The ‘San Francisco consensus’ is Silicon Valley’s latest buzzword—and it echoes an earlier moment in international development: the Washington Consensus that reshaped economies from Morocco to much of the Global South. The parallel is more than rhetorical. A “consensus” signals a shared framework for organizing economic life and technological change. Yet the differences are as telling as the similarities.

The Washington Consensus was driven by governments and international institutions, promoting liberalization, privatization and integration into global markets. The San Francisco Consensus, by contrast, is taking shape within the technology sector itself, propelled less by public policy than by private corporate power and the competitive logic of innovation. As debates around artificial intelligence—and, increasingly, the idea of recursive self-improvement have gained momentum over the past two years, this new Consensus has begun to crystallize. Set side by side, these two moments pose a sharper question: how technological frontiers are made, and who they are made for.

California has been one such frontier before. In the mid-19th century, it was a landscape of extraction. Gold drew people westward, fortunes were made overnight, and the terrain itself was reorganized around the promise of sudden wealth. A century later, another frontier emerged in the same place—less visible, but no less transformative. Silicon Valley did not mine the earth so much as it mined information, and, increasingly, human behavior.

Today, that infrastructure is no longer merely connective; it is extractive and predictive. A handful of firms do not simply facilitate  communication but actively model and shape human behavior at scale. The shift from open networks to platform capitalism has produced new forms of concentration, surveillance and control. What was once framed as liberation now feels more like atmosphere, inescapable and privately governed. Power has not vanished; it has been reconfigured. A Silicon Valley elite—engineers, executives and investors—now commands the digital means of production, while a dispersed global workforce performs the hidden labor that sustains these systems. From within this landscape, a new sensibility has emerged, less utopian, more anxious. This is what some technology leaders, such as Eric Schmidt, have begun to call the San Francisco Consensus.

At its core, this consensus reflects not simply the expectation of automation, but the anticipation of acceleration. The horizon is no longer machines that assist or replace workers, but systems that begin to improve themselves. What engineers increasingly anticipate is a shift toward recursive self-improvement: AI systems capable of refining their own code, directing their own learning, and extending their capabilities without human instruction. Early signs are already visible in systems that write software, generate mathematical conjectures, and discover new knowledge. Once improvement becomes self-directed, technological change ceases to follow familiar trajectories. It becomes recursive—and potentially self-reinforcing.

This possibility intensifies the anxiety surrounding AI. The concern is not simply that machines will take jobs, but that they will outpace human capacity to remain economically relevant. Within the industry itself, displacement is no longer treated as an unintended side effect but as a benchmark of progress. Models are evaluated by their ability to perform real-world tasks, often as proxies for replacing human workers. Companies, responding to both competitive pressure and investor expectations, are already reorganizing: laying off employees, flattening career ladders, and reorienting toward a future in which human input is increasingly marginal.

Yet labor does not vanish—it is reorganized. Beneath the rhetoric of automation lies a fragmented global workforce performing the invisible “ghost work” that makes AI systems function. These workers—moderating content, labelling data, correcting outputs, remain essential, even as their labor is obscured. Automation, in this sense, displaces labor rather than eliminating it, embedding human effort within systems that present themselves as autonomous.

If the Washington Consensus once promised that integration into global markets would generate development, the San Francisco Consensus suggests a more unsettling possibility: that the market itself may no longer require most people. And if recursive self-improvement accelerates, that transition may unfold far more quickly than earlier technological shifts.

For much of the past three decades, digital technologies have appeared to expand opportunity in the Global South. Mobile phones, the internet and digital platforms enabled what came to be known as ‘leapfrogging’: bypassing older stages of development by adopting new technologies directly. In practice, this meant sidestepping costly infrastructure—landlines, banking systems, legacy media, and moving into more flexible systems. Mobile banking enabled financial inclusion, while digital platforms opened access to global markets.

Leapfrogging depended not only on technology, but on timing and openness. Earlier digital systems were sufficiently modular to allow adaptation. They enabled improvisation, with local actors repurposing global tools and building hybrid systems suited to their own contexts. Even within unequal systems, there remained space for maneuvering.

Morocco offers a telling example. Over the past two decades, it has invested in digital infrastructure, renewable energy and industrial platforms that have positioned it as a regional hub linking Africa, Europe and the Atlantic world. Initiatives in e-governance, fintech and logistics allowed it to bypass certain legacy systems while integrating into global value chains on strategic terms. These efforts did not eliminate dependency, but they demonstrated how targeted investment could create room for maneuvering.

The emerging San Francisco Consensus around artificial intelligence casts Morocco’s development trajectory in a new light. As the World Bank’s 2026 report, Scaling the Atlas: Growth and Jobs for a Prosperous Morocco, argues, Morocco has maintained macroeconomic stability, developed world-class infrastructure, and expanded its integration into global markets, yet it continues to struggle with weak job creation, stagnant productivity, and declining labor-force participation, especially among women and youth. At the heart of this dilemma is a growth model heavily reliant on public-sector-led capital accumulation. While investment has modernized key sectors of the economy, productivity gains have remained uneven, and private-sector dynamism and labor absorption have lagged behind.

The relevance to the San Francisco consensus is striking. As AI systems accelerate productivity, they may also intensify a broader global pattern in which economic growth becomes increasingly decoupled from human labor and employment creation. Morocco’s experience offers an early warning: technological modernization alone does not guarantee broad-based prosperity. The report ultimately argues that only deeper structural reforms, fostering competitive markets, dynamic firms and greater workforce inclusion, can transform growth into high-quality employment. In the age of AI, that lesson extends far beyond Morocco.

Artificial intelligence changes these conditions. The barriers to entry are no longer primarily infrastructural but computational and epistemic. Building advanced AI systems requires access to water and electricity, large-scale data, specialized hardware and technical expertise—resources that are tightly concentrated. The relative openness that enabled earlier forms of leapfrogging has given way to enclosure. And with the prospect of recursive self-improvement, that enclosure may deepen, as those who control the most advanced systems also control the pace and direction of innovation.

Leapfrogging, however, does not disappear—it changes form. Instead of bypassing infrastructure, countries may seek to reconfigure it: investing in regional data centers, developing local language models, or building public digital systems that prioritize collective ownership. Leapfrogging in the age of AI becomes less technological than institutional—experimenting with governance frameworks and data regimes that depart from dominant models. In some cases, it may involve strategic refusal: choosing not to adopt systems that reproduce dependency.

The consequences extend beyond  economics. Algorithmic systems now shape domains once governed by human judgment—credit, policing, and welfare— replacing discretion with calculation. Yet these systems often reproduce historical inequalities embedded in their data. What appears as objectivity can function as a new form of classification and control.

These shifts are also visible in labor and migration. In the Global North, AI is eroding professions once considered secure. In the Global South, it threatens pathways into global labor markets—outsourcing, manufacturing, and service work—before they have fully matured. The result is a paradox: economic and environmental pressure increases the desire to migrate, even as automation reduces demand for labor. More people seek movement; fewer opportunities exist.

At the same time, digitally mediated labor creates new forms of participation. Workers engage in global markets without moving physically. Yet this “virtual migration” often reproduces asymmetries of power. Work is governed by opaque systems that can include or exclude without explanation. Access fluctuates; livelihoods remain fragile.

In sum, these dynamics point to a shift from opportunity to constraint. Where earlier technologies enabled leapfrogging, AI risks producing lock-in. Countries that do not control the infrastructure, data or models that underpin AI become dependent on external systems— not only materially, but in how knowledge itself is produced and circulated.

This brings the question of sovereignty into focus. The challenge is not simply to adopt AI, but to shape it. Who owns the data? Who builds the models? Who governs their use? Without meaningful answers, participation risks becoming extraction.

What, then, would it mean to shape this transformation? Not a single policy, but a strategic reorientation: investing in public data infrastructures, regional compute capacity, and education systems that treat AI as something to be built and contested. It requires collaboration—South–South alliances, open-source ecosystems, shared linguistic datasets, and governance frameworks that treat data as a collective resource. Without such efforts, the trajectory is clear: the Global South risks becoming a source of raw input rather than a producer of intelligence.

Seen in this light, the shift from the Californian Ideology to the San Francisco Consensus is less a story of progress than of reconfigured power. What began as a vision of decentralization has consolidated into a system of concentration. The frontier has not disappeared; it has been internalized—encoded in algorithms and scaled globally.

The Californian dream has always been future-facing. But futures are made, not found—and unevenly so. The question is no longer whether AI will transform the world—it already is—but who will shape that transformation.

For the Global South, this is an immediate horizon. Whether AI deepens dependency or enables autonomy will depend on decisions being made now—about infrastructure, governance and ownership. As Morocco’s earlier trajectory suggests, space can be created within unequal systems—but only through deliberate strategy.

If the age of artificial intelligence becomes one of algorithmic dominance, it will not be because it had to be. It will be because existing structures of power were left intact. But if those structures are contested—if alternative infrastructures and alliances take shape—then the future of AI may yet be plural rather than unequal.

Also published in Morocco World News

Ilahiane, a Moroccan-born author and applied cultural anthropologist, is a Professor at the School of Middle Eastern and North African Studies and the W. A. Franke Honors College at the University of Arizona.