🕶️ From Jobs’ memorabilia to Jensen’s waiting lists

Wearable health gadgets are quietly becoming a massive e-waste problem. A new study estimates that by 2050, these devices could generate over a million tons of discarded electronics, driven by demand rising to nearly 2 billion units a year. The carbon cost is just as steep—around 100 million tons of CO₂ over that period.

The surprising part isn’t the plastic bands or casings. Most of the environmental damage comes from the electronics inside. Printed circuit boards—packed with chips and hard-to-source metals—account for roughly 70% of the footprint. In other words, a compostable strap doesn’t fix a disposable device.

For startups, the takeaway is practical. Build wearables so the core electronics can be reused, refurbished, or swapped out. Plan for take-back programs, longer software support, and slower upgrade cycles. Regulations are heading this way anyway, and buyers are starting to ask for hard data on e-waste and emissions. Done right, reuse can lower costs, reduce reliance on scarce materials, and unlock subscription-style hardware models.

Are we repeating 1929—or just rhyming with it? The Roaring Twenties ran on debt, tech hype, speculation, and inequality. Sound familiar? This episode compares the core triggers of the Great Depression with today’s world heading into 2029—leverage, AI hype, global debt, and a widening wealth gap—and asks what’s truly similar, what’s different, and whether modern guardrails are enough to stop history from repeating.

Founders make hundreds of decisions every day. What to wear shouldn’t be one of them. WFH Studio built a 5-day T-shirt system for entrepreneurs who value focus: clean for video calls, comfortable all day, 100% cotton.

One shirt per weekday. Zero mental overhead.

Nvidia unveiled Rubin, its next-generation chip platform for large AI systems. It’s already in production, with volume ramping later this year, and promises big jumps in speed and efficiency over today’s chips. In theory, that should make AI cheaper—if you can get access.

In reality, the biggest cloud players will likely grab supply first, leaving startups dealing with tight capacity into 2026. Inference should get cheaper before training does, which favors agent-style systems. For now, Nvidia—and its manufacturing bottlenecks—still set the tempo of the AI economy.