Today at Bloomberg Tech San Francisco, Anthropic President Daniela Amodei discussed the growing AI wealth gap and her recommendation for redistribution. We’ve seen the direction this new economy is taking, just last month Sam Altman walked into a Y Combinator dinner and made what one YC partner called a “mic drop” offer: $2 million in OpenAI tokens, in exchange for equity, for every startup in the current batch. Compute is no longer a line item on an AWS bill. It is the asset class. At NVIDIA GTC 2026, Jensen Huang mentioned that in salary negotiations, prospective engineers are asking how many tokens come with their jobs. Compute access is becoming a determining factor in evaluating compensation packages, with technically skilled workers treating it as a non-negotiable. This insight into the evolving Silicon Valley hiring landscape offers a window into how compute access is being distributed, and who is getting left out of the conversation as a labor displacement curve that many forecasters aren’t covering begins to grow. 

The World Economic Forum’s Future of Jobs Report 2025 projects 92 million roles displaced by 2030, with Goldman Sachs expecting generative AI to affect 300 million full-time jobs. These reports also project creation of roughly 170 million jobs, but the headlines don’t cover the specific roles being eliminated. Administrative, clerical, and customer-facing roles are among those most impacted by increased workplace automation, whereas AI development and deep technical work are seemingly more “safe”. The overlap between populations holding the respective roles, by most labor data, is minimal. 

According to the UNDP’s December 2025 report, “The Next Great Divergence,” roughly two in three people in high-income economies use AI tools regularly, whereas approximately just 5% in many low-income countries use the same tools. The IMF similarly projected AI’s benefits for advanced economies could be more than two times that of the growth benefits for low-income economies. It’s clear that access to agentic systems can unlock new streams of income and advance economic opportunities, but if barriers to this technology result in those that stand most to benefit lacking access while corporations and wealthy individuals profit, then this ultimately is a failure. This is why Universal Basic Compute, the agentic era’s answer to Universal Basic Income, can be incredibly powerful if executed correctly at scale.

Universal Basic Compute would bring democratized access to people throughout the world and ensure that everyone, everywhere can harness the power of AI. To distribute compute at global scale to billions, ensuring real humans that qualify are able to get it was the central distribution challenge – until the recent emergence of proof-of-human and identity solutions.

I’ve spent years as the co-founder of Self.xyz working to solve this problem, addressing how an accessible, effective, and privacy-first approach to human verification can reach users across the world. While we’ve seen large-scale efforts from some of technology’s brightest minds, tools that require external hardware or that compromise user data privacy ultimately cannot, and in some cases should not, reach widespread adoption. The answer is here: leveraging what most people already own, a smartphone, their government-issued ID, with underlying technology built to ensure data privacy.

Biometric passports and national IDs, issued to international standards across 174 countries, contain NFC chips with digitally signed identity data. By connecting this with zero-knowledge infrastructure, individuals can verify their unique humanity and identifiers (age, country of residence, etc.) without ever sharing that underlying information with any third party. 

Industry leaders like Google have already deployed the technology, with pilots signaling the viability of using it for Universal Basic Compute. Through the forthcoming Google Cloud USA₮ Faucet, the US Dollar-backed stablecoins will be accessible for free to those who verify their humanity and eligibility. This platform is a blueprint for how Universal Basic Compute or even Universal Basic Income could be structured, eliminating the risk of bots draining funds and ensuring that real humans meet criteria such as 18 or older, or not a resident of an OFAC-sanctioned country. 

The case might be made that there is no economic incentive for hyperscalers to distribute compute freely, especially as the market demand is a core revenue driver. However, the cost for companies generating billions in revenue to allocate compute is negligible relative to the cost individual companies will face if sanctioned for restricting access, and more importantly the societal cost of millions left behind by the rapidly developing AI driven economy. Broadening access would help address the EU AI Act, proposed US AI regulation, and global antitrust concerns with outsized positive impact for communities worldwide.