What Does Xi Jinping’s Landmark AI Speech Mean for the Global South?
A future of Digital Westphalia or techno-colonialism? The former aligns with an equitable, sustainable digital order; the latter risks repeating history’s injustices in technological garb.
By Warwick Powell
On 17 July 2026, Chinese President Xi Jinping delivered a keynote address at the World AI Conference (WAIC) in Shanghai, articulating a vision for global AI governance centered on openness, inclusivity, human control, and shared prosperity.
His speech emphasized a people-centered approach to AI, rejecting technological monopolies and digital divides that could replicate historical injustices. Xi announced concrete measures: 5,000 AI training opportunities for developing countries over five years, international AI application cooperation centers with organizations including ASEAN, the African Union, BRICS, the Shanghai Cooperation Organisation, and others, and the rollout of China’s AI-powered meteorological warning system (MAZU) to 30 countries. He also highlighted the establishment of the World Artificial Intelligence Cooperation Organization (WAICO) in Shanghai.
This moment marks a pivotal contrast in the global AI landscape. It pits a Chinese-led model of collaborative, open-source empowerment — positioning China as what I have previously described as a “great enabling power” aligned with the Global South — against an American approach that prioritizes proprietary control, national security exceptionalism, and the pursuit of AGI as a western civilizational project.
The stakes are high, revolving around moving from a US-dominated order of technological rents, cross-border expropriation, and hegemony toward a multipolar digital sovereignty that echoes the Westphalian principles of state autonomy and non-interference, adapted to the information age.
Drawing on my previous analyses, this essay examines the structural, economic, philosophical, and geopolitical dimensions of this shift. It argues that the US tech stack — characterized by high costs, security risks, low sovereignty, and entropic pressures — is facing systemic limits, while China’s open-source ethos, hardware-software co-optimization, energy advantages, and capacity-building initiatives offer a viable alternative.
The Global South faces a clear choice: empowerment through accessible, sovereign AI or lock-in to a costly, extractive model that exacerbates inequalities. It’s also a “choice” citizens of the Global North need to confront, before it’s too late for their own societies and economies.
Techno-Colonialism: The US Model of Hegemony and Rents
The American AI paradigm embodies what Cedric Durand and Yanis Varoufakis describe as “techno-feudalism” extended globally; I call this techno-colonialism. In this model, a handful of firms (OpenAI, Google, NVIDIA, etc.) capture economic rents through proprietary systems, data monopolies, and control over critical infrastructure. This mirrors historical colonialism: extraction of value (data, compute, and talent) from peripheries to cores, with technological dependency enforcing political alignment.
JD Vance’s 2025 remarks in Europe underscored this: an “American AI order, or nothing else.” This aligns with figures associated with the “dark enlightenment,” including Peter Thiel, who frame AGI as a quasi-theological quest — a “God made by Humans.” The pursuit prioritizes frontier scaling and speculative breakthroughs over grounded, deployable applications.
And yet, this gives rise to the “paradox of AI with American characteristics”: massive capital inflows into hyperscale data centers and models yield limited measurable ROI (MIT surveys suggest 95% of businesses see zero return), while driving entropy through energy consumption, infrastructure strain, and speculative bubbles.
AI is “embodied electricity.” US data centre demand is surging, but the grid faces severe constraints. Secretary of Energy Chris Wright’s call for 100 GW of new firm capacity in five years is a “mirage,” given permitting delays (7–10+ years for transmission), transformer shortages (lead times of 2–5 years, with China dominating supply), skilled labor gaps, and regulatory hurdles.
I have previously detailed how these create Dutch Disease effects: AI competes for electricity, water, and equipment, inflating costs and hollowing out other industries. Wholesale prices have risen sharply; lower-income households and communities bear the brunt.
China’s advantages compound this asymmetry. Industrial electricity in key regions costs ~$0.04/kWh vs. ~$0.12/kWh in the US. Combined with hardware-software co-optimization (e.g., Huawei’s Ascend and CANN stack, open-sourced elements), China achieves lower cost-per-token and faster deployment. Analogue and neuromorphic chips further enable industrial-scale edge AI at fractions of the TCO.
The US stack relies on closed ecosystems (NVIDIA CUDA dominance), creating vendor lock-in. This high-cost, high-risk model exposes users to supply chain vulnerabilities, backdoors, and extraterritorial controls (e.g., sanctions, data access via US firms). For Global South nations, adopting this means ceding sovereignty over data, algorithms, and critical infrastructure — in other words, digital neo-colonialism.
Philosophically, the AGI quest reflects a hubristic drive for transcendence, detached from human-centered governance. Xi’s speech counters this: AI must remain “under human control,” with risks managed through laws, monitoring, and multilateralism, opposing the over-extension of national security pretexts.
The Dark Enlightenment Philosophy and Its Risks in the American AI Project
A deeper ideological current underpins the American AI order, as I have previously argued; namely, the influence of “Dark Enlightenment” thought among segments of the techno-elite. Associated with figures like Curtis Yarvin (Mencius Moldbug), and resonating in networks linked to Peter Thiel, JD Vance, and entities such as Palantir, this philosophy rejects egalitarian democracy in favor of hierarchical, accelerationist, and often anti-egalitarian visions of governance. Technology — particularly AGI — is framed not merely as a tool for prosperity but as an instrument for civilizational dominance, elite sovereignty, and a reconfiguration of power away from public accountability toward concentrated corporate-techno authority.
In this worldview, AI represents a mechanism to “win” existential contests, where a select vanguard directs societal outcomes, surveillance enhances state/corporate power, and democratic constraints are seen as obstacles to progress. Palantir’s emphasis on AI for defense, intelligence, and “winning” aligns with these currents, as do broader discourses skeptical of mass politics in favor of sovereign, CEO-like governance structures.
I have previously highlighted how this manifests in coordinated influence campaigns — super PACs tied to OpenAI, Palantir, and Andreessen Horowitz funding influencers to promote “American AI” supremacy while stoking fears of Chinese data risks and job losses. Such efforts sustain the narrative of an existential race, justifying subsidies, lax regulation, and procurement while deflecting from domestic bottlenecks.
Risks of This Philosophical Turn include:
1. Centralized Power and Accountability Erosion: Dark Enlightenment-adjacent visions prioritize elite direction over distributed sovereignty. This risks entrenching techno-oligarchic control, where AI systems serve narrow interests rather than broad human empowerment. In contrast to Xi’s insistence on AI remaining “under human control” and governed through multilateral consensus, this approach invites unaccountable decision-making by algorithms shaped by a narrow ideological cadre. The weaponization of data and platforms (echoing the US “Underground Empire” of fiber optics, SWIFT, and surveillance as documented by Newman and Farrell) extends this domestically and internationally, raising profound risks to civil liberties and pluralism.
2. Accelerationism and Existential Hazards: The pursuit of AGI as a near-theological imperative — “a God made by Humans” — carries unexamined risks of loss-of-control scenarios. While Xi’s speech stresses risk-awareness, early-warning systems, and opposition to overstretching national security pretexts that could justify unilateralism, Dark Enlightenment accelerationism downplays safeguards in favor of rapid dominance. I have also noted the thermodynamic and systemic entropy this generates: high-burn-rate models with questionable ROI, energy profligacy, and speculative bubbles that prioritize frontier hype over viable, deployable intelligence.
3. Social and Global Fragmentation: By framing competition in zero-sum, civilizational terms, this ideology exacerbates divides. Influence operations amplify “Team USA” patriotism while ignoring how US export controls have backfired — accelerating China’s self-reliant stack (e.g., NVIDIA’s reported zero market share in China’s AI accelerators, alongside DeepSeek and Qwen advances).
For the Global South, it offers a stark binary: subordinate to an exclusionary order or embrace more distributed alternatives. Equity concerns compound this; US data center siting has reportedly shifted burdens toward marginalized communities, echoing broader distributional failures.
4. Economic and Systemic Instability: High valuations (e.g., OpenAI’s pursuit of $850B+ amid massive projected losses) rest on monopoly narratives vulnerable to cheaper, open-weight Chinese models achieving practical parity at 5–10% of the cost. A correction could ripple through markets propped up by AI hype, affecting pensions and broader confidence. We can contrast this “hare” sprint — reliant on scarce premium hardware and energy — with China’s “tortoise” emphasis on efficiency, co-optimization, and accessibility.
China’s Alternative: Open-Source, People-Centered, and Sovereign AI
Xi’s address framed AI as a “new engine of world economic growth” and accelerator for shared prosperity. Key principles: openness and win-win innovation; security and controllability; inclusiveness and civilizational mutual learning; solidarity in global governance. China promotes open-source to democratize access, contrasting US proprietary barriers.
Beyond rhetoric, announcements include training programs, regional cooperation centers, and MAZU deployment. WAICO institutionalizes this. China positions itself “with” and “of” the Global South, bridging digital divides rather than widening them.
I have previously argued that China methodically builds systemic advantages: energy abundance, full-stack ecosystems (chips to applications), and rapid commercialization over pure scale. It is these systemic “whole stack” advantages that make it possible for American AI’s techno-colonial ambitions to be challenged.
Huawei’s τ (Tau) Scaling Law exemplifies the shift — optimizing for compute/energy/time rather than transistor shrinkage alone. This yields better power-performance-area (PPA) metrics, aligning with thermodynamic realities. This links to thermoeconomics: viable information maximizes negentropy (order and utility) relative to energetic costs.
China’s stack produces more viable information; US efforts risk unviable noise and entropy. Low-cost Chinese models (e.g., DeepSeek, Alibaba) are gaining traction globally, including in India and even the US, due to cost advantages amid transformer shortages and energy constraints. Yicai Global and SCMP reports highlight China’s pivot to full-system ecosystems and commercial deployment.
This can be understood as the reassertion of nation-state sovereignty in digital realms: data localization, open-source interoperability, collective truths over fragmented “truth tribes,” distributed ledgers for trust, and data ecologies as public goods. Huawei’s HarmonyOS and ecosystem contributions (e.g., Linux kernel, RISC-V) enable this multipolar architecture, countering US “Underground Empire” surveillance and weaponization (SWIFT precedents extended to tech). I have called this Digital Westphalia. Far from being isolationist, this speaks of a future of sovereign interoperability, which prevents hegemony while allowing collaboration on common standards via multilateral forums like the UN.
China’s approach to AI should not be misunderstood as a narrowly technocratic or state-driven response to geopolitical pressures. Rather, it is a response that is deeply rooted in enduring philosophical traditions, particularly Daoism (Taoism), which emphasizes harmony with the natural order (Dao), effortless action (wu wei), adaptability, simplicity, and the dynamic interplay of opposites (yin-yang).
This cultural-philosophical grounding contrasts sharply with Western framings often dominated by dualistic tensions — mind versus body, human versus machine, control versus chaos — and a Promethean quest for transcendence through AGI.
In Daoist thought, as articulated in the Dao De Jing and Zhuangzi, the Dao is an immanent, flowing principle underlying the universe. Wisdom lies not in imposing rigid order or conquering nature but in aligning with its patterns — observing, adapting, and allowing emergent processes to unfold. Applied to AI, this manifests as a preference for systems that integrate seamlessly with human society, physical infrastructure, and ecological realities rather than seeking to supplant or dominate them. Xi’s repeated calls for “people-centered” AI, “human control,” and AI that “empowers humanity writ large” echo this: technology as a harmonious extension of human endeavor, not a rival intelligence aspiring to godhood.
China’s shift “from size to speed” (e.g., Huawei’s τ Scaling Law), prioritizing compute/energy/time efficiency and full-stack co-optimization over raw scale points to a pragmatic emphasis that incorporates harmony and grounded deployment at the very heart of AI. This reflects Daoist pragmatism — wu wei as strategic non-forcing: working with material constraints (energy abundance, mature nodes, analogue chips) rather than against them through heroic over-engineering. Chinese models like DeepSeek and Qwen achieve practical parity or superiority in cost-effectiveness and deployability, enabling rapid commercialization and industrial integration (“AI+” initiative). This is “grounded AI”: embedded in the physical world, supporting intelligent manufacturing, agriculture, and public services without the entropic excesses of hyperscale Western data centers.
Daoist harmony (he) further informs the emphasis on inclusivity and civilizational mutual learning in Xi’s speech. AI should not erode cultural diversity but enrich the “garden of civilizations.” This stands in opposition to Western tendencies toward universalist imposition or apocalyptic dualisms (e.g., singularity fears or accelerationist hierarchies). Where some Dark Enlightenment influences in US circles envision AI as a tool for elite-directed civilizational overhaul or existential dominance, the Daoist-inflected Chinese path seeks balance: humans and machines in mutual reinforcement, yin (contemplative, ethical governance) and yang (innovative deployment) in dynamic equilibrium.
By way of contrast, Western AI discourse often inherits Judeo-Christian and Enlightenment legacies — creation ex nihilo, mastery over nature, and linear progress toward transcendence. The AGI pursuit embodies a Faustian bargain: humanity playing God, risking hubris and loss of control. Philosophical critiques (e.g., Hubert Dreyfus’ famous arguments about embodiment as the sine qua non of human intelligence in the world) expose the limits of disembodied computation, yet the cultural momentum persists.
Daoism, by contrast, views intelligence as emergent from relational flows, not isolated symbolic manipulation. AI becomes a companion in the ecosystem — adaptable, decentralized, and attuned to context — rather than an autonomous overlord.
This philosophical divergence yields tangible advantages. China’s stack produces what I call more viable information when put into thermoeconomic terms: lower energetic costs relative to systemic negentropy (order, utility and coordination). Open-source dissemination aligns with Daoist simplicity and sharing, democratizing access for the Global South and countering techno-feudal rents. It enables sovereignty-compatible development — nations adapting AI to local Dao — rather than lock-in to proprietary, high-cost ecosystems.
Critics may view state guidance as antithetical to Daoist spontaneity, yet in practice, China’s “effective market + well-functioning government” model creates conditions for emergent innovation while steering toward harmonious outcomes (e.g., safety, inclusivity and ecological integration being key design components). Xi’s vision of AI as a “symphony of international cooperation” rather than solo performance resonates with Daoist orchestration of collective flows.
Xi’s vision counters the Dark Enlightenment risks by advocating inclusiveness, civilizational mutual learning, and AI as a “common asset” of humanity. It rejects monopolies and promotes capacity-building that respects diversity rather than imposing a singular ideological or technological hegemony. This can be framed as a broader contest of visions for humanity’s future, and not be reduced to a crude nation-state rivalry. The contrast being drawn is between concentrated elite visions and sovereignty-compatible, distributed development. Digital Westphalia — open-source interoperability, data sovereignty, and public-good ecologies — offers a stabilizing alternative that mitigates the dark underbelly of unaccountable accelerationism.
This philosophical divergence amplifies material contrasts. Where Dark Enlightenment thinking fuels exclusionary races, China’s grounded, people-centered approach — evident in practical announcements at WAIC — prioritizes shared tools for human flourishing. The Global South, and indeed ordinary citizens in the US facing rising energy costs and inequality, stand to gain from the latter path. Ignoring these ideological risks risks not only technological lock-in but a deeper erosion of pluralistic, accountable futures.
Daoist roots provide China with a resilient philosophical operating system for AI: pragmatic, harmonious and human-grounded. This complements material strengths (energy, co-optimization and open architectures) and positions China’s alternative as not only competitive but civilizational — offering the Global South tools for endogenous modernization rather than imported transcendence or dependency.This grounded path merits study in the West, where philosophical overreach risks compounding thermodynamic and social entropy. The choice is between flowing with reality or forcing a singular, fragile vision upon it.
Global South Choice: Empowerment vs. Lock-In
For developing nations, the calculus is stark. US models promise cutting-edge capability but deliver dependency, high costs, and vulnerability to sanctions or policy shifts (e.g., NSF restrictions on Chinese collaborations are a recent case in point). Chinese offerings provide accessible tools for local adaptation: cheaper inference/training, edge computing for agriculture/industry, and governance frameworks respecting cultural diversity.
Xi’s emphasis on “AI for all” and avoiding “new historical injustice” directly addresses colonial legacies. Chinese advantages in analogue chips, energy, and co-optimization enable broad industrial transformation — thousands of nodes per factory at low TCO — versus constrained US deployments.
In the US and the West, AI runs the risk of exacerbating ever-widening inequalities. Data centers strain grids, raising costs for ordinary households while benefits accrue to tech elites. Global South adoption of the US stack risks similar hollowing-out without sovereignty gains. China’s model supports “intelligent manufacturing” and modernization suited to local contexts.
Toward a Just Digital Order
Xi advocates shaping AI values with “humanity’s common values” while preserving civilizational diversity. This contrasts with Western AGI pursuits tinged with exceptionalism and transcendence fantasies. This pursuit is as dangerous as it is philosophically flawed and entropically costly. A grounded, human-in-the-loop approach — prioritizing safety, ethics, and applicability — better serves humanity writ large.
“Techno-feudalism” globalizes into techno-colonialism via rents and expropriation. Digital Westphalia restores balance: states as primary actors, technology as enabler rather than dominator.
Xi’s Shanghai speech and initiatives signal China’s commitment to a multipolar AI future. By contrasting open, sovereign, people-centered development with proprietary lock-in, it offers the Global South agency. The transition from techno-colonialism to Digital Westphalia is underway. It promises reduced rents, enhanced sovereignty, distributed empowerment and broader prosperity — if nations and communities in them seize the opportunity. For ordinary people in the US and the West more broadly, escaping high-cost, entropic AI dynamics also holds appeal. Humanity’s collective wisdom, as Xi noted, should guide AI toward shared flourishing, and not be the seeds of division.
Before us are critical choices. Rosa Luxemburg once pitched the question of the times as one between socialism or barbarism; for her, the barbarism of the 1920s was the rising tide of Fascism. A century on, we again confront the risks and threats of our own contemporary throw-back of barbarism; techno-colonialism or a future of Digital Westphalia in which collaborative enablement gets the better of hegemonic dependence. The latter aligns with an equitable, sustainable digital order; the former risks repeating history’s injustices in technological garb.
About the Author
Warwick Powell is an adjunct professor at the Queensland University of Technology, and a Guancha columnist.
Absolutely marvellous article about an absolutely exceptional speech. To me, the entire world—Europe included—should jump on the opportunity to 'follow the way' into the future.