The AI 'arms race' fallacy
The future of globalization in the age of AI
Marc Andreessen calls DeepSeek AI’s Sputnik moment.
Yes, it did catch the US unprepared but that’s where the ‘space race’ analogy ends.
Today's AI race is not merely an 'arms race' nor is DeepSeek easily explained away as just a Sputnik moment.
This race is playing out against the larger backdrop of more than a decade of technology infrastructure export by the largest economies around the world. - whether it is India's export of digital public infrastructure, cloud export by the US BigTech, or China's Digital Silk Road working alongside its Belt and Road project.
And that’s what makes this truly interesting!
This combination of tech infrastructure exports combined with leverage through complementary AI capabilities creates a new format of globalization - standards-based globalization - something that most people don’t yet fully understand.
This post goes into three parts:
Chapter 1: The USA
Chapter 2: China
Chapter 3: Europe
But first, let’s start at the very beginning - roughly 10 days back.
Chapter 1: The USA
It’s been a heady few weeks.
First, we had the U.S. government’s Stargate announcement: $500 billion for what’s likely the largest AI infrastructure project in history.
This was likely Peak Altman. A supposedly half-trillion-dollar moonshot to secure America’s dominance in the AI ‘arms race’.
It might be Peak Altman, but it’s also Masa-as-usual, who’s still looking to undo his mistake of not investing in Facebook thinking it was overvalued.
Almost at the same time, DeepSeek appeared on the scene - no breathless media tour though - and stole the show.
While Sam Altman was busy promising that AI would cure cancer (and maybe solve world hunger and teenage boredom - two other wicked problems), DeepSeek quietly threw in a better, faster, more efficient way to train and run models.
And open-sourced it for good measure.
So OpenAI, which has always had an identity crisis of sorts with its openness and its profit-making, was now out-open-sourced by another player who ironically improved on its own models.
Sandwich Economics - What’s really happening
In my most recent post, I explained the concept of Sandwich Economics.
This primarily involved the creation of dominant positions over critical layers of the economy. In particular, the two layers that matter the most are
- market-facing aggregators, which control access to the end consumption at scale.
- cross-industry infrastructures, which control access to critical production capabilities at scale.
And, with that, they sandwich the rest of the economy in between - squeezing power and profits from them.
The most significant BigTech innovation is the “Sandwich” they create and impose on entire industries, forcing these industries to re-organize within it.
The U.S. is going all-in on a similar playbook for global dominance.
The Aggregator Layer:
The first row at the Trump inauguration was dominated by the heads of today’s leading Big Tech Aggregators - Meta, Amazon, Google, X etc. That was the Aggregator Layer right there.
The Infrastructure Layer:
Shortly after, Trump announced Stargate - the most ambitious AI infrastructure project - and the other part of the sandwich.
With Sandwich Economics, one ‘player’ specifies an economic framework and imposes it on the rest of the industry. Or in this case, the rest of the global economy.
The “sandwich” as an economic framework represents the logic by which all other players within a target economy create and capture value.
In other words, the ‘sandwicher’ - in this case, the USA - doesn’t just work on making itself more competitive, it also works on changing the rules of competition for everyone else.
The difference - is the power of competing with an economic framework (sandwiching) vs competing within it (getting sandwiched).
And with co-opting the leading aggregators while also announcing plans for the most ambitious AI infrastructure project, the US is all-in on sandwiching.
Last week, I posted the first in a series of videos explaining Sandwich Economics (also included again towards the end of this post).
Here’s the second video on Sandwich Economics explaining how sandwiching works and how economies get crushed in between these layers:
Chapter 2: China
Speaking of AI in terms of an ‘arms race’ misses the larger point. This is not a competition as much as a new form of globalization.
Over the past 15 years, we’ve seen massive digital infrastructure exports from countries - whether it is India's export of digital public infrastructure, cloud export by the US BigTech, or China's Digital Silk Road working alongside its Belt and Road project.
China has been playing this tech infrastructure export game in full swing since the days of Huawei and ZTE exporting their 5G infrastructure to run smart city infrastructure across cities along the Belt and Road - although where Huawei and ZTE started was more of component export. With AI, we see the power of infrastructure export.
My research at Brookings back in 2020 explained how China was leveraging a combination of AI and technology export to embed itself across the world:
China is aggressively exporting its digital infrastructure, playing a critical role in the development of technical standards, and developing unique points of control in the digital economy. Much like Google established itself as a dominant player in the smartphone ecosystem, China is attempting to do the same in an increasingly digital geopolitical landscape.
Understanding this is key to understanding the role AI plays in today’s geopolitical context.
Take China’s smart city strategy, for instance. Extracts from the Brookings report below:
Leading Chinese firms - Alibaba, Huawei, ZTE, and others - provide technology infrastructure for managing smart cities, including cloud-hosted services to integrate city management databases as well as central AI capabilities to manage city operations. Huawei smart city systems are active in more than 200 cities around the world. Alibaba provides a similar smart city management system and is currently piloting its “City Brain” project in Kuala Lumpur, Malaysia, where it uses traffic light information and traffic camera feeds, as well as data from the ride-hailing service Grab, to predict traffic patterns and reduce traffic congestion.
Smart city infrastructure needs 5G technology, and Chinese firms have been at the forefront of the 5G race. In countries where ZTE and Huawei deploy 5G networks, these companies will have greater leverage to promote their smart city infrastructure.
Along with China Mobile, these firms have increased their participation in international standard-setting bodies for 5G. By setting the standards and providing the infrastructure, China establishes important control points over the equipment, the technical services, and the shape of future technology across participating countries. Essentially, the full stack!
As providers of a smart city’s data operations, Huawei’s “Intelligent Operation Center” and Alibaba’s “City Brain” project gain visibility into patterns of citizen behavior and city infrastructure utilization. By aggregating data flows across these global smart city implementations—particularly those along the Belt and Road—Chinese companies like Huawei are well positioned to centralize smart city AI and create control points over cities along the Belt and Road. By creating the best trained and centralized AI brain for city operations, these companies could directly influence city-level governance and decision-making, including by controlling the profiling of citizens, citizen access to city services, and city infrastructure management.
The combination of smart city infrastructure exports, investment in 5G standards, and smart city management AI sets up China’s platform ambitions in urban infrastructure.
One key issue to consider is the shift from exporting components to exporting entire infrastructures.
In the past, companies like Huawei and ZTE focused on exporting 5G components, which made it relatively easy for other countries to counter - components can be swapped out and replaced.
However, what we’re seeing now is the export of complete infrastructures. This creates a much deeper level of dependency because once countries build their systems on top of these infrastructures, disentangling from them becomes far more complex.
If China hadn’t cracked down on its own big tech sector, this shift could have accelerated even faster.
And this creates the basis for standards-based globalization.
Standard-based globalization
When multiple countries build on the same underlying infrastructure, all exported from a single country, they also adopt the policies and standards embedded within that infrastructure.
As more nations import and layer their systems on top of it, they implicitly align with those standards.
Over time, the standard tied to the most widely adopted technology infrastructure becomes the foundation for globalization.
Moreover, the ability to manage complementary AI that makes this infrastructure more valuable but is controlled by the exporting country gives that country powerful control over these standards, effectively enforcing them with an iron grip.
This is the basis of standards-based globalization - a price that AI laggards will pay when the only option they have left is to build on the infrastructure exported by others.
Chapter 3: Europe
Europe seems like the un-cool uncle in all of this - refusing to participate in the race while asking others to stop running on its lawn.
But that’s misunderstanding what’s really at stake.
In a race, your worst outcome is only that you get left behind. If you’re an ‘uncle’ only happy when you set the unruly kids right, then that’s not a bad outcome.
But in a world of technology exports, standards-setting, and sandwich economics, your worst outcome is that you get sandwiched within someone else’s economic framework without even realising it.
DeepSeek will be one of many innovations that will force Europe and other AI laggards to confront an uncomfortable reality:
The future of globalization is being written in algorithms, in models, in protocols, not in treaties and regulations.
And in that future, Europe’s obsession with being the rule-maker, will - in all likelihood - just leave it stranded as the rule-taker.
Android, but for countries
Google used a three-pronged strategy. First, it drove the adoption of Android among smartphone manufacturers such as Samsung by open-sourcing the operating system. This was Google’s entry strategy.
Second, it controlled unique IP (intellectual property) in the form of Google Maps and the Google Play app store. This IP served as Google’s key differentiator. Every smartphone manufacturer needed to license this IP. Google continues to invest in improving its mapping data and growing its app store as these two sources of IP make Android more attractive as the standard.
Third, Google leveraged its neutral position in the smartphone industry to invest in standards development.
This three-pronged strategy – an adoption strategy with partners, control of unique IP, and the benefits of promoting standards – established Android as the dominant standard.
Countries that take this playbook will follow a similar strategy: open infrastructure, AI-based control points, and promotion of standards.
The ‘arms race’ fallacy
This is the ‘arms race’ fallacy.
Countries are not just locked in a race to own the best AI. They are simultaneously exporting digital infrastructure on which other countries build their economic and social activity.
It’s easy to frame the rise of AI as an ‘arms race’ - another chapter in global rivalry - East vs. West, Global North vs. Global South. But that’s missing the point.
The longer we cling to Cold War metaphors, the more we blind ourselves to what’s really happening. This isn’t a battle between superpowers for ideological supremacy - between democracy and authoritarianism - it’s a race between tech progressives and laggards, between those who export standards and those who adopt them.
The EU is falling behind the US, China, and India, which are also taking the lead in setting the standards that will shape the future.
The real divide is between the countries shaping AI and those who will merely consume it.
The real power struggle isn’t geopolitical - it’s architectural.
It’s about who gets to write the standards and control the infrastructure everyone else builds on. The ones in the lead will rewrite the rules of global competition and collaboration. The other will be left to adopt AI on someone else’s terms, inheriting its priorities and policies along the way.
That is the essential idea of sandwich economics.
And if you’d like to get into a deeper case study of how these companies set up a sandwich, there’s no better starting point than the case study of Reliance Jio in the video below:
We, the people, will make the standards. Kwaai.ai
Really enjoying this series. Its been super useful to start to give terminology to the shifts we can see around us.