The AnewZ Opinion section provides a platform for independent voices to share expert perspectives on global and regional issues. The views expressed are solely those of the authors and do not represent the official position of AnewZ
The U.S. State Department has drafted a letter urging 35 countries to choose sides in the global AI race, but the rapid spread of open models highlights the limits of efforts to divide an increasingly interconnected technology ecosystem.
According to the draft reviewed by Reuters, countries participating in Washington’s Pax Silica framework cannot also join rival initiatives with conflicting requirements - and may face exclusion from U.S.-led collaboration if they try to do both.
But while diplomats debate the proposed ultimatum, a different contest is playing out among developers. Across development platforms, academic institutions and private companies, engineers and researchers are making their own choices based not on geopolitical instructions, but on practical utility and technical merit.
That raises a fundamental question: can diplomatic pressure force technology into rigid camps, or has innovation reached a point where it can continue to flow across borders despite political efforts to contain it?
The contrast between diplomatic manoeuvring and market reality is increasingly stark. Recent Hugging Face data show that Qwen-based models account for 151,448 derivatives on the platform - 2.6 times the total footprint of Meta models. Repositories with declared parameter counts recorded more than two billion Qwen downloads between January and July 2026.
Those figures reflect not government mandates, but the accumulated choices of developers responding to capability, cost and accessibility.
OpenRouter data tell a similar story. The platform’s own analysis reported that Chinese-origin models had overtaken U.S.-origin models in token share by early June 2026. The figures measure traffic on one major model-routing platform rather than worldwide AI demand, but they capture actual usage rather than political alignment.
That does not mean U.S. AI companies lack sophistication or reach. The United States attracted about 83 per cent of global private AI investment in 2025 and retains a commanding position in hyperscale computing infrastructure. Yet on open development platforms, a growing number of developers are voting with their downloads and computational cycles.
The effectiveness of exclusionary diplomatic pressure has always depended on a critical variable: how essential the excluded party becomes. In previous eras of technological competition, this leverage was substantial. Access to proprietary systems, specialised hardware or exclusive partnerships could meaningfully constrain development elsewhere.
AI presents a different landscape. Capable open-weight models can now run on consumer-grade devices after quantisation, while smaller systems can be adapted for local use without constant access to a foreign cloud service.
Once model weights are released under permissive licences and downloaded widely, recalling or suppressing them globally becomes extremely difficult. The idea that advanced AI always requires choosing a geopolitical side becomes harder to sustain when useful models can run locally.
Open licensing is central to this shift. Leading Chinese releases such as Alibaba’s Qwen models and DeepSeek’s systems have used Apache 2.0 or MIT licences, allowing broad modification and commercial deployment.
The distinction is not absolute: U.S. companies also publish open-weight models. However, many leading American systems remain closed services or operate under custom restrictions, giving developers an incentive to gravitate towards ecosystems offering greater control.
Commercial infrastructure also crosses political boundaries. Nvidia publishes its own open models while providing optimisation tools for Chinese-origin systems such as DeepSeek. AMD’s software ecosystem supports inference for models including Qwen and DeepSeek.
Exclusive coalitions may appear clearly defined in diplomatic documents, but the technology industry does not map neatly onto those divisions.
History offers cautionary examples of the unintended costs of supply-chain fragmentation. The Information Technology and Innovation Foundation estimates that U.S. restrictions on doing business with Huawei cost American companies at least $33 billion in sales between 2021 and 2024.
The estimate does not demonstrate that export controls are always ineffective, but it does underline that exclusion can carry significant costs for the countries imposing it.
The potential cost of deeper semiconductor fragmentation is greater still. A Boston Consulting Group and Semiconductor Industry Association study modelled a hypothetical world of parallel, fully self-sufficient regional chip supply chains. It estimated that such duplication could raise semiconductor prices by 35 to 65 per cent, with those costs ultimately passed on to businesses and consumers.
For middle powers and developing economies, fragmentation carries particular consequences. When countries are pressured to choose exclusive blocs, they risk losing the option to build domestic AI capacity using the best tools and models from multiple sources.
Diverse access matters most when countries are trying to develop local expertise rather than remain dependent on a single foreign system.
Increasingly, part of the answer lies not in diplomatic offices but with developers making technical decisions every day. Every download, fine-tuning run and adaptation of an open model is a vote for a particular combination of accessibility, cost and performance. Those choices cannot easily be reversed by diplomatic ultimatums.
Once developers have established that capable AI systems are available without exclusive geopolitical alignment, demands for binary loyalty risk narrowing a policy space that countries may seek to preserve for themselves.
This matters particularly for countries positioned between great powers. Exclusion from one diplomatic framework does not necessarily mean complete technological isolation. A country can build genuine AI competence by using open models, training local talent and contributing to global development communities.
But openness does not eliminate every dependency. Access to advanced chips, capital, cloud infrastructure and specialised expertise remains critical.
Diplomatic letters can draw political lines. They cannot easily sever an ecosystem that is already deeply interconnected. When model weights move across borders, permissive licences allow modification and reuse, and developers retain agency in their technical choices, rigid bloc boundaries become harder to enforce.
The vision of AI divided into exclusive camps assumes that technology flows only through officially designated channels. In practice, technology rarely remains neatly contained when capable and accessible alternatives spread widely.
The architecture of the AI era is being shaped not only by decisions in capital cities, but by the cumulative choices of engineers, researchers and companies building applications, solving problems and selecting tools according to what works.
That does not make geopolitical competition irrelevant. Talent retention, investment patterns, security, standards and access to advanced hardware remain genuinely important. But it suggests that the eventual shape of global AI may be determined less by diplomatic pressure alone than by the economics of openness, accessibility and technical merit.
For Pakistan and similar nations, the lesson is straightforward. The era in which joining an exclusive bloc was the main route to technological access is giving way to one in which genuine capability increasingly depends on diverse tools, local talent, infrastructure and participation in global technical communities.
Diplomatic pressure from any direction still matters, but so does understanding - and preparing for - this structural shift.
Many of the most consequential AI choices in 2026 are not being made in government offices. They are being made by developers each time they download a model, modify its weights and deploy it to solve a real problem.
That distributed decision-making may ultimately prove more consequential to the shape of the global AI ecosystem than any draft letter addressed to 35 governments.
The writer is a technology policy analyst focused on global AI governance and digital infrastructure development.
U.S. efforts to squeeze Iran’s economy through an oil blockade and sanctions are becoming increasingly difficult for Tehran to withstand, acc...
U.S. negotiators Steve Witkoff and Jared Kushner will visit Russia then Ukraine over the weekend, Russian state news agency TASS has reported, citi...
The 83rd Venice Film Festival has opened on the Lido, bringing Hollywood stars and leading filmmakers together for a programme dominated by war, po...
Russian President Vladimir Putin has said that he thinks there is a chance of a peace being reached with Ukraine. He told the Eastern Economic Forum in Vladivostok, Russia that he believed there was "a chance of finding a solution," to the conflict.
The Iranian Health Ministry said at least 18 people were killed and 142 others injured in U.S. airstrikes carried out between 30 August and 2 September.
U.S. negotiators Steve Witkoff and Jared Kushner will visit Russia then Ukraine over the weekend, Russian state news agency TASS has reported, citing an unnamed source. Ukrainian President Volodymyr Zelenskyy has also said American envoys will visit the capitals of both countries.
Start your day informed with the AnewZ Morning Brief. Here are the top stories for the 3rd of September, covering the latest developments.
The Iranian Army has again struck at U.S. bases in Kuwait and the United Arab Emirates in retaliation for U.S. attacks on southern and western parts of Iran late on Tuesday, which reportedly killed 18 people and injured more than 100 others.
“Once Upon a Time in the Middle East” has drawn large audiences in Chinese cinemas, and for good reason.
When the South Korean container ship PanStar Acro left Busan for Europe through the Arctic in August, its voyage seemed far removed from the war between the United States, Israel and Iran. In reality, the two developments are increasingly connected.
On 1 September 2026, the Non-Aligned Movement marked the 65th anniversary of its first conference. Belgrade once again hosted a high-level commemorative gathering at the movement's birthplace.
Something unusual is happening across societies otherwise separated by geography, political culture, economic development and historical experience.
In late April, Mali faced one of its most serious coordinated assaults in years.
You can download the AnewZ application from Play Store and the App Store.
What is your opinion on this topic?
Leave the first comment