DOSSIER

TECH

How big tech companies learned to stop worrying and love the bombs

Until recently, many big tech companies opposed the militarization of AI, but that now seems like ancient history, as they are forging partnerships with arms companies. The prospect of generous Pentagon funding for AI is too tempting to refuse.

In any list of "known unknowns" that the world will face in 2026, artificial intelligence will certainly be among the top ones. Are predictions of widespread AI adoption, which will replace hundreds of millions of workers, about to come true? Will the AI ​​bubble burst? Will the United States or China win the race for "general artificial intelligence"?

Nick Srnicek's book, Silicon Empires, doesn't directly answer any of these questions, but, as the author states, it "offers a map of the terrain we must fight on." By carefully mapping the development of AI within its proper economic and geopolitical context, and encompassing analyses from both the US and China, Srnicek's AI guide can help us maintain a realistic, long-term perspective on the technology's likely trajectory.

Beyond bubbles and chatbots...It's no longer a fringe idea to say that there's an AI bubble, as this has been acknowledged even by important figures in the industry, such as Jeff Bezos and Bill Gates. OpenAI CEO Sam Altman seems to already be preparing his company for a state bailout. One estimate of the AI ​​bubble suggests it is seventeen times larger than the dot-com bubble and four times larger than the subprime real estate bubble that triggered the 2008 financial crisis. A crisis is clearly brewing.

It's no longer a fringe idea to say that there's an AI bubble, as this has been acknowledged even by important figures in the industry, such as Jeff Bezos and Bill Gates.

Srnicek's sober analysis encourages us to look beyond the bubble. The fact that AI is going through a difficult initial period is neither new nor surprising: the history of technological advancements is marked by struggles and difficulties before success. Furthermore, it is highly unlikely that any crisis will bring down the large technology companies, which are primarily responsible for AI development due to the strength of their market position and their intrinsic importance to the global digital infrastructure.

As Srnicek states: If an AI winter sets in, it is unlikely to be long-lasting. The technology's potential remains too high, and the importance of first-mover advantages is too great, for large technology companies to voluntarily relinquish control over the direction of AI development...thinking in terms of bubbles severely limits the view of AI's impact.

Renewed questions have arisen about the true potential of AI, with skeptics pointing to the slowdown in progress in the latest version of OpenAI's ChatGPT as a case study of the limitations of the "scaling" model that has brought generative AI to this point. For Srnicek, focusing on chatbots like ChatGPT is looking in the wrong direction. Investors place their hopes on the potential of industry-specific AI "agents" that can go far beyond simply answering a question and actually execute actions to achieve a goal—automating workflows across the economy. "Chatbots are an inadequate guide to where AI is heading, and both critics and opponents should ensure they have the right target in mind," he argues.

What is perhaps missing from Srnicek's analysis is an exploration of the macroeconomic conditions under which AI agents could be adopted across the economy. Economist Michael Roberts has convincingly argued that a mountain of "zombie" capitalist companies, kept afloat by cheap credit since 2008, is incapable of investing heavily in AI. The global economy would have to undergo a seismic process of "creative destruction" to forge the space in which new actors willing to fully embrace AI agents can emerge. The development of AI is ultimately conditioned by the dynamics of the capitalist political economy.

The AI ​​Strategies of Big Tech Companies...Srnicek's 2016 book, "Platform Capitalism," stood out by conceptualizing the breadth of digital platform business models that have begun to dominate nearly every sector, from "lean" platforms that outsource everything except core software, like Uber, to "industrial" platforms like Siemens, a company that builds digital hardware and software infrastructure in industry. Similarly, a major strength of "Silicon Empires" is the clarity with which it explains the different strategies that big tech companies are adopting in the field of AI. The differences in approach are significant and may ultimately determine which companies will win the race to dominate AI.

Artificial intelligence (AI), like the steam engine and electricity, is a general-purpose technology (GPT). All GPTs are characterized by their applicability across the economy, requiring widespread dissemination to develop. Typically, the value of technological advancements is captured later, when they are transformed into industry-specific products.

This is why states have historically been fundamental to R&D, as they can afford to invest in advances in global technologies without aiming for profit. This was the case with the internet and semiconductors. In the case of AI, large technology companies lead innovation, but they need to operate with business models that aim for profit. The attempt to reconcile these two aspects has led to the emergence of four strategies. First, the infrastructure strategy seeks to dominate the foundations of the AI ​​economy, upon which other companies can build. Amazon and Microsoft are key players in this scenario, consolidating their oligopolistic positions in cloud computing markets. For these companies, the huge investments in data centers represent an investment in the future growth of AI, as they prepare to receive cloud rents from industry-specific products that will depend on their infrastructure to operate.

For those who benefit from the infrastructure strategy, the more widespread AI is, the better. Microsoft CEO Satya Nadella praised the chatbot from the Chinese company DeepSeek, which has capabilities similar to ChatGPT but at a much lower cost, as a major step towards "ubiquitous" AI. Microsoft partnered with a US education nonprofit, offering free use of the chatbot to teachers "with the goal of integrating the American education system with Microsoft servers."

The second strategy is to lead at the frontiers of AI innovation. OpenAI, Anthropic, and DeepSeek are developers of cutting-edge AI models. For those adopting a pioneering strategy, staying one step ahead of the competition is essential to capturing value, as this innovation advantage is the only thing that can place the company's intellectual property at the center of a broader development ecosystem.

Everything and anything...The challenge faced by leading companies is that the costs of innovation are enormous due to the amount of computing power needed to drive AI innovation. Meanwhile, the task of commercializing these technological advances is fraught with difficulties, and when greater emphasis is placed on commercial implementation, research can be hampered.

Leading companies are betting on artificial general intelligence (AGI), the Holy Grail of AI that reporter Karen Hao discovered to be a flimsy excuse for OpenAI CEO Sam Altman to dismiss all criticism of his company's business practices. For Srnicek, we should simply understand AGI as an AI model that can be “applied across all sectors.” This would eliminate at once the difficulties that leading AI companies face in capturing value from their innovations due to the need for sector-specific tools. Srnicek describes the potential of AGI as “immense,” but it is important that we remain skeptical about its viability.

The conglomerate strategy, the third of its kind, represents an attempt to build sector-specific AI products, aiming to dominate the market in the same way as conglomerates of the past: through ownership and acquisitions. Google is at the forefront of this strategy, having developed as many fundamental AI models as its next three largest competitors (OpenAI, Microsoft, and Meta) combined.

Google's quest for AI dominance requires the company to possess capabilities across the entire AI value chain: positioned at the forefront of research, with a solid infrastructure base, and capable of building high-quality products for diverse sectors. The company's launch of a series of AI tools for the healthcare sector in recent years, from personal health to drug development, exemplifies how this strategy is being applied in practice. In China, Huawei is at the forefront of a group of large technology companies that are adopting this comprehensive approach to AI development.

Finally, there is the open strategy, with Meta in the United States and Alibaba and DeepSeek in China as the main implementers. As the name suggests, the open strategy involves making AI models available so that other developers can improve them. In the case of Meta's "Llama" models, this does not meet the open-source standard, as there is still a significant lack of transparency in the training data and the algorithms behind the models. Even so, the weights used in the modeling are publicly available, which facilitates access and modification of the models by others.

What advantage does Meta gain from the open strategy? Other large technology companies are building protective walls around their intellectual property, creating an exclusive zone of interaction with selected partners. Meta, on the other hand, manages to build a broad ecosystem around its intellectual property, which naturally attracts researchers and developers. These, in turn, will make their own improvements and discoveries, which "can then be easily incorporated into Meta's internal systems." This is a strategy that can significantly reduce Mark Zuckerberg's company's costs in the long term.

The Rise of the “Technological-Industrial Complex”...In his farewell address in January 2025, Joe Biden warned of the risks of a growing “technological-industrial complex” in the United States. This consciously echoed the words of Dwight Eisenhower as he left the White House in 1961, when he memorably expressed his fears about a “military-industrial complex” that could dominate American democracy.

Like the military-industrial complex, the technological-industrial complex combines powerful vested interests within the state, primarily the War Department, with the largest private market players, which today are the big tech companies. This is a class alliance that has consolidated very recently. As Srnicek points out, Google, Meta, OpenAI, and Anthropic opposed the use of AI tools for military purposes in early 2024. All of these companies changed their positions in less than a year, and some quickly formed partnerships with companies in the defense sector.

The drastic shift in posture is partly due to economic necessity. AI development is expensive, and the military offers the prospect of substantial, long-term funding. But the geopolitical turn has deeper roots. There has been a notable ideological shift among the tech elites in the US, moving away from what Srnicek calls the "Silicon Valley Consensus" toward "technonationalism."

The Silicon Valley Consensus was essentially a compromise among tech elites with US-led neoliberal globalization. Politicians and CEOs of tech companies shared a belief in technology's ability to create a world of borderless commerce and data, led by the United States. The lax regulation of the tech sector meant that Silicon Valley had little reason to worry about state interference. Abroad, Washington helped keep foreign economies open to American technology and limited the imposition of foreign taxes and regulations on large US technology companies, while value chains between all major technology companies stretched from China to the United States, keeping costs low.

What ended the Silicon Valley Consensus was the rise of China, which opened a new constellation of class and interest conflicts. Chinese tech giants began to become real competitors to their American rivals, shifting Silicon Valley's calculations. Meanwhile, since at least Donald Trump's first term, the state has prioritized American technological dominance at the expense of global interconnectedness. This continued during Biden's presidency, with the tightening of sanctions on critical technologies like semiconductors, and under Trump's second term, flourished in what Srnicek calls a "technonationalist vision of American supremacy and unrestricted innovation."

The level of integration between large technology companies and the state is now undeniable. A $9 billion Pentagon contract for a “joint cloud computing capability for military purposes” includes all the major US cloud players: Amazon, Google, Microsoft, and Oracle. Ties between tech companies and the military have rapidly increased. Srnicek notes that it is no coincidence that the emergence of the tech-industrial complex has coincided with the “war waged by big tech companies against their workers,” many of whom have sought to resist militarization.

The rise of technonationalism in the United States has found a parallel in China. Just as in the US, the elites of the Chinese Communist Party initially adopted a non-interventionist stance toward the emergence of large and powerful digital platforms in China, seeking to encourage the sector's growth. However, with increasing tensions with the United States, Chinese President Xi Jinping has begun to steer tech companies toward state priorities. This involved cracking down on many companies focused on facilitating consumption, such as the sharing economy platforms Meituan and DiDi, while simultaneously pressuring technology companies to contribute to industrial development, as this is the raison d'être of the Chinese Communist Party government.

Thus, in both the US and China, we observe the emergence of a potential new hegemonic order “due to the dismantling of class coalitions between the economic interests of the State, the security interests of the State, and the interests of platform capitalism.” Srnicek is cautious about the prospects for consolidating this new order, highlighting the opposing trends that move away from militarized technonationalism and the relative independence of large technology companies from the State. But the era of neoliberal globalization has clearly come to an end, and the consequent fusion between the State and large technology companies around a nationalist vision for AI entails extreme dangers for everyone.

What's next?...In a contest between the United States and China for supremacy in artificial intelligence, which country is more likely to emerge victorious? Srnicek's analysis leans toward the idea that China, despite exhibiting many weaknesses compared to the US, may very well win the technological race.

The reasoning is surprisingly simple: while the US tech industry focuses on innovation, China's priority is adoption, and adoption is likely to be decisive in the long run due to the need for a general-purpose technology like AI to spread throughout the economy to reach its full potential: In previous industrial revolutions, major power transitions led to these transitions not because one country monopolized the profits, but rather because one country excelled at adopting a new technology and used it to dramatically transform its entire economy in terms of productivity and growth. This widespread transformation of the entire economy—and not of a single leading sector—is what allows emerging major powers to overtake and surpass established hegemonies.

Regardless of the outcome of this dispute, it is unlikely that technology will decisively divide into two hemispheres, East and West, due to the complex interaction of international value chains. Instead, there will be an "overlap of different geopolitical [technological] layers," with the pursuit of a balance between American and Chinese power being a viable strategy for many countries, albeit challenging to implement.

Unlike Srnicek's 2015 book (co-authored with Alex Williams), *Inventing the Future*, which encouraged the left to embrace automation as part of a post-capitalist vision, *Silicon Empires* avoids developing left-wing policies for AI. Srnicek restricts himself to just two demands: no war between the United States and China, and large technology companies should not be allowed to dominate AI development.

These are useful starting points to guide the left regarding AI, but ultimately, a more ambitious agenda will be needed. Any self-respecting contemporary socialist program must be able to explain what role AI should play in the economy and society, how it should be governed, and what its relationship should be with the state and between states. Regardless of what happens in 2026 with the AI ​​bubble, the political challenges posed by this powerful technology will only increase over time.

Authors: Ben Wray is the co-author, along with Neil Davidson and James Foley, of Scotland After Britain: The Two Souls of Scottish Independence (Verso Books, 2022).