# ARK Invest's Big Ideas 2026 Every year, ARK Invest publishes [Big Ideas](https://research.ark-invest.com/hubfs/1_Download_Files_ARK-Invest/Big_Ideas/ARKInvest%20BigIdeas2026.pdf), a report on how they think the future will look like. ARK publishes the most fascinating metrics, and I find the way they frame problems genuinely thought-provoking — even when I disagree. ## Two Overarching Themes Two ideas run through the entire report. **Convergence:** the idea that technologies are now catalyzing each other. Reusable rockets enable space-based compute for AI. Falling battery costs enable autonomous robots. ARK's "Convergence Network Strength" metric, which measures how much disruptive technologies are feeding into one another, rose 35% in 2025. I agree with them on convergence. At the very least, the catalyzing is more visible now than it was in the past. **Wright's Law**: as cumulative production of something doubles, the cost to produce it drops by a consistent percentage. Compare this to Moore's Law, which says computing power doubles roughly every two years. Wright's Law connects every section of the report: produce more, costs fall, demand grows, produce even more. ## What Stood Out ### AI Not-So-Slowly Takes Jobs ARK says AI now reliably handles about 31 minutes of an average worker's daily tasks. That means a $20/month subscription pays for itself in less than half a day for roughly half the working population. It's great to look at something other than the usual AI benchmarks. They're underestimating this. AI can already handle far more than 30 minutes of daily work. In January, I built a tool that automates about 600 hours of work a year at my office. In February, another that automates 100 more. That's more than 31 minutes a day. I've started to design my process around AI doing as much of my work as possible. ### AI Bubble Today's companies spend a lot more paying humans than they do paying for software. As AI starts doing more of an average worker's daily tasks, the amount companies become willing to pay for AI also increases by a lot. Eventually, companies pay more for software and less for humans, and this is how ARK says AI companies generate a lot more revenue. Current AI CapEx matches telecom-bubble-era spending, but market valuations are far less, implying we're not in a bubble. ARK compares old valuations to the "Mag 6," conveniently excluding Tesla, which is the most aggressively valued company in the Mag 7 and held by ARK. ARK also predicts that AI companies, who mostly make money from subscriptions, will soon generate multiples of their current revenue from selling leads and advertisements. This prediction makes their not-a-bubble position slightly more defensible. ### AI Adoption ARK highlights that AI adoption is outpacing internet adoption, and presents this comparison as a metric for AI's transformative potential. AI is certainly more transformative than the internet, and the most transformative technology to ever be diffused at scale. But the adoption curve says more about access than potential. Not everyone had a computer when the internet launched. Everyone had a phone when ChatGPT launched. AI access is also free. I don't have to pay extra like I do for the internet. ### No Neural Interfaces? ARK traces the evolution of how we interact with computers: typing commands, clicking, tapping, and now prompting. They don't mention thought-based interfaces. This is a glaring omission because brain-computer interfaces are a natural and imaginable progression, and ARK, if any, firm should mention this. ### AI Commerce ARK predicts AI will drive 25% of digital commerce transactions by 2030, up from 2% in 2025. Is ARK bullish on AI independently making purchasing decisions on behalf of consumers? "Help me find the best 24oz water bottle" will happen a lot more than "Find and buy the best 24oz water bottle for me". Their prediction that AI search will capture 60% of query share by 2030 suggests they're thinking more about AI shaping discovery and intent, which is more reasonable than full agent-driven purchasing. ### Tokenization The market for tokenized assets tripled to $19 billion in 2025. ARK projects $11 trillion by 2030 — a 500x increase. Sounds absurd unless public equities, sovereign debt, and bank deposits are tokenized, which is what ARK predicts. $11T is plausible, not probable, and I've always been long Ethereum at heart. DeFi applications are also more formidable than I thought. Tether at $180B and Circle at $80B in assets aren't too far off from Robinhood's $310B or Coinbase's $510B. Publicly listed DeFi applications are valuable. The percentage of crypto derivatives being traded on DeFi is rising. I always assumed DeFi meant worse UI and onboarding compared to CeFi. Their increased prevalence suggests the gap is closing. ### Robots ARK says generalized robots will create jobs we can't even imagine. I agree. We haven't even fully automated the majority of specific, repeatable tasks, which we can do with specialized robots. More drones and robots mean falling costs, and cost per delivery trends toward zero. The market for autonomous vehicles is much bigger than the market for ride-hailing. Autonomous vehicles are competing for all miles driven on the road, and ARK argues only Tesla has the capacity to produce such vehicles at scale. I agree with this, but Tesla's RoboTaxi rollout has underperformed my lowest expectations, and does not instill confidence.