- Product & Design Pulse
- Posts
- Product & Design Pulse v87
Product & Design Pulse v87
Show Your Hand 🃏
Malcom Jones
April 13, 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 showing your hand — and in some cases, changing it entirely. Anthropic launched Project Glasswing, a coalition with AWS, Apple, Google, Microsoft, and eight other partners built around its unreleased Claude Mythos Preview model, which has autonomously found thousands of zero-day vulnerabilities in every major operating system and browser — turning an embarrassing CMS leak into a $100M defensive cybersecurity initiative backed by the industry's biggest names. The company also revealed its revenue run rate has hit $30 billion and signed a 3.5-gigawatt compute deal with Google and Broadcom, while The Verge declared AI coding the first truly mainstream AI use case and the central battleground for every major lab's IPO ambitions. Meta made the week's most dramatic strategic reversal, launching Muse Spark as its first fully closed-source model just eighteen months after Zuckerberg championed open source — an acknowledgment that Chinese competitors used the Llama playbook to overtake it. Meanwhile, Epic Games pinned its post-layoff recovery on a Disney extraction shooter that internal testers have already flagged for unoriginal mechanics. The pattern is unmistakable: the companies that built their positions through openness in 2025 are now racing to close them, because in 2026 the question isn't who can build the most but who can control what they've built.
🎧 Audio Overview
For those who don’t have time to read 😁 |
Last week…
Anthropic announced Project Glasswing, a coalition with AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks, built around an unreleased frontier model called Claude Mythos Preview that has autonomously discovered thousands of zero-day vulnerabilities — including some surviving decades of human review — in every major operating system and web browser. The model scores 83.1% on CyberGym (vs. 66.6% for Opus 4.6) and 93.9% on SWE-bench Verified (vs. 80.8%), with Anthropic committing $100M in usage credits and $4M to open-source security organizations while explicitly choosing not to make Mythos Preview generally available due to its dual-use cybersecurity risks. For the industry, this is Anthropic converting its most controversial capability — the same one leaked via its CMS weeks ago — into a strategic asset, positioning itself as the indispensable defensive infrastructure layer while simultaneously making the case that frontier models require controlled distribution rather than open access.
Epic Games Pins Fortnite Recovery on Disney Partnership as Multiple Game Modes Shut Down
Bloomberg reports that Epic is developing a Disney-themed extraction shooter — the first game from its $1.5 billion Disney partnership — targeting a November 2026 launch, while simultaneously shutting down multiple underperforming Fortnite modes including Ballistic, Festival Battle Stage, and eventually Rocket Racing after acknowledging that many recent updates and games had flopped. Internal playtesters have raised concerns that the Disney game's mechanics lack originality, and Disney has reportedly expressed disappointment with progress on the other two partnership games, prompting resource shifts between projects. For product leaders, this is a case study in the risk of staking recovery on a single high-profile partnership when the underlying product execution challenges remain unresolved.
Fubo Upgrades Mobile Apps With AI-Powered Sports Highlights and Real-Time Game Alerts
Fubo unveiled a major upgrade to its iOS and Android apps using proprietary AI to transform mobile sports streaming from passive viewing to an active, moment-driven experience — featuring live video in home screen carousels, AI-generated vertical highlight clips for Team Channels, one-tap push notifications that jump directly to key plays like home runs, and breaking news alerts with instant live coverage access. The update reflects Fubo's strategy of extending beyond the living room to capture the on-the-go sports audience that increasingly relies on smartphones for quick updates between full broadcasts. For streaming product teams, this is a useful example of how AI-powered personalization and clip-level content indexing can turn a linear TV app into something closer to a sports-first social feed — the kind of behavior shift that could matter more for retention than adding new channels.
Anthropic signed an expanded agreement with Google and Broadcom for 3.5 gigawatts of next-generation TPU compute capacity coming online in 2027, while revealing that its annualized revenue run rate has surged to $30 billion — up from $9 billion at the end of 2025 — with enterprise customers spending over $1 million annually more than doubling from 500 to over 1,000 in less than two months. The deal extends Anthropic's diversified chip strategy across NVIDIA GPUs, Google TPUs, and Amazon Trainium, making Claude the only frontier model available across all three major cloud platforms. For industry watchers, this announcement crystallizes how quickly the AI infrastructure race is escalating: Anthropic is now making compute commitments at hyperscaler scale, and its growth rate suggests Claude Code and enterprise adoption are driving demand that even the company's aggressive fundraising may struggle to keep pace with.
Meta Superintelligence Labs, led by former Scale AI CEO Alexandr Wang after Meta's $14.3 billion investment, released Muse Spark — a natively multimodal reasoning model with a novel "Contemplating" mode that runs multiple sub-agents in parallel — as an entirely proprietary product with no public weights, breaking from the open-source Llama strategy that Mark Zuckerberg championed as recently as 2024. The model scores competitively but trails Gemini 3.1 Pro and GPT-5.4 on most benchmarks, while leading on health-related tasks thanks to training data curated with over 1,000 physicians, and Meta claims it achieves Llama 4-level performance with 10x less compute through a technique called "thought compression." For the AI ecosystem, this is the most significant strategic reversal in open-source AI: the company that built the most widely downloaded open-weight model family has concluded that open source became a competitive liability — largely because Chinese competitors like Qwen used Llama's playbook to overtake it — and is now locking its frontier work behind the same walls it once criticized rivals for building.
The Verge argues that AI-assisted coding has emerged as the first genuinely mainstream AI use case and the primary commercial battleground for OpenAI, Google, and Anthropic — driven by Claude Code's viral adoption in late 2025, which coincided with an explosion in Anthropic's revenue and prompted OpenAI executives to explicitly refocus on competing with Claude Code rather than pursuing "side quests" like Sora. All three companies are now trying to build AI super apps with coding at the center, while simultaneously moving to close the broader ecosystem — as seen in Anthropic's recent attempt to ban OpenClaw — to lock developers into their own platforms. For product leaders, this is the clearest signal yet that developer tools, not consumer chatbots, are where AI companies see their path to IPO-ready revenues — and that the window for independent AI coding startups like Cursor and Windsurf to remain competitive is narrowing as the platform companies consolidate.
