Big Tech’s largest platforms are starting to resemble the old-line computing empires they once claimed to disrupt. Their dominance in infrastructure, their spending on hardware, and the regulatory pressure gathering around them all echo the era when IBM set the rules for the digital economy. The comparison is not perfect, but the structural rhymes are close enough that investors, regulators, and customers should pay attention.
In the 1960s, IBM’s mainframes were the center of gravity for corporate computing. Today, cloud data centers and AI clusters run by a handful of giants play a similar role for everything from media to retail. The question is whether Alphabet, Microsoft, Amazon, Meta, and their peers are building a durable foundation, or repeating the same concentration and complacency that eventually humbled IBM.
The new mainframes: hyperscalers as infrastructure landlords
The core of the analogy lies in infrastructure. In the 1960s, IBM sold powerful but non-interactive mainframes that sat at the heart of corporate IT, and access to computing flowed through its machines and service contracts. Earlier in that decade, IBM had already turned itself into the default supplier of commercial computers, after UNIVAC briefly led the market, only to see IBM capture 85% of computer sales by the mid‑1950s. That level of concentration meant that buying computing power was, in practice, buying IBM.
Today’s cloud and AI “hyperscalers” are playing a similar role. The four Big Tech giants most associated with hyperscale infrastructure, including Microsoft, Alphabet, Amazon, and Meta, are collectively committing staggering sums to data centers and chips, with $650 billion in capital spending expected this year alone. That investment is aimed at AI clusters, networking, and data center infrastructure that everyone else must rent rather than own. As one analysis of these Big Tech “hyperscalers” notes, the race is not just to build AI models, but to control the physical substrate on which everyone else’s models will run.
AI spending binge and the IBM-style arms race
The AI boom has turned that infrastructure push into something closer to an arms race. Big Tech is set to spend More than half a trillion US dollars on AI infrastructure this year, a figure that includes specialized chips, new data centers, and the power and cooling to run them. The same four hyperscalers, including Microsoft, Alphab, Amazon, and Meta, are at the center of this surge, and their capital plans are reshaping global supply chains for semiconductors and networking gear. In the 1960s, IBM made a similar “bet the company” move with its System/360 line, a multibillion‑dollar platform investment that locked customers into its ecosystem for a generation.
Investors are already asking when these AI bets will pay off. Recent earnings coverage has highlighted that Adding computing capacity is not cheap, and Microsoft, Amazon, Alphabet and Met are pouring cash into GPUs and data centers long before the revenue from AI copilots and enterprise subscriptions fully materializes. Reporting on the AI build‑out notes that Big Tech is set to spend $650 billion in 2026 as AI investments soar, and that the race to dominate the burgeoning AI market has drawn in IBM, META, GOOG, NVDA, and AMZN, with Laura Bratton pointing to leading chipmaker Nvidia (NVDA) as a key beneficiary. The scale and front‑loaded nature of this spending look very much like IBM’s 1960s gamble, which required patience from shareholders and deep pockets from the company itself.
IBM’s rise, fall, and the warning for today’s giants
To understand the risk in this strategy, it helps to recall how IBM’s dominance unraveled. In the 1980s, In the 1980s, IBM’s profit margins suffered a steep decline, and because the company’s costs remained level, profits dropped sharply. Critics argued that IBM had become too attached to its legacy hardware and service model, even as the personal computer and client‑server architectures shifted value away from centralized mainframes. The company had to shed less profitable portions of its business and rethink its role in the computing stack.
The seeds of that vulnerability were planted in the 1960s, when IBM’s System/360 became the standard corporate computer. The company’s own archives and later analysis describe how IBM built the 360 Series of computers as a unified platform, locking in customers but also locking itself into a particular architecture. A later account of the PC era shows how the IBM PC, which initially helped the company win the personal computer market, eventually slipped from its control as others cloned the hardware and operating system. The lesson for today’s platforms is that even overwhelming market share and technical leadership can erode quickly if the underlying architecture shifts or if value migrates to more modular, open systems.
Regulators revisit the IBM and Microsoft playbook
Regulation is another area where history is starting to rhyme. IBM spent years under antitrust scrutiny for its dominance in mainframes, and later, Microsoft faced its own landmark case over Windows and Internet Explorer. A retrospective on those battles notes that The judge overseeing the Microsoft case sought to fast‑track it, but there was no guarantee the appeals court or the Supreme Court would act swiftly, underscoring how slowly structural remedies can move compared with the pace of technology. Those cases nonetheless set important precedents about tying, bundling, and the responsibilities of dominant platforms.
Today’s Big Tech firms are facing a new wave of antitrust scrutiny that explicitly references their control over infrastructure. One analysis of the current enforcement landscape asks what would happen If Amazon were to lose its case against the FTC, and whether a judge would resist a requested divestiture of Amazon Marketplace given how central it has become. Another legal preview notes that Jan court opinions on remedies in tech cases now contain extended discussions of generative AI, and that judges are trying to account for how quickly the market landscape can change. The regulatory questions are no longer just about search or social media, but about who owns the AI and cloud infrastructure that everyone else must use.
Vertical integration and the return of the full-stack giant
One of the most striking parallels with the IBM era is the renewed push for vertical integration. In the 1960s, IBM controlled chips, hardware, operating systems, and services, a strategy that a later retrospective described as a Gamble In The 60 that nonetheless Continues To Reap Rewards decades later. Forty years after the launch of System/360, IBM was still benefiting from the ecosystem and customer relationships that platform created. Today’s Big Tech firms are similarly trying to own everything from custom AI accelerators to cloud platforms and consumer applications, creating end‑to‑end stacks that are hard for rivals to dislodge.
Chip designers see this trend clearly. One industry analysis notes that And the reason hyperscalers are building their own silicon is the efficiency and performance gains of building those completely vertically integrated solutions. They are doing that because controlling the compute substrate lets them tune hardware for their own AI workloads and lock in customers to proprietary APIs and services. In effect, the cloud is becoming the new mainframe, and the companies that own it are reviving a model that looks very familiar to anyone who watched IBM’s rise.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
