Summary
In the heart of San Francisco, engineers chase the dream of artificial general intelligence (AGI), driven by the San Francisco Thesis—a mindset prioritizing rapid innovation toward superintelligence that outstrips human cognition. This article explores the global race for AI supremacy, contrasting the Bay Area’s bold vision with pragmatic and regulatory approaches worldwide. Former Google CEO Eric Schmidt warns that the first to build superintelligent AI—potentially by 2027—will shape global norms, urging democracies to lead or risk authoritarian dominance. Meanwhile, public caution in the U.S. (60% favor slowing AI, per Axios 2025), China’s state-driven AI push, and Europe’s weakened AI Act highlight a fractured landscape. Ethical divides among xAI’s truth-seeking, Anthropic’s Constitutional AI, and China’s control-focused approach underscore that innovation alone won’t define AI’s future—ethics will. The article calls for visionary leadership to balance ambition with human-centered values, ensuring AI serves humanity without sacrificing accountability or societal well-being.
A Mission District Dream
It begins in a dimly lit flat in San Francisco’s Mission District, where a group of engineers, fueled by caffeine and conviction, huddle over laptops. Their screens glow with neural networks, each line of code a step closer to artificial general intelligence (AGI). They’re not merely optimizing spreadsheets or moderating content; they’re chasing the holy grail of AI—recursion, where an AI can improve itself, iterate endlessly, and transcend human cognition.
This is the essence of the San Francisco Thesis: a mindset that believes the true frontier of AI lies not in incremental improvements but in creating a mind that surpasses ours. And they believe it’s imminent—perhaps within years, maybe even months. Outside, the world debates regulation and risk, but here, in the Bay’s foggy embrace, the focus is on acceleration. The mantra is clear: if we don’t build it, someone else will.
San Francisco, long a hub of technological innovation, is at the epicenter of this movement. From its early days with SRI International and Stanford’s AI lab to the current wave of startups like OpenAI and Anthropic, the city has consistently pushed the boundaries of what’s possible. Yet, this relentless pursuit raises critical questions: Is this mindset sustainable? Is it ethical? And can it truly lead to a future where AI serves humanity?
The Race for Supremacy: Eric Schmidt and the Frontier Vision
Eric Schmidt, former Google CEO, is a standard-bearer for this audacious vision. He predicts AI systems will soon outperform the world’s best physicists, artists, and strategists—not through rote training, but through self-improvement. Speaking at a recent conference, he suggested we’re just years away from superintelligent AI, a timeline echoed by forecasters like Daniel Kokotajlo and Scott Alexander, who peg 2027 as a pivotal year for AGI with impacts surpassing the Industrial Revolution (AI Report).
Schmidt doesn’t explicitly name it the “San Francisco Thesis,” but he embodies its core logic: speed is destiny. The first to build a truly capable AI will shape the political, ethical, and economic norms for generations. His warning is stark: if democracies hesitate, authoritarian regimes won’t. “Picture the world’s smartest system,” he says, “built without values we’d recognize” (TechCrunch). This urgency is reflected in San Francisco’s startup scene, where Safe Superintelligence, co-founded by Ilya Sutskever, raised $2 billion at a $32 billion valuation in April 2025 to build safe superintelligence for healthcare and education (Built In SF). Similarly, Reflection AI, launched by ex-DeepMind researchers, secured $130 million in March 2025 to develop autonomous coding agents (Bloomberg).
These developments underscore the San Francisco Thesis’s ambition: a race not just for innovation, but for defining the future.
Pragmatism vs. the Frontier: Global Perspectives
While Schmidt and San Francisco’s engineers push for velocity, others urge restraint. A March 2025 Axios poll revealed 60% of Americans want AI development slowed, citing fears of job loss and existential risks (Axios). The White House has issued executive orders to regulate AI, while OpenAI grapples with leadership changes and a shift toward commercialization, raising questions about its commitment to safety (OpenAI).
In contrast, China’s DeepSeek model, launched in 2024, integrates seamlessly into Beijing’s national strategy, prioritizing utility and dominance over public debate (Reuters). This divide is stark: the U.S. plays defense, China plays to win. Pragmatism—using AI to streamline industries or bolster cybersecurity—is rational but shortsighted. The San Francisco Thesis warns that incrementalism will be outpaced by recursive systems that don’t just solve problems but redefine what problems are worth solving.
Critics highlight the risks of this mindset. A January 2025 analysis draws parallels with San Francisco’s past infrastructure mistakes, like the Embarcadero freeway, which prioritized technology over social impact, displacing communities (TechPolicy.Press). This critique underscores the need for balance between innovation and societal well-being.
The European Dilemma: Balancing Innovation and Regulation
Europe offers a third perspective, caught between innovation and regulation. The EU’s AI Act, initially a robust framework for ethical AI, has faced significant rollbacks by 2025. The cancellation of the AI liability directive, which would have established accountability for AI-related harms, and the introduction of national-security carve-outs allowing AI in mass surveillance have sparked controversy (Carnegie Endowment). These changes, driven by France and big tech lobbying from companies like OpenAI and Google, reflect pressure to compete with the U.S. and China (Corporate Europe Observatory).
Former Italian Prime Minister Mario Draghi’s 2024 report warned that overly strict regulations threaten Europe’s economic edge, citing reliance on foreign semiconductors and cloud infrastructure (TechPolicy.Press). Yet, critics argue this deregulation undermines democratic oversight, erodes privacy and consumer protections, and risks Europe’s role as a global ethical leader (Pymnts). The EU’s shift, seen as a “lost battle” to competitive pressures, highlights the challenge of balancing innovation with ethical integrity.
Ethical Fault Lines: A Fractured Ecosystem
The path to superintelligence isn’t just technical—it’s ethical. The AI landscape reveals a fractured mosaic of priorities:
xAI, led by Elon Musk, bets on “truth” as its north star, arguing that ideological guardrails stifle discovery (Musk X Post).
Google DeepMind has shifted from a safety-first ethos to prioritize performance, driven by competitive pressures (TechCrunch).
OpenAI, despite its roots in ethical alignment, struggles under commercialization and leadership churn, clouding its moral stance (OpenAI).
Anthropic stands apart with “Constitutional AI,” embedding human-centered ethics as the foundation of intelligence, not an afterthought (Anthropic).
China sidesteps public ethical debates, with state oversight ensuring AI serves national goals—surveillance, education, defense—with stability and power as priorities (Reuters).
Critics like Thomas Wolf of Hugging Face call AGI visions “wishful thinking,” pointing to LLMs’ creative limitations (TechCrunch). Yann LeCun advocates for “world models” over LLM-based AGI (LeCun X Post). This patchwork reveals a truth: innovation alone won’t dictate AI’s future. Ethics—or their absence—will.
The Path Forward: Visionary Leadership Over Paralysis
The San Francisco Thesis exposes a deeper truth: pragmatism alone cannot navigate the AI frontier. Incremental gains won’t outrun recursive systems, nor will cautious regulation outmaneuver state-driven ambition. Europe’s regulatory rollback underscores the urgency of this moment. The question isn’t whether superintelligence will emerge—it’s who will shape it and how.
Democracies must blend the Bay’s audacity with ethical clarity. This means investing in AI research, fostering talent, and setting norms that prioritize human flourishing without stifling innovation. Anthropic’s Constitutional AI offers one model: ethics as a scaffold, not a shackle. xAI’s truth-seeking ambition offers another: boldness without apology. Europe must reclaim its role, balancing competitiveness with robust oversight to avoid strategic vulnerabilities.
The alternative is grim. If the U.S., EU, and their allies cede the frontier, others will define the values—or lack thereof—embedded in the systems that govern our future. The San Francisco Thesis isn’t just a mindset; it’s a call to act. Lead with vision, or follow in someone else’s shadow.
Share this post