. Now the heavy stuff. Oracle cut up to 30,000 people this week via a 6 a.m. email so it could take on $58 billion in debt to build AI data centers. A Los Angeles jury found Meta and YouTube liable for designing platforms that are deliberately addictive, applying the same legal framework used for tobacco to interface design decisions. The job market data keeps showing design roles flat while PM and engineering surge. And AI enforcement is intensifying across a dozen states even without a federal law.
A lot happened. Here’s what I’ve been tracking this week.
Oracle’s $58 billion answer to “what are people worth?”
The numbers are stark enough on their own. Oracle cut between 20,000 and 30,000 employees this week, roughly 18% of its global workforce. The hardest-hit divisions were Revenue and Health Sciences SaaS, each losing about 30% of staff. The method was a mass email at 6 a.m. EST on Tuesday, signed only “Oracle Leadership,” with system access revoked within hours and severance details promised via DocuSign.
But the number that turns this from a layoffs story into a systems story is $58 billion. That’s how much new debt Oracle has taken on in two months to fund AI infrastructure. TD Cowen analysts estimated the headcount reduction could generate $8 to $10 billion in incremental free cash flow, which means the math here is explicit: the company calculated exactly how many human salaries it takes to finance an AI buildout, and then it cut them.
The mechanism is the same one buried in every “AI transformation” pitch deck that lands on a leadership team’s desk, even if nobody says it out loud. Oracle just said it out loud.
A jury just called social media a defective product
On March 25, a Los Angeles jury found Meta and YouTube liable for designing platforms that are deliberately addictive, and ruled that both companies knew their products harmed minors and failed to warn users. The plaintiff, identified as KGM, began using YouTube at age 6 and Instagram at age 9, posting 284 videos before finishing elementary school. She developed body dysmorphia, anxiety, and depression by age 10, with a therapist testifying that her social media use and self-image were inseparable.
The damages were $6 million total, split 70/30 between Meta and YouTube. The dollar amount is almost irrelevant compared to the legal logic underneath it: the jury treated infinite scroll, autoplay, algorithmic recommendations, and persistent notifications as design defects, the same framework that held tobacco companies liable for engineering addiction, now applied to interface design decisions.
Internal Meta documents entered into evidence showed executives discussing efforts to bring in users as young as possible, with one memo noting that 11-year-olds were four times as likely to keep returning to Instagram as users of competing apps. Another referenced the company’s strategy to “win big with teens” by “bringing them in as tweens.”
There are roughly 2,000 similar lawsuits pending. Amnesty International called for the verdict to lead to structural design changes. Both companies have said they’ll appeal, but the precedent is set: a jury of twelve people looked at how these products were designed and concluded the design itself was the problem. If you’ve ever worked on engagement metrics, growth loops, or retention features, that framing should land differently now.
Jury finds Meta, YouTube liable for social media addiction — Al Jazeera
Jury finds Meta and Google negligent in landmark trial — NPR
Landmark verdict must lead to design changes — Amnesty International
What the verdict means for social media design — Scientific American
The standoff nobody wants to name
Fast Company ran a piece this week asking why designers, engineers, and PMs are in a “three-way standoff,” and the framing is off but the observation is right. The actual dynamic looks a lot more like a merger that nobody agreed to and nobody’s leading.
The data from Lenny’s report that I flagged last week keeps connecting to new things. PM openings are at a three-year high (7,300+), engineering is at 67,000+ globally, and design has been flat since 2023. But the really interesting data point from the Fast Company piece is about team ratios: some companies have moved from 1 PM per 4 engineers to 2 PMs per 1 engineer, which isn’t a tweak to the existing model so much as a completely different organizational theory about what product work actually is.
Meanwhile, 64% of product builders now identify with two or more roles, according to Figma’s annual report. PMs are expected to understand design and data. Designers are expected to understand constraints and metrics. Engineers are closer to discovery and customer conversations than they’ve ever been. The boundaries that used to define these jobs are dissolving, and the teams that are winning are the ones that stopped trying to police the overlap.
The part that sticks with me: a lot of design’s organizational leverage came from being the only function that could do a certain kind of work. But when everyone can produce a version of that work with AI-assisted tools, the leverage has to come from somewhere else. Strategy, systems thinking, the judgment calls that AI can’t make and that don’t fit neatly in a portfolio but are impossible to ship without. I wrote about this shift in detail in From Craftsperson to Conductor, and the data from this week keeps reinforcing the same thesis: the role is changing shape whether we’re ready or not, and the people who adapt fastest are the ones treating the overlap as the job rather than a threat to it.
Why are designers, engineers, and product managers in a ‘three-way standoff’? — Fast Company
State of the Product Job Market in Early 2026 — Lenny’s Newsletter
AI enforcement is here, with or without a federal law
Morgan Lewis published a thorough landscape analysis this week on AI enforcement, and the headline finding runs against most people’s assumptions: there’s no comprehensive federal AI law, but enforcement is intensifying anyway. The FTC is going after companies making misleading AI claims under Section 5. The SEC is targeting “AI washing” in investor disclosures. State attorneys general are using existing consumer protection statutes to prosecute AI-related harms. And seven distinct categories of AI litigation are emerging, from algorithmic pricing antitrust cases to copyright claims over training data to biometric privacy violations.
The pattern is familiar to anyone who’s watched regulatory cycles: when the legislature stalls, enforcement fills the gap through creative interpretation of existing authority. For product teams, the question has moved well past “is there a law?” and into “which of the fourteen overlapping enforcement frameworks applies to what we’re building?” The Colorado AI Act takes effect June 30, Indiana’s privacy enforcement grace period just started April 1, and universal opt-out mechanisms are now required in 12 states. None of it is theoretical.
Agentic AI meets the DMV
Deloitte’s Government Trends 2026 report landed this week, and the most interesting section is on how agentic AI is starting to reshape government service delivery. Not in a theoretical way. Estonia’s Bürokratt system is coordinating services across agencies in real time, and Abu Dhabi’s TAMM platform now connects over 1,000 services from more than 90 providers into a single experience that adapts to individual needs.
The shift here is from siloed, form-driven government websites to systems that figure out what you actually need and coordinate the response across departments you’d never know existed. If you’ve ever spent an afternoon bouncing between three different government websites trying to figure out which agency handles your specific problem, you understand why this matters. And if you work in service design or civic tech, the pattern is one you’ve been advocating for: stop designing around the org chart and start designing around the person.
The catch, and Deloitte is clear about this, is that agentic systems handling sensitive government data need governance frameworks that don’t exist yet in most jurisdictions. The technology is ahead of the policy, which should sound familiar by now.
Experiential retail becomes the baseline
A NetChoice report this week confirmed something that’s been building for a while: 46% of retail leaders now rank omnichannel experience as their top growth priority, and the stores getting the most investment aren’t the ones optimizing square footage for product density. They’re the ones turning floor space into experience.
The most visible example right now is Wake The Tiger, which is opening an 80,000-square-foot immersive art and retail space at Westfield London. But the trend goes deeper than flagship spectacles. Stores are being redesigned as fulfillment hubs, community spaces, and experience centers where the primary function of the physical space is to do something a screen can’t. The retail teams that are winning this aren’t thinking about “in-store vs. online” anymore, they’re thinking about what the physical environment can offer that justifies someone getting off the couch, and then connecting that moment to everything digital.
If you’re in spatial design, experience design, or working on the physical/digital boundary, the job is changing fast. The stores being built right now look more like exhibits than shops, and the design challenge is closer to environmental storytelling than traditional merchandising.
One more signal
Salone del Mobile 2026 runs April 21-26 in Milan, and the theme shift is worth noting. Last year was human-centric. This year is “A Matter of Salone,” which moves the conversation from the person to the raw material itself, with circularity, design-for-disassembly, and material passports that trace the origin and lifecycle of every component taking center stage. Exhibitors are expected to bring biomaterials like mycelium and algae-based plastics alongside a revival of low-impact craft techniques, and SaloneSatellite’s theme is “New Crafts for New Worlds.” OMA is designing the master plan.
1,900 exhibitors from 32 countries. If you’re in industrial or physical product design, this is the event that sets the vocabulary for the next two years, and this year it’s saying we’ve moved past debating whether sustainability matters and into how it gets built into the material itself.
Coming this week: The next three issues of the Design Your Next Move series are in the pipeline, and they pick up right where From Craftsperson to Conductor left off. First up is The New Skill Stack, which gets specific about which hard skills are actually shifting and which soft skills are about to matter more than ever — generative UI, non-deterministic systems, and why the designers getting hired are the ones working at the intersection of judgment and capability. After that, Make the Invisible Visible digs into the work nobody sees but everyone depends on: stakeholder navigation, team protection, problem reframing, and a simple framework for documenting the contributions that never make it into a portfolio. And then The Career Lattice lays out four paths beyond the default ladder, because the assumption that more authority equals more fulfillment stopped being true for a lot of us a while ago.
See you tomorrow — Justin