AI

The AI Reckoning: Why Meta and Microsoft Are Cutting Up to 23,000 Jobs While Pouring Billions into Artificial Intelligence

Published

4 hours ago

April 24, 2026

On Thursday, April 24, 2026, two of the world’s most powerful technology companies delivered remarkably similar messages to their workforces, framed in the polished bureaucratic language of “efficiency” and “investment prioritization.” Meta announced it would eliminate roughly 8,000 jobs — 10 percent of its global workforce — while simultaneously canceling 6,000 open positions, effective May 20. Microsoft, on the very same day, offered voluntary retirement buyouts to approximately 8,750 U.S. employees, or about 7 percent of its domestic workforce, in what is described as the first program of its kind in the company’s 51-year history.

Together, the moves affect up to 23,000 positions across two of the most profitable companies ever to exist. That is not a quarterly adjustment. That is an industrial reckoning.

The surface-level paradox is arresting: Meta expects to spend between $115 billion and $135 billion on capital expenditures in 2026 alone, more than double the $72.2 billion it spent in 2025. Microsoft recently committed over $80 billion to AI infrastructure and is reporting quarterly revenues of $81.3 billion. These are not struggling enterprises trimming costs in a downturn. They are dominant, cash-rich platforms undergoing a fundamental reorganization of what “work” inside a technology company actually means.

The deeper question — the one that boards, economists, policymakers, and frankly every mid-career software engineer should be grappling with — is whether this represents a rational, healthy recalibration for a new era of productivity, or the opening act of a structural displacement whose downstream effects we are only beginning to comprehend.

The Arithmetic of the AI Economy

To understand what Meta and Microsoft are doing, you need to understand the economics they are navigating. The business case for large language models and AI-driven automation is, at its core, a substitution argument: AI can perform certain cognitive and creative tasks at near-zero marginal cost once the infrastructure is built. The infrastructure, however, is extraordinarily expensive — requiring massive GPU clusters, purpose-built data centers, enormous electricity contracts, and a relatively small number of extremely specialized engineers.

This creates a peculiar arithmetic. Capital expenditure explodes. Operational headcount — particularly in middle layers of the organization — becomes a liability rather than an asset.

Meta’s internal memo from Chief People Officer Janelle Gale frames the layoffs explicitly around this logic. The reductions are, she wrote, “part of our continued effort to run the company more efficiently and to allow us to offset the other investments we’re making.” Notably, the company is also restructuring its entire organizational model around AI-focused “pods,” creating new internal roles — “AI builder,” “AI pod lead,” “AI org lead” — while transferring engineers from across the business into an expanded Applied AI organization. This is not simply headcount reduction; it is a deliberate rewiring of the corporate organism around machine intelligence.

Microsoft’s approach is more architecturally elegant — and, arguably, more revealing. The “Rule of 70” program targets employees whose age and years of service sum to at least 70, at the senior director level and below. It is, in effect, a precision instrument designed to thin the layer of experienced, expensive, institutionally knowledgeable staff — precisely the cohort that, in prior decades, would have been the most insulated from layoffs. CEO Satya Nadella noted at Microsoft’s Build conference last year that approximately 30 percent of the company’s code is now written by AI tools. When a machine can replicate a senior engineer’s output at scale, institutional knowledge loses some of its traditional premium.

Why Meta Is Cutting 8,000 Jobs — and What That Actually Signals

The May 2026 cuts are not Meta’s first. They are, in fact, the third wave of workforce reductions this year alone, following approximately 2,000 earlier eliminations. Reuters reported last week that additional cuts are planned for the second half of 2026. This is less a single event than a sustained, deliberate, multi-phase reorganization.

Context matters here. Meta’s 2022 layoffs — 11,000 people, or 13 percent of its workforce — were driven by a revenue shock following Apple’s privacy changes and the market’s rejection of the metaverse bet. The 2023 round, another 10,000 jobs, was part of what Mark Zuckerberg branded the “Year of Efficiency.” This time, the framing is different. Revenue is not the problem. Meta’s total expected expenses for 2026 are projected between $162 billion and $169 billion, driven by AI infrastructure and talent acquisition — and those expenses are being funded by a profitable, growing business.

That distinction matters enormously. When companies lay off employees during revenue crises, the calculus is forced and defensive. When they do so during record investment cycles, it is strategic and, in a meaningful sense, voluntary. Meta is not cutting because it cannot afford to pay these people. It is cutting because it has decided those people are less valuable than the AI systems it is building to replace aspects of their functions.

There is something worth sitting with in that distinction. These are not performance-based terminations. The memo explicitly acknowledges that affected employees “have made meaningful contributions.” They are being let go because the direction of the organization has fundamentally changed around them — not because they failed, but because the map of valued capability has been redrawn.

Microsoft’s First-Ever Voluntary Buyout: A Blueprint, or a Bellwether?

Microsoft’s decision to deploy voluntary buyouts — a mechanism more commonly associated with legacy industrial companies managing generational transitions than with a cloud-computing titan — deserves particular attention. The company has conducted multiple rounds of involuntary layoffs in recent years, cutting 9,000 positions as recently as last summer. The pivot toward a voluntary program represents a different kind of strategic signal.

By offering long-tenured employees a financially dignified exit, Microsoft accomplishes several things simultaneously. It reduces payroll costs weighted toward senior-level salaries and legacy compensation structures. It creates runway to hire a new generation of AI-native engineers without inflating total headcount. And it does so in a manner that — for now — avoids the morale craters and employer-brand damage that accompany involuntary mass layoffs.

The structural elegance of the Rule of 70 formula, however, should not obscure its human complexity. The employees targeted are those whose decades of service once represented job security. In an environment where Azure AI can digest institutional documentation in seconds, the implicit argument is that the value of accumulated human knowledge is being repriced. Rapidly.

Whether all 8,750 eligible employees will accept the offer is an open question. Many will calculate that their internal leverage — built over years of relationships, proprietary context, and organizational navigation — remains irreplaceable in ways that models cannot yet fully emulate. They may be right. They may also be underestimating the pace of substitution.

The Productivity Paradox, Revisited

Economists have long wrestled with what Robert Solow famously observed in 1987: “You can see the computer age everywhere but in the productivity statistics.” The first wave of digitization promised enormous efficiency gains that took decades to materialize in aggregate economic data. There is genuine, serious debate about whether AI will repeat this pattern — delivering micro-level efficiencies at the firm level while broader societal productivity gains remain elusive, displaced by transition costs, retraining friction, and the concentration of gains among capital holders.

What Meta and Microsoft are demonstrating is a clear answer to one part of that question: at the firm level, AI is already powerful enough to justify eliminating significant portions of a highly paid, highly skilled workforce. The question of whether the displaced workers find equivalent employment elsewhere — whether the historical promise of technology, that it creates as many jobs as it destroys, holds in this iteration — is one that macroeconomists and policymakers cannot answer with confidence in April 2026.

Historical analogies are imperfect but instructive. The automation of manufacturing in the mid-20th century did eventually produce new categories of employment, but the transition was measured in decades and extracted enormous social costs from specific geographies and communities. Technology sector layoffs feel different — the affected workers are highly educated, geographically mobile, and better resourced than factory workers of the 1970s — but the structural dynamic has more in common with those earlier transitions than comfortable Silicon Valley narratives tend to acknowledge.

The Talent Concentration Problem

Perhaps the most underappreciated dimension of this moment is what it implies for talent distribution and long-term innovation capacity. Meta is splurging on acqui-hires and elite AI researchers — it recently acquired buzzy AI startups including Moltbook and Manus, and has been assembling a superintelligence laboratory with eye-watering compensation packages. Microsoft has explicitly exempted AI-focused teams from its hiring freeze. Amazon and Google are doing analogous things.

The result is an intensifying concentration of AI talent and infrastructure capital within a handful of firms that already dominate their respective markets. When 23,000 experienced technology workers are released into a labor market simultaneously, some will land well. A portion will find roles at smaller firms, startups, or in adjacent sectors. But a meaningful cohort will struggle, particularly those in roles — project management, middle-layer software engineering, content operations, HR — that AI is demonstrably eroding across the board.

Meanwhile, the engineers who remain inside these companies, and those being recruited to join, are becoming increasingly specialized and increasingly expensive. This narrows the distribution of who benefits from the AI boom in ways that have implications not just for income inequality but for the diversity of perspectives shaping the most consequential technology in a generation.

The Regulatory Vacuum

Governments, with a few notable exceptions, have not caught up. The European Union’s AI Act introduces tiered requirements around transparency and accountability but does not directly address workforce displacement mechanisms. The United States has no coherent federal framework addressing AI’s labor market effects at all. Individual countries are experimenting — some with AI taxes, others with retraining levies — but none has yet devised policy interventions commensurate with the scale and speed of the shift underway.

This is not an argument for reflexive regulation. Heavy-handed intervention in technology development carries its own costs, and there are real risks in designing policy around yesterday’s AI rather than tomorrow’s. But the absence of any serious public-sector engagement with questions of workforce transition, anti-competitive talent concentration, and the distributional effects of AI-driven corporate restructuring represents a significant governance gap — one that will become harder to fill the longer it persists.

The companies themselves are not passive actors here. They lobby actively against labor market regulations, fund think tanks that favor their preferred policy frameworks, and have become extraordinarily adept at shaping public narratives around AI’s job creation potential. That narrative deserves skepticism, not reflexive hostility — but scrutiny, proportionate to the power these firms wield.

Right-Sizing or Structural Rupture? A Reasoned Assessment

Is what Meta and Microsoft are doing a legitimate, healthy recalibration for the AI era — or something more troubling?

The honest answer contains both.

There is a genuine case that some portion of these cuts reflects normal organizational evolution. Companies periodically need to realign their workforce with their strategic direction. AI genuinely does enable certain tasks to be performed with fewer people. Organizations that fail to adapt to technological shifts tend to lose competitive position, which ultimately destroys more jobs than it preserves. The argument for efficiency is not cynical.

But the speed, scale, and simultaneity of this transition — across not just Meta and Microsoft but Amazon, Google, Snap, and dozens of other firms in recent months — point to something more structural than a routine restructuring. When the largest technology companies in the world are all, simultaneously, reducing their human workforce while dramatically increasing their capital investment in AI systems, that is not a collection of independent firm-level decisions. It is a coordinated inflection point in the relationship between capital and labor in knowledge work.

The risks are real and underweighted in current discourse. Employee morale inside these organizations — among those who remain, not just those who leave — is a genuine concern. Trust in large institutions takes years to build and can erode in a single earnings cycle. The innovation that emerges from diverse teams working in psychologically secure environments is qualitatively different from what emerges from a high-surveillance, high-anxiety “pod” structure where engineers know their output is being benchmarked against AI tools. Meta’s recent disclosure that it has been tracking employee keystrokes and mouse movements to train AI systems — which some staff reportedly criticized — offers an unsettling preview of where the logic of substitution leads.

What Business Leaders and Policymakers Should Take From This

For corporate leaders navigating similar decisions, the strategic imperative is clarity over comfort. Workforce transitions managed with transparency, genuine dignity, and robust support — including retraining investment, not just severance — tend to preserve the organizational culture and employer brand that sustain long-term competitive advantage. The companies that will emerge strongest from this decade are those that treat the humans they are releasing as alumni rather than liabilities.

For policymakers, the agenda is more urgent. Universal retraining infrastructure, portable benefits independent of employer tenure, and serious investment in understanding AI’s net labor market effects are not luxuries for a later policy cycle. They are present-tense governance responsibilities. The European Commission’s early moves toward an AI liability framework, and some U.S. states’ exploration of technology workforce transition funds, are directionally correct — but structurally insufficient.

For the 23,000 individuals directly affected — and the many more who will follow in subsequent waves across the industry — the immediate reality is one of uncertainty. Some will thrive. The labor market for experienced technology workers, while tightening in certain specializations, remains reasonably absorptive at the aggregate level. But “aggregate” is cold comfort to a 54-year-old senior engineer with a Rule-of-70 number and a severance package measuring weeks, not the decades of career that precede it.

Conclusion: The Bill We Have Not Yet Paid

The AI revolution being financed by Meta’s $135 billion and Microsoft’s $80-plus billion infrastructure buildout will almost certainly generate enormous economic value. The productivity gains, once they propagate through the broader economy, may well exceed the disruptions they cause. That is the optimistic case, and it is not baseless.

But revolutions do not distribute their benefits automatically or equitably. The costs of this transition are being paid now, in real time, by specific individuals with specific families and mortgages and professional identities. The gains are being accrued, for the moment, primarily by shareholders, a narrow band of AI researchers, and the infrastructure firms supplying the data center components of this buildout.

That asymmetry — between who bears the transition cost and who captures the productivity gain — is the central moral and economic challenge of the AI era. April 24, 2026 will not be remembered as the day two tech companies cut 23,000 jobs. It will be remembered, if we are honest about it, as the day the reckoning became impossible to look away from.

The question is not whether the AI era requires a workforce transformation. It plainly does. The question is whether we have the institutional imagination and political will to ensure that transformation is navigated with something approaching justice.

That question remains, conspicuously, unanswered.

Key Data Points at a Glance

Meta layoffs 2026: ~8,000 jobs eliminated (10% of workforce), effective May 20, 2026; 6,000 open roles canceled; third wave of 2026 cuts, with more planned for H2
Meta AI spending 2026: $115–135 billion in capital expenditure (up from $72.2B in 2025); total projected expenses of $162–169 billion
Microsoft voluntary buyouts: ~8,750 U.S. employees eligible (7% of 125,000 U.S. staff); Rule of 70 formula (age + years of service ≥ 70); first program of its kind in the company’s 51-year history; details arriving May 7 with 30-day decision window
Microsoft AI infrastructure: $80+ billion committed to AI data center buildout; $81.3 billion in quarterly revenue; approximately 30% of code now AI-generated per Satya Nadella
Combined impact: Up to ~23,000 positions affected across the two companies
Broader context: Amazon, Google, and Snap have conducted parallel workforce reductions in 2026, all citing AI-era restructuring

Discover more from The Economy

Subscribe to get the latest posts sent to your email.

Up Next

KPMG and EY Demote Partners: The Definitive End of the Big Four Job-for-Life Model

Don't Miss

The Tax That Quietly Grew: OECD Wage Levies Hit Their Highest Point in Nearly a Decade

Click to comment

Analysis

Bezos’s Project Prometheus Nears $38 Billion Valuation: The Real AI Race Is Just Beginning

Published

3 days ago

April 21, 2026

Abdul Rahman

A $10 billion funding round—his first operational role since Amazon—signals a shift from digital chatbots to the physical world. But as AI funding hits $242 billion in a single quarter, is the real bubble in our power grid?

Introduction

In Greek mythology, Prometheus stole fire from the gods and gave it to humanity. Today, Jeff Bezos is attempting a similar act of technological transference—not with a fennel stalk, but with a $10 billion checkbook.

According to a report first published by the Financial Times, Bezos’s secretive AI lab, code-named Project Prometheus, is on the verge of closing a massive funding round that values the startup at roughly $38 billion. The round, which includes heavyweights like JPMorgan and BlackRock, is reportedly being upsized due to “strong investor demand”.

This isn’t just another tech funding story. It marks Bezos’s first operational role since stepping down as Amazon CEO in 2021—and it is a deliberate, high-stakes bet that the next trillion-dollar opportunity in artificial intelligence lies not in writing better poetry or generating fake images, but in bending the physical laws of manufacturing, aerospace, and construction to our will .

The $38 Billion Bet on the Real World

For the last two years, the AI narrative has been dominated by large language models (LLMs) and the battle between OpenAI, Google DeepMind, and Anthropic. These models excel in the digital ether. Project Prometheus, by contrast, is targeting “physical AI”—systems designed to understand the laws of physics and revolutionize industries where atoms, not just bits, matter.

Co-founded with scientist Vik Bajaj (formerly of Google X), the venture is focused on applications in engineering, aerospace, semiconductors, and even drug discovery . Imagine an AI that can simulate the airflow over a new jet wing, predict material fatigue in a bridge, or optimize a factory floor in real-time—all without the costly, time-consuming cycle of physical prototyping. As Pete Schlampp, CEO of Luminary, recently noted, “AI is changing that by allowing” faster, cheaper digital testing.

The $38 billion valuation is staggering for an early-stage company, but it pales in comparison to the capital being mobilized around it. Bezos is reportedly also raising a separate $100 billion fund to acquire manufacturing companies outright and infuse them with Prometheus’s technology—a strategy that effectively creates a captive market for his lab’s innovations.

A Deluge of Dollars, A Scarcity of Power

To understand the significance of Bezos’s move, one must look at the broader macroeconomic context: the AI funding boom has reached a fever pitch. In the first quarter of 2026 alone, AI companies vacuumed up $242 billion in venture capital, accounting for a staggering 80% of all global startup investment during that period.

This is not just a trend; it is a financial singularity. The AI sector raised more money in three months than it did in all of 2025 combined. This capital influx is concentrated among a few “super rounds”: OpenAI raised $122 billion, Anthropic secured $30 billion, and xAI closed $20 billion .

However, the macro story reveals a critical vulnerability that makes Bezos’s physical AI pivot particularly shrewd. While money is abundant, physical infrastructure is not. A recent Bloomberg report found that roughly half of the AI data centers planned for 2026 in the U.S. have been delayed or canceled. The bottlenecks are not software glitches but tangible hardware: transformer shortages, grid strain, and supply chain paralysis. Only about one-third of the projected 12 GW of new computing capacity is actually under active construction.

The Competitive Chessboard: Why Bezos Is Building His Own Fire

Bezos’s move with Project Prometheus also needs to be read in the context of Amazon’s complex AI allegiances. The e-commerce giant is deeply entwined with Anthropic, having recently committed up to $25 billion in new investment into the Claude maker—a deal that reportedly values Anthropic at up to $3.8 trillion in private markets. Meanwhile, Amazon has also pledged $500 billion to OpenAI for a joint venture focused on stateful AI systems.

In this environment, relying solely on external partners—even those you’ve heavily funded—is a strategic risk. Prometheus gives Bezos a proprietary, in-house engine for the industrial revolution he envisions. It is a classic Bezos move: vertical integration via massive capital expenditure. The lab has already begun “snapping up office space in San Francisco” and “luring away top talent from OpenAI and Google DeepMind”. If you can’t buy the future, you build it yourself.

The Human Cost and the Political Backlash

The fire of Prometheus has always come with a warning. Bezos’s parallel $100 billion plan to acquire and automate factories—replacing human workers with AI-driven robots—has already drawn political fire. The narrative that AI will create more jobs than it destroys is being tested by the sheer scale and speed of this capital deployment.

On the political stage, figures like Senator Bernie Sanders are warning of “AI Oligarchs” planning to spend $300 million on the 2026 midterm elections, while Elon Musk and Andrew Yang debate the necessity of a federal “universal high income” to offset automation-driven job loss. The $38 billion valuation of Project Prometheus is not just a number on a term sheet; it is a geopolitical and socioeconomic fault line.

Conclusion: Fire from the Gods, Grounded in Reality

Bezos’s Project Prometheus nearing a $38 billion valuation is more than a fundraising milestone; it is a directional signal for global capital markets. It confirms that while the first wave of generative AI was about software eating the world, the second wave will be about AI rebuilding the physical world.

For investors, the lesson is clear: the highest returns will not come from funding the next clone of a chatbot but from solving the hardest problems in physics and engineering. For policymakers, the challenge is equally stark: the infrastructure to power this AI future does not exist yet. And for the rest of us, it is a reminder that even as we fret about what AI might do to our jobs, the real bottleneck isn’t the algorithm—it’s the electrical grid.

Bezos is betting $38 billion that he can steal this fire. The question is whether the rest of us are ready to live with the heat.

Discover more from The Economy

Subscribe to get the latest posts sent to your email.

AI

Apple’s Next Chief Ternus Faces Defining AI Moment: Tim Cook’s Replacement Must Lead iPhone-Maker Through Industry Shift

Published

3 days ago

April 21, 2026

Abdul Rahman

The tectonic plates of Silicon Valley shifted unequivocally on April 20, 2026. After a historic 15-year tenure that propelled the iPhone maker to an unprecedented $4 trillion valuation, Tim Cook announced he will step down on September 1, transitioning to the role of Executive Chairman. The keys to the kingdom now pass to John Ternus, the 51-year-old hardware engineering savant who has spent a quarter-century architecting the physical foundation of Apple’s most iconic modern devices.

Yet, as the dust settles on this long-anticipated Apple CEO succession plan, a stark reality emerges. Ternus is inheriting a radically different landscape than the one Cook received from Steve Jobs in 2011. Cook was tasked with scaling an undisputed hardware monopoly; Ternus is tasked with defending it against an existential software threat.

As Tim Cook’s replacement, Ternus assumes the mantle at the exact moment the technology sector pivots from the mobile era to the generative artificial intelligence epoch. His success will not be measured by supply chain efficiencies or incremental hardware upgrades, but by his ability to define and execute a winning Apple Intelligence strategy in an increasingly hostile, hyper-competitive market.

The Dawn of the Ternus Era: From Operations Titan to Hardware Visionary

To understand the trajectory of the John Ternus Apple CEO era, one must examine the fundamental differences in leadership DNA between the outgoing and incoming chief executives. Tim Cook is, at his core, an operational genius. His legacy is defined by mastery of global supply chains, geopolitical diplomacy, and the methodical extraction of maximum margin from the iPhone ecosystem.

Ternus, conversely, is an engineer’s engineer. Having overseen the iPad, the AirPods, and the monumental transition of the Mac to Apple Silicon, he deeply understands the intersection of silicon and user experience. Insiders report that Ternus brings a decisively different management style to the C-suite. Where Cook historically preferred a Socratic, hands-off approach to product development—acting as a consensus-builder among top brass—Ternus is known for making swift, definitive product choices.

This decisive edge is precisely what the company requires as it navigates its most pressing vulnerability: its artificial intelligence deficit. A recent Reuters report on Apple’s corporate governance and succession highlights that Ternus’s mandate is to aggressively reinvent the product lineup to meet modern consumer expectations. However, being a hardware visionary is no longer sufficient. The modern device is merely an empty vessel without a pervasive, context-aware intelligence layer running beneath the glass.

The Intelligence Deficit: Combating the Decline in Apple AI Market Share

Apple’s entry into the artificial intelligence arms race has been characterized by uncharacteristic hesitation and strategic missteps. While Microsoft, Google, and Meta sprinted ahead with large language models (LLMs) and advanced neural architectures, Apple opted for a walled-garden, on-device approach that has struggled to keep pace with cloud-based capabilities.

The Apple AI market share currently lags behind its chief rivals, largely due to a fragmented rollout and technological bottlenecks. The initial deployment of Apple Intelligence was marred by delayed features and an overly cautious integration of third-party tools. Most notably, in late March 2026, a botched, accidental rollout of Apple Intelligence in China—a market where Apple lacks the requisite regulatory approvals and relies heavily on local partners to bypass restrictions—highlighted the immense logistical hurdles the company faces.

As highlighted by Bloomberg’s recent analysis on Apple’s AI deployments, Apple’s decision to integrate Google’s Gemini model to power a revamped Siri underscores a painful truth: the company cannot win the AI war in isolation. Ternus must immediately stabilize these partnerships while simultaneously accelerating Apple’s in-house foundational models. He inherits an AI division that saw the departure of key leadership in late 2025, leaving a strategic vacuum that the new CEO must fill with undeniable urgency.

Recalibrating the Apple Intelligence Strategy

The challenge for Ternus is twofold: he must merge his innate understanding of hardware architecture with an aggressive software and cloud strategy. According to a Gartner report on AI adoption and edge computing, the future of enterprise and consumer tech lies in a hybrid model—balancing the privacy and speed of edge computing (processing on the device) with the raw, expansive power of cloud-based LLMs.

Ternus’s immediate priority will be launching iOS 27 and the anticipated overhaul of Siri. It is no longer enough for Siri to be a reactive voice assistant; it must evolve into a proactive, system-wide autonomous agent capable of reasoning, executing complex in-app tasks, and seamlessly analyzing user data without compromising Apple’s rigid privacy standards.

This is where Ternus’s decisive nature will be tested. He must be willing to cannibalize legacy software structures and perhaps even open the iOS ecosystem to deeper third-party AI integrations than Apple is historically comfortable with. The Apple Intelligence strategy must pivot from being a defensive moat to an offensive spear.

The Future of Apple Hardware: AI-First Architecture

Because Ternus is rooted in hardware, his most significant leverage lies in reimagining the physical devices that will house these new AI models. The future of Apple hardware is inextricably linked to the evolution of neural processing units (NPUs).

In tandem with Ternus’s promotion, Apple elevated its silicon architect, Johny Srouji, to Chief Hardware Officer. This alignment is not coincidental. It signals a unified front where hardware and silicon are co-developed exclusively to run massive AI workloads. We can expect future iterations of the iPhone and Mac to feature a radical redesign of thermal management and memory bandwidth, specifically tailored to support on-device inference for generative AI.

Furthermore, Ternus—who reportedly expressed caution regarding the high-risk development of the Vision Pro and the now-cancelled Apple Car—will likely ruthlessly prioritize form factors that deliver immediate AI value. We are likely to see a convergence of wearables and AI, where devices like AirPods and the Apple Watch act as persistent, ambient interfaces for Apple Intelligence, rather than relying solely on the iPhone screen.

Silicon Valley Geopolitics: The Burden of the $4 Trillion Crown

Beyond the silicon and software, Ternus faces a daunting geopolitical landscape. Tim Cook was a master statesman, successfully navigating the treacherous waters of the US-China trade wars, negotiating with consecutive presidential administrations, and maintaining a fragile equilibrium with international regulators. As The Wall Street Journal’s ongoing coverage of tech monopolies points out, global regulatory bodies are increasingly hostile toward Big Tech’s walled gardens.

With Cook serving as Executive Chairman and managing international policy, Ternus has a temporary shield. However, the ultimate responsibility for antitrust compliance, App Store regulations, and navigating the complex AI compliance laws of the European Union and China will soon rest entirely on his shoulders.

Conclusion: The Decisive Leadership Required for Apple’s Next Decade

As September 1 approaches, the global markets are watching with bated breath. John Ternus is not stepping into a role that requires a steady hand to maintain the status quo; he is stepping into a crucible that requires a wartime CEO mentality.

The transition from Tim Cook to John Ternus marks the end of Apple’s era of operational perfectionism and the beginning of its most critical existential challenge since the brink of bankruptcy in the late 1990s. To justify its $4 trillion valuation, the future of Apple hardware must become the undisputed premier vessel for consumer artificial intelligence.

Ternus possesses the engineering pedigree, the institutional respect, and the decisive operational mindset required for the job. Now, he must prove he possesses the visionary foresight to lead the iPhone maker through the most disruptive industry shift in a generation. The hardware is set; the intelligence is pending.

Discover more from The Economy

Subscribe to get the latest posts sent to your email.

AI

Could AI’s Leading Men Become as Powerful as Ford or Rockefeller? For Now, They Are Still a Long Way Behind.

Published

6 days ago

April 18, 2026

Abdul Rahman

The five men reshaping intelligence — Dario Amodei, Demis Hassabis, Elon Musk, Mark Zuckerberg, and Sam Altman — command wealth, attention, and technological leverage that no previous generation of innovators has enjoyed. Yet the distance between their present dominance and the systemic, civilization-bending grip once exercised by John D. Rockefeller or Henry Ford remains vast — and poorly understood.

Imagine a boardroom meeting in 2035. The agenda is simple: who controls the infrastructure of thought itself? A decade earlier, five men launched what many called the most consequential technological disruption since electricity. By 2026, their companies had collectively captured trillions of dollars in market value, reshaped labor markets across three continents, and triggered geopolitical confrontations from Brussels to Beijing. And yet, if you measure their power by the standards history reserves for its true industrial titans — the men who didn’t just build industries but became them — the five AI leading men of our era still have a very long way to go.

That is not a comfortable argument to make. The numbers alone seem to render it absurd. Elon Musk’s net worth now exceeds $811 billion, a figure that surpasses the GDP of Poland. Musk’s February 2026 all-stock merger of SpaceX and xAI created a combined entity valued at $1.25 trillion — a single transaction larger than the entire U.S. defense budget. OpenAI, now valued at approximately $500 billion, counts some 800 million weekly active users of ChatGPT, a number that would have seemed science fiction five years ago. Anthropic — founded by Dario Amodei and his sister Daniela — reached a valuation of $380 billion in early 2026, while Meta has committed to spending $115 to $135 billion in capital expenditure in 2026 alone, with an astonishing $600 billion pledged toward data centers through 2028.

These are not ordinary fortunes. They are structurally new categories of wealth concentration. And still, the Rockefeller comparison fails — and fails instructively.

What Made a Tycoon a Tycoon: The Three Pillars of Historical Power

To understand why AI tycoons remain a long way behind their Gilded Age predecessors, one must first understand what actually made Rockefeller and Ford so uniquely dangerous to the social order of their time. It was not simply their wealth. Adjusted for GDP, Rockefeller’s peak fortune has been estimated at roughly $400 billion in today’s dollars — comfortably surpassed by Musk. What made Standard Oil a civilizational force was something more specific and more structural: the simultaneous control of physical infrastructure, political capture, and cultural monopoly.

Rockefeller didn’t just refine oil; he controlled approximately 91% of United States oil refining capacity by the mid-1880s through ownership of the pipelines, the railroad rebates, and the pricing mechanisms that every competitor had to use to survive. He didn’t lobby Congress — he owned the conversation. Ford, similarly, didn’t just manufacture cars; he built company towns, set wages for an entire economy, and deployed a private security apparatus — the Ford Service Department — to enforce his will on a captive workforce. Both men bent the physical world to their models in ways that left no exit for competitors, workers, or governments.

That is the three-pillar framework that the AI quintet has not yet replicated: physical infrastructure lock-in, political capture, and cultural monopoly. The gap between aspiration and achievement on each of these dimensions is the real story of power in 2026.

Infrastructure: Who Controls the Pipes?

The most important question in any era of technological transformation is not who builds the smartest machine, but who controls the plumbing. Rockefeller’s genius was not chemistry — it was logistics. He understood that the pipeline was more powerful than the refinery.

In the AI economy, the equivalent of the pipeline is the data center, the chip, and the undersea cable. Here the picture for the quintet is mixed at best. Mark Zuckerberg’s Meta is building on the most ambitious scale — two mega-clusters that dwarf any corporate construction project in a generation — but the silicon in those data centers is manufactured almost entirely by NVIDIA, a company none of the five control. Musk’s SpaceX-xAI merger is the most vertically integrated attempt to replicate Rockefeller’s pipeline logic: orbital data centers fed by Starlink satellites, in theory giving xAI the physical substrate to train and deploy models without dependence on third-party cloud providers. But as of 2026, that vision remains largely prospective. xAI’s Grok competes credibly against ChatGPT and Claude, but it does not yet possess the proprietary infrastructure advantage that would make it structurally inescapable.

Sam Altman, for his part, has no direct equity in OpenAI, earning a nominal salary of roughly $65,000 per year. His influence derives almost entirely from his position at the helm of the world’s most recognizable AI brand — a form of power that is real, but brittle. The moment a better or cheaper model displaces GPT, the institutional moat begins to crack. Rockefeller, by contrast, had no such vulnerability: he owned the pipes regardless of whose oil flowed through them.

Dario Amodei’s Anthropic presents a different case. With a $380 billion valuation, enterprise AI revenues reportedly growing at exponential rates, and a model — Claude — that has captured an estimated 40% of enterprise large language model spending in the United States, Anthropic is the most quietly formidable player in the quintet. Amodei has also demonstrated a rare form of institutional courage: in February 2026, he refused a Pentagon demand to remove contractual prohibitions on Claude’s use for mass domestic surveillance, even as the Trump administration labeled Anthropic a “supply-chain risk” and ordered agencies to stop using the model. That is not the behavior of a man who has captured the state. It is the behavior of a man trying not to be captured by it.

Political Power: Proximity Is Not Capture

The AI leading men have achieved unprecedented proximity to political power. Altman donated to Trump’s inaugural fund, sat on San Francisco’s mayoral transition team, and has testified repeatedly before Congress. Musk, as an architect of the Department of Government Efficiency, has arguably achieved more direct influence over federal bureaucracy than any private citizen since Bernard Baruch. Zuckerberg has reoriented Meta’s content moderation in ways that reflect political calculation as much as principled policy.

And yet proximity is not capture. Rockefeller’s Standard Oil didn’t merely lobby regulators — it effectively set the regulatory agenda in oil-producing states for two decades. The steel and railroad barons didn’t just meet with senators; they funded them in ways that made legislative independence a legal fiction.

Today’s AI executives remain subject to forces their predecessors never faced. The European Union’s AI Act imposes binding constraints that no 19th-century robber baron ever encountered. Antitrust scrutiny from both the Department of Justice and the EU threatens the integration strategies of both Google DeepMind and Meta. Anthropic’s standoff with the Pentagon demonstrates that even the most safety-focused AI lab cannot escape the gravitational pull of geopolitical competition. The five men are powerful political actors — but they are actors on a stage with many more directors than Rockefeller ever faced.

The Cognition Economy: A New Kind of Monopoly Risk

Where the AI quintet is converging toward something genuinely Rockefellerian is in what might be called the cognition economy — the emerging marketplace where intelligence itself, not oil or steel, is the resource being extracted, refined, and sold.

Demis Hassabis, the Nobel Prize–winning CEO of Google DeepMind, said at Davos 2026 that today’s AI systems are “nowhere near” human-level AGI, placing the milestone at “five to ten years” away. Amodei, characteristically more bullish, has predicted that AI will reach “Nobel-level” scientific research capability within two years, and has described the coming AI cluster as “a country of geniuses in a data center” running at superhuman speeds. If either is even partially correct, the downstream consequences for labor markets, knowledge production, and institutional power are more profound than anything the Industrial Revolution generated.

The danger is not that one of these five men will own the world’s intelligence outright. It is that the economic logic of AI — massive upfront compute costs, proprietary training data, and compounding capability advantages — tends toward the same concentration dynamics that produced Standard Oil. A model that is marginally better attracts more users; more users generate more data; more data enables further improvement; the loop closes. This is not metaphor. Meta’s Llama 5, released in April 2026, was explicitly designed to commoditize proprietary AI — Zuckerberg’s theory being that if intelligence becomes free, the company that distributes it through 3.5 billion social media users wins by default. That is not so different from Rockefeller’s insight that the real money was never in the oil itself, but in making yourself indispensable to everyone who wanted to transport it.

Cultural Monopoly: The Unfinished Frontier

Henry Ford didn’t just build cars. He built a culture. The five-dollar day, the $40 workweek — Ford shaped how Americans understood the relationship between labor, leisure, and consumption. His prejudices, published in the Dearborn Independent and later praised by Adolf Hitler, exercised a cultural influence that no modern tech executive has approached, for better or for worse.

The AI quintet has, so far, produced nothing comparable to that kind of cultural ownership. ChatGPT is used by hundreds of millions, but it has not yet redefined the terms of civic life in the way that Ford’s assembly lines redefined time itself. The AI leading men give TED talks and publish essays — Amodei’s “Machines of Loving Grace” and its sequel “The Adolescence of Technology” are genuine intellectual contributions — but they have not yet built the durable cultural institutions that the Carnegies and Fords used to launder their economic power into social legitimacy. The Carnegie libraries are still standing. The Ford Foundation still funds democracy initiatives. What will Sam Altman’s equivalent be? We do not yet know.

This gap may close faster than we expect. If AI agents do begin displacing 50% of white-collar jobs — as Amodei and others predict within five years — the resulting social disruption will demand new cultural narratives. The men who shape those narratives will wield a form of power that makes their current wealth look like a down payment.

Why the Gap Matters — And Why It Is Narrowing

The distance between the AI tycoons of 2026 and the historical robber barons is real, but it is not permanent. Three trends are accelerating the convergence.

First, physical infrastructure is being built at unprecedented speed. Meta’s $600 billion data center pledge, Musk’s orbital computing vision, and the arms-race dynamics of semiconductor procurement are creating the structural lock-in that historically defines industrial monopoly. The company that owns the compute wins — not just the model race, but the infrastructure race.

Second, regulatory arbitrage is becoming a competitive strategy. Just as Rockefeller used the legal patchwork of late-19th-century interstate commerce to outmaneuver state-level regulators, AI companies are exploiting the gap between national regulatory frameworks to deploy capabilities that no single jurisdiction can constrain. The Trump administration’s rollback of Biden-era AI safety executive orders has already opened space for more aggressive deployment by American companies.

Third, the feedback loops of AI capability are compounding in ways that no previous technology has. When Anthropic’s own engineers have largely stopped writing code themselves — directing AI-generated code as product managers rather than authors — the productivity advantages of leading AI labs over their competitors begin to resemble Standard Oil’s pipeline advantages over independent refiners. Not yet identical. But structurally rhyming.

The View from 2035: A Question of Institutions

The most important distinction between Ford, Rockefeller, and today’s AI leading men may ultimately be institutional rather than technological. The Gilded Age tycoons operated in a world with weak antitrust frameworks, no administrative state to speak of, and a political economy that had not yet developed the tools to constrain concentrated private power. The Progressive Era — Teddy Roosevelt’s trust-busting, the Sherman Act, the eventual dissolution of Standard Oil — was the institutional response. It took a generation.

We may be at the beginning of a similar reckoning. Whether the five men who currently lead the AI revolution become as powerful as Ford or Rockefeller depends less on their own ambitions — which are extraordinary — than on the speed and coherence of the institutional response. Policymakers who wait for the infrastructure to be fully built before acting will find themselves in the same position as the regulators who confronted Standard Oil in 1911: arriving at the scene of a revolution already completed.

The AI leading men are not, today, as powerful as Rockefeller. But they are building the conditions under which someone very like them could be. That is the moment for executives, investors, and policymakers to pay attention — not when the resemblance is complete, but now, while the architecture is still under construction and the pipes have not yet been welded shut.