Cold War 2.0

George Takach is a Canadian technology lawyer by profession — for decades a senior partner at McCarthy Tétrault handling cross-border technology deals — and that background shapes the vantage point of Cold War 2.0: Artificial Intelligence in the New Digital Battleground. The book, published in 2024, lands in a conversation already thick with contributions from political scientists, former defence officials and venture capitalists arguing that the United States and its allies have entered a technology-driven standoff with a bloc of authoritarian states led by Beijing and Moscow. Takach’s own entry point is the deal flow: the licensing agreements, the export-control filings, the national-security reviews that have made his working life steadily more fraught since the mid-2010s. The resulting book reads less like a Washington policy paper and more like a tour of the fault lines a transactional lawyer has watched open under his feet, with the stated ambition of persuading a general reader that the comforting language of “great-power competition” understates the stakes.

The central argument is that the world has slipped, over roughly a decade, from the hopeful interdependence that followed 1989 into a bipolar contest that Takach labels Cold War 2.0. The adversary bloc, in his telling, is anchored by the Chinese Communist Party under Xi Jinping and includes Russia, Iran, North Korea and a shifting cast of client states. The democracies — the United States, the European Union, the United Kingdom, Canada, Australia, Japan, South Korea, Taiwan and India in some domains — form the other pole. The decisive terrain is technological rather than territorial. Whichever bloc leads in a handful of foundational capabilities, Takach argues, will set the terms of global order for the second half of the century. He puts artificial intelligence at the centre of that list, but treats it as inseparable from the semiconductors it runs on, the quantum systems that may one day break its cryptography, the biotechnology it accelerates, and the cyber and space infrastructures that carry it. The recurring warning is that democracies, having spent a generation offshoring the industrial base for most of these technologies, risk discovering their dependence only when it is too late.

The structure moves outward from this thesis in layered chapters. An opening section lays out what Takach means by Cold War 2.0 and why he thinks the analogy to the original stand-off is useful rather than lazy. He argues that both contests are driven by irreconcilable political systems rather than by misunderstanding, that both are global rather than regional, and that both will likely be settled without direct great-power combat but through proxy wars, economic pressure, propaganda and technological lead. From there the book works through the technology domains in turn. Semiconductors receive the longest treatment, which is in keeping with their position as the chokepoint of everything downstream. Then come artificial intelligence and machine learning as a cluster — training compute, foundation models, the applications in surveillance and in weapons. Quantum computing follows, framed primarily as a cryptographic threat and a sensing opportunity. Biotechnology — gene editing, synthetic biology, pandemic-adjacent research — anchors a middle chapter. The final technical chapters cover cyber operations, space launch and the growing military dependence on commercial satellite constellations. The book then closes with policy chapters on export controls, investment screening, alliance architecture and the public-private partnerships Takach believes the democracies will need to build.

The concrete examples are where the book earns its length. On semiconductors, Takach walks the reader through the ASML extreme-ultraviolet lithography machine as the single most important industrial artefact in the world, built in the Netherlands with American optics and sold under tightly negotiated export licences to a handful of foundries. He traces how TSMC in Taiwan came to produce the leading-edge logic chips that train the frontier AI models and run the latest missiles, and why the island’s position has become the pivot of the global semiconductor supply chain. He revisits the Huawei affair — the 5G equipment contracts, the extradition proceedings against Meng Wanzhou, the eventual American sanctions that cut Huawei off from Taiwan Semiconductor Manufacturing Company — as a case study in how a Western-trained general-purpose technology was quietly turned to the service of a party-state and then slowly fenced off by Washington. The passage of the CHIPS and Science Act in 2022 and the Biden administration’s October 2022 export-control package on advanced computing to China are treated as watershed moments: imperfect, legally fragile, but the first serious attempt in decades to run industrial policy at the frontier.

On artificial intelligence, Takach’s treatment is less technical than that of specialist books but ranges wider across applications. He points to the Chinese internal security apparatus — the facial-recognition networks in Xinjiang, the social-credit experiments in cities like Rongcheng, the integration of AI into the Great Firewall — as a demonstration of what state-directed AI looks like when a government is unencumbered by civil-liberties constraints. He contrasts this with the more fragmented, private-sector-led American ecosystem anchored by OpenAI, Anthropic, Google DeepMind and Meta. The book argues that the democracies’ advantage in frontier model research is real but narrowing, and that it rests on three things: access to the leading-edge chips, access to the global pool of talent, and the openness that makes allied scientists want to work in democratic countries in the first place. Military applications get their own treatment — loitering munitions in Ukraine, Project Maven and its successors in the United States Department of Defense, the autonomy debates around the Defense Innovation Unit and the Replicator initiative. Takach is cautious about the hype around fully autonomous lethal systems but direct about the near-term reality: machines are already selecting targets in Ukraine, and the question of meaningful human control is being settled in the field rather than in Geneva.

The Ukraine war threads through the whole book. Takach returns to it as the proving ground on which the thesis is being tested in real time: Russian drones built with Western microcontrollers slipped through sanctions, Ukrainian software teams iterating weekly on target recognition, Iranian Shahed drones launched in waves against civilian infrastructure, North Korean artillery shells travelling the Trans-Siberian railway to the front. The implication he draws is uncomfortable for European readers in particular. A continent that spent the post-Cold War decades treating defence as a residual budget item is now watching Ukrainians fight for its security with equipment the continent cannot replace at the rate it is being destroyed. The book is not alarmist on this point so much as patient — Takach does not call for mobilisation, he describes a drift toward unseriousness that the democracies have not yet reversed.

Quantum computing receives a shorter and more speculative chapter. Takach is careful not to oversell the state of the art — he notes that cryptographically relevant quantum computers remain years away by the published estimates of IBM, Google and the Chinese Academy of Sciences — but he presses the harvest-now-decrypt-later problem: adversaries can be assumed to be collecting encrypted traffic today against the day their machines can read it. The policy conclusion is that the migration to post-quantum cryptography is urgent even if the quantum machines themselves are not yet here. On biotechnology, the chapter touches CRISPR and the gain-of-function debates sharpened by COVID-19, and argues that the biosecurity dimension of the US-China contest is seriously underpoliced compared with the semiconductor one. Space and cyber are treated more briefly: the role of SpaceX’s Starlink in Ukraine, the strategic exposure created by one company’s commercial satellite dominance, the near-constant Chinese and Russian cyber intrusions into Western infrastructure.

The policy chapters at the end turn from description to prescription. Takach argues that the democracies need a formal technology-alliance architecture that goes beyond the existing patchwork of G7 statements, the Chip 4 informal grouping, AUKUS and the Trade and Technology Council. He calls for a durable multilateral export-control regime with teeth — closer to the old COCOM framework that governed Western technology exports to the Soviet bloc, with staffed secretariats and compliance monitoring, rather than ad-hoc coordination out of Washington. He wants investment screening harmonised across the democracies, standards work in bodies like the International Telecommunication Union organised rather than drifting, and research-security measures in universities taken seriously after decades of neglect. He argues for a strategic use of government procurement to seed national champions in critical technologies — not as a general industrial policy, but as a narrow tool against the specific chokepoints the adversary bloc has identified. And he closes with a defence of the democratic model itself: not the specific policies of any administration, but the proposition that open societies with real courts, free press and competitive politics outperform authoritarian ones over any horizon longer than a decade, because they adapt and the autocracies cannot.

The book’s reception, broadly, has followed the fault lines in the field it joined. Reviewers aligned with the hawkish turn in Washington and allied capitals — the China-focused staff at the Center for Strategic and International Studies, writers at Foreign Affairs and Foreign Policy, the semiconductor-geopolitics commentators grouped around Chris Miller’s Chip War — have treated it as a competent synthesis for the policy-literate general reader. Those who dissent from the Cold War framing have been sharper. Critics argue that treating China as a monolithic strategic adversary rather than a trading partner with whom disagreements can be managed is a self-fulfilling prophecy, that the historical analogy to the Soviet Union flatters a more complex situation, and that export controls will prove counterproductive by accelerating Chinese indigenous capability. Takach engages with the self-fulfilling-prophecy objection directly in the book; he is less generous with the second criticism, which he treats as a wishful residue of the engagement-era consensus.

Within the specialist literature on AI in war the book sits as a general-reader on-ramp rather than a technical reference. It does not do what Paul Scharre’s Four Battlegrounds does on the specific competitive dynamics of AI compute and talent, and it does not do what Audrey Kurth Cronin’s Power to the People does on the diffusion of lethal capability to small actors. It covers more ground than any of those specialist works and expects less of the reader. The pairing that makes the most sense is probably with Chris Miller’s Chip War, which Takach cites, for the semiconductor history; with Scharre for the AI-specific competitive picture; and with any of the running journalism on Ukraine for the operational reality the book keeps returning to.

The judgement to make about Cold War 2.0 a few years after publication is where the thesis has aged well and where the surface detail has already dated. The thesis has aged well. A second Trump administration arrived in Washington, the European Chips Act moved into implementation, Taiwan’s political situation hardened, and the restrictions on advanced computing exports to China have been repeatedly tightened rather than relaxed. The specific technical details have aged less well, as any book writing about frontier AI compute at a given moment is destined to. The gap between the leading American labs and the leading Chinese labs that Takach reports as substantial in early 2024 has narrowed considerably in some benchmarks since. The ASML export-control picture has moved. The CHIPS Act implementation has unfolded in ways the book could only sketch. A reader coming to Cold War 2.0 today should treat the policy chapters as a snapshot of a conversation that has since moved and the framing chapters as the enduring argument the book was really making. The framing is the thing worth keeping, and that framing — that the democracies are in a long contest they cannot opt out of, and that the technologies on which the contest will turn are already in play — is where the book is still worth the hours.