- Product & Design Pulse
- Posts
- Product & Design Pulse v85
Product & Design Pulse v85
Consequences Catch Up ⚖️
Malcom Jones
March 30, 2026
Welcome to this week’s edition of Product & Design Pulse, where we explore the latest in tech, product, design, and innovation! Last week was about consequences — the kind that arrive all at once after years of deferred reckoning. A Los Angeles jury found Meta and Google liable for designing addictive platforms that harmed a young user's mental health, establishing for the first time that product design choices can be treated as defective products and opening the floodgates to thousands of similar claims in what's being called Big Tech's Big Tobacco moment. OpenAI killed Sora and exited AI video generation entirely, taking Disney's $1 billion investment deal down with it — the same day Epic Games laid off 1,000 people, leaving Disney's new CEO watching $2.5 billion in tech partnerships unravel before he'd finished his first week. Meta continued its own painful restructuring, cutting 700 employees while awarding top executives up to $921 million each in stock options, and Anthropic — the company that built its brand on safety — accidentally leaked details of its most powerful unreleased model through an unsecured CMS. But amid the wreckage, the strategic picture sharpened: Apple paid rare retention bonuses to stop OpenAI from poaching its iPhone designers, signaling that the AI talent war has moved to hardware; YouTube's Neal Mohan declared the platform untouchable even as the addiction verdict loomed; and Ben Thompson argued at GTC that AI agents have changed the compute demand equation so fundamentally that we're not in a bubble at all. The week's lesson was stark — the companies that shipped too much, promised too much, or secured too little are now paying the price, while the ones that stayed focused are pulling ahead.
🎧 Audio Overview
For those who don’t have time to read 😁 |
Last week…
A Los Angeles jury found Meta (70% liable) and Google's YouTube (30% liable) negligent in designing their platforms to be addictive, awarding $6 million in combined compensatory and punitive damages to a 20-year-old plaintiff who testified that features like infinite scroll, autoplay, and algorithmic recommendations worsened her depression, body dysmorphia, and suicidal thoughts beginning when she was a child. The verdict — which came one day after a separate New Mexico jury ordered Meta to pay $375 million for failing to protect children from predators — was the first bellwether case in coordinated litigation involving thousands of similar claims, and its focus on product design liability rather than content successfully sidestepped Section 230 protections. For the industry, the dollar amount is negligible but the precedent is seismic: it establishes that platform design choices can be treated as defective products, opening the door to what many are calling Big Tech's Big Tobacco moment.
Apple awarded out-of-cycle retention bonuses worth $200,000–$400,000 in stock units vesting over four years to iPhone hardware designers, a direct response to OpenAI recruiting over 40 former Apple engineers — including former VP of product design Tang Tan — to build its AI hardware devices. Bloomberg's Mark Gurman framed the bonuses alongside Apple's broader AI strategy shift: the company is recommitting to its core hardware-and-services model, betting that distribution through billions of devices and system-level integration will matter more than building proprietary frontier models. For product leaders, this is Apple quietly acknowledging that the AI talent war has moved from software engineers to hardware designers — and that the biggest threat to its ecosystem may not be a better phone but an entirely new category of AI-native device.
OpenAI Shuts Down Sora, Killing Disney's $1 Billion Investment and Exiting AI Video Generation
OpenAI announced it is shutting down its standalone Sora video app and exiting the AI video generation business entirely, citing the need to redirect compute resources toward coding, enterprise, and reasoning capabilities ahead of a potential IPO — a move that simultaneously killed Disney's planned $1 billion investment and three-year character licensing partnership before any money changed hands. The decision reflects Fidji Simo's internal mandate to eliminate "side quests" and consolidate around OpenAI's enterprise-focused superapp strategy, as the company faces intensifying pressure from Anthropic's focused approach. For the entertainment industry, Sora's collapse is a reality check: the most hyped AI video product of the past two years couldn't survive contact with the economics of compute allocation, leaving Google as effectively the only scaled player in the space.
In a wide-ranging New York Times interview, YouTube CEO Neal Mohan argued that top creators will never fully leave the platform even as Netflix and other streamers recruit them, noting that no major YouTuber has ever completely pulled their content and that rival platforms inevitably "acquiesce" to what creators know is right for them long-term. Mohan also addressed AI slop, the addiction trial verdict, and YouTube's dominance — the platform is now the leading way Americans watch video, with 90% of U.S. teenagers on it and connected TV as its primary consumption surface. The interview reads as a victory lap from a platform that has become the default infrastructure of the creator economy, but the addiction verdict looming in the background raises the question of whether YouTube's scale is becoming a liability as much as an asset.
Fortune discovered that Anthropic left nearly 3,000 unpublished assets — including a draft blog post announcing a new model called Claude Mythos and a new "Capybara" tier above Opus — in a publicly accessible, unsecured content management system due to human error in CMS configuration. The draft described Mythos as "by far the most powerful AI model we've ever developed," with dramatically higher scores on coding, reasoning, and cybersecurity benchmarks than Opus 4.6, while also warning it poses "unprecedented cybersecurity risks." The irony of an AI safety company leaking its most sensitive model details through a basic infrastructure misconfiguration underscores a persistent gap between frontier capability and operational discipline — and it handed the Pentagon a public relations gift in its ongoing standoff with Anthropic.
Meta Lays Off 700 Employees While Awarding Top Executives Up to $921 Million in Stock Options
Meta laid off approximately 700 employees across Reality Labs, Facebook, recruiting, sales, and global operations — the second round of cuts this year — while simultaneously disclosing SEC filings showing it granted stock options worth up to $921 million each to four senior executives including CTO Andrew Bosworth and CFO Susan Li, in the first such awards since the company's 2012 IPO. The layoffs hit senior AI executives and directors as part of a consolidation of decision-making under Zuckerberg, with one researcher warning the reorganization risks turning Meta's AI division into "a feature factory." The juxtaposition of cutting hundreds of jobs while granting nearly a billion dollars in executive retention packages crystallizes the zero-sum economics of the AI transition: companies are spending more on compute and leadership retention while employing fewer people to justify it.
New Disney CEO Josh D'Amaro, barely a week into the job, watched two of the company's biggest technology partnerships collapse in a single day: OpenAI shuttered Sora — killing Disney's $1 billion investment and character licensing deal — while Epic Games laid off over 1,000 employees after declining Fortnite engagement threatened the $1.5 billion Disney-invested metaverse project that has yet to produce anything beyond a temporary minigame island. The back-to-back failures expose a deeper strategic problem: Disney has been writing billion-dollar checks to platform companies for participatory fan experiences without having a working vehicle to deliver them. For product leaders, it's a cautionary tale about the difference between announcing transformative partnerships and actually shipping products — and a reminder that the companies with the best IP don't automatically win in distribution formats they don't control.
Ben Thompson Argues AI Agents Make the Case That We're Not in a Bubble
Writing on the morning of NVIDIA's GTC 2026, Ben Thompson argues that the emergence of AI agents represents a third LLM inflection point — after ChatGPT and o1 — that fundamentally changes the demand equation for compute because agents operate autonomously, consume resources continuously, and generate economic returns on both the top and bottom line, making hyperscaler capex plans that blow past Wall Street expectations a rational response to real demand rather than bubble behavior. Thompson contends that agents will replace "human cogs in the organizational machine" not by eliminating jobs through a single decision but by making it economically impossible for large organizations to resist, fueling even more compute demand in a self-reinforcing cycle. The essay reframes the entire AI spending debate: if agents work as promised, the question isn't whether we're overinvesting — it's whether we're investing fast enough.
