From sneakers to servers: What Allbirds tells us about the AI gold rush

Hi AI Futurists,

A shoe company sold its brand for $39 million and rebranded as a GPU leasing business. Stock up 300%. If you want to understand where the AI market is right now, this is the story to watch.

Let's take a look.

Our agenda.

Best,
Lex Sokolin

P.S. Enjoying the newsletter? Have a suggestion? Hit reply and let us know!

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide

📊 Stanford AI Index 2026: China Nearly Closes the Performance Gap — The US lead over China's top model narrowed from 9.26% to 1.70% in a single year. Meanwhile, generative AI hit 53% population adoption. Faster than the PC or the internet.

📝 OpenAI Internal Memo: Microsoft "Limited Our Ability" to Reach Enterprise — OpenAI's CRO called inbound demand from the new Amazon/Bedrock partnership "frankly staggering." Nothing says partnership like writing memos about how your partner held you back.

🎬 OpenAI's Sora App Shutting Down April 26 — Active users collapsed below 500,000 while the product burned an estimated $15M per day in compute. Disney dropped its planned $1B investment. Turns out making video is easier than making money from it.

🧠 Meta Launches Muse Spark From New Superintelligence Labs — Led by Alexandr Wang, the first model from Meta's $14B AI overhaul is natively multimodal with three thinking modes. Outperforms GPT-5.4 on HealthBench Hard. Meta is done playing catch-up.

👓 Apple Testing Four Designs for AI Smart Glasses — Camera, microphone, and Siri that can analyze your surroundings. Apple is coming for Meta's Ray-Bans, and they brought four prototypes to the fitting.

🔬 Human Scientists Still Outperform AI on Complex Research — A Nature study found human researchers consistently beat the best AI agents on multi-step, open-ended scientific tasks. The benchmarks say one thing. The lab says another.

📉 Stanford: AI Transparency Scores Dropped Sharply — The Foundation Model Transparency Index fell from 58 to 40 points year over year. Organizational AI adoption hit 88%. The gap between deployment and accountability is widening.

💻 Google Releases Gemma 4 Open-Source Models — Built for reasoning and agentic workflows under Apache 2.0. Google's answer to Meta's open-model strategy, aimed squarely at developers building agents.

⚖️ Court Rules Platforms May Be Liable for AI-Assembled Ad Fraud — When a platform's AI has "ultimate authority" over ad content, the platform may own the liability. That affects Meta, Alphabet, Snap, TikTok, and X Corp.

📈 MIT: AI Benchmarks Are Broken — Coding benchmarks went from 60% to near 100% in a single year. Frontier models now match humans on PhD-level science. The capability curve is still accelerating, and the yardstick can't keep up.

Here's the setup. Allbirds, the sustainable shoe company that IPO'd at a $4 billion valuation in 2021, just sold its entire footwear brand for $39 million. Revenue had dropped 25% in 2024 and another 20% in 2025. Instead of winding down, the company raised $50 million in convertible financing, renamed itself NewBird AI, and announced a pivot into GPU-as-a-Service. The stock surged over 460%.

We've seen this exact pattern before. In 2017 and 2018, a wave of struggling public companies pivoted to blockchain. Long Island Iced Tea became Long Blockchain Corp. Kodak announced KodakCoin. The stocks popped. The businesses didn't. What makes this moment different, and in some ways more dangerous, is that AI infrastructure is a real market with real revenue. GPU supply is genuinely constrained. There's actual demand behind the narrative. So when a company like Allbirds announces a pivot, it sounds more plausible than a beverage company pivoting to distributed ledgers.

But the question is the same one we always ask: where is the margin, and why is it durable? Leasing GPUs works if you have proprietary procurement, long-term contracts with guaranteed utilization, and operational expertise in data center infrastructure. CoreWeave, Lambda Labs, and Crusoe Energy built those capabilities over years. Allbirds has a public listing and a press release.

What the capital markets are actually pricing here is the shell, not the business. A publicly traded entity that can raise capital, file an S-1, and access institutional money is worth something on its own. Attach it to the hottest sector in the market and the stock re-rates before a single GPU is racked. When the narrative is strong enough that a shoe company can raise $50 million by saying the word "GPU," capital is flowing on story, not substance.

For context, look at the Sora shutdown happening simultaneously. OpenAI built one of the most technically impressive video generation models in the world and still couldn't make the economics work. $15M per day in compute costs. If OpenAI can't make AI infrastructure profitable at their scale, a sneaker company with $50M isn't going to crack it. The AI boom is creating both real companies and real mirages. Telling the difference is the whole game right now.

Figma for Agents — Design with AI agents, connected to your design system

Fathom 3.0 — AI meeting notes: now bot-free, in ChatGPT and Claude

Luma Agents — Agents that plan, iterate, and refine with full creative context

This 158-page research report provides the first comprehensive taxonomy of public companies, private ventures, and tokenized protocols building the infrastructure for autonomous AI systems. Compiled by Lex Sokolin, former Chief Economist at ConsenSys, fintech strategist at Autonomous Research, and current Managing Partner at Generative Ventures, this report delivers institutional-grade analysis of 100+ companies across 14 critical infrastructure layers. Learn more here.

That’s all for today, folks!