On 26 February 2026, Block CEO Jack Dorsey announced a 40% reduction in headcount, cutting the company’s workforce from more than 10,000 employees to just under 6,000. The move came despite strong performance. Gross profit rose 17% year over year, and guidance increased. Markets rewarded the move. The stock rose over 20% following the announcement. On the surface, the move appeared counterintuitive.

The explanation lies in what has already been happening inside the company. Behind the scenes, internal AI systems were quietly transforming the workspace. Dorsey’s internal AI coding agent, Goose, was reportedly saving engineers eight to ten hours a week. By the time the restructuring was announced, roughly 90% of new code submissions were being generated by AI systems. “Intelligence tools have changed what it means to build and run a company,” Dorsey told investors. “A significantly smaller team, using the tools we’re building, can do more and do it better.”

This is not an isolated case. Across industries, companies are beginning to operate under different constraints that differ from those that shaped traditional organisations. As access to information expands, coordination becomes easier, and execution accelerates. If the conditions that justified hierarchy are changing, does the structure itself still make sense in its current form?

For more than a century, the dominant model of corporate organisation has been the pyramid. At its narrow peak sits the CEO and a small circle of senior executives; beneath them layer after layer of middle managers and supervisors descend to the wide base of frontline employees. Long before modern corporations, in ancient Egypt, the construction of the pyramids itself relied on a similarly structured hierarchy: architects and planners at the top, overseers and coordinators in the middle, and a vast workforce at the base.

In such systems, decisions are concentrated at the upper levels and cascade downward through the organisation. Strategy, resource allocation, major investments, large-scale hiring, and tactical choices are handled by senior leadership and their closest teams. Lower levels receive directives with limited autonomy. This top-down flow reflects the foundations of industrial-era management, heavily influenced by military-style command structures and refined as corporations scaled throughout the twentieth century. 

Information asymmetry is the source of power. Senior leaders control the data pipelines, market forecasts, financial models, competitive intelligence, and long-term plans. Knowledge is then filtered and rationed as it moves down the pyramid. Those at the summit are seen as having the broadest view, and therefore the clearest right to decide.

This model, however, was built for a world where information was costly to process, and slow to circulate. Entire organisational layers emerged to manage this constraint. Analysts gathered data, managers consolidated it, and senior executives interpreted it before decisions were made. Each step added structure but also delay.

As highlighted by Jack Dorsey and Roelof Botha in their article “From Hierarchy to Intelligence”, modern enterprises remain deeply shaped by these legacy patterns, where coordination, reporting, and alignment are embedded in layers of management. In many cases, complexity itself became part of the system’s resilience. Organisations built entire workflows around complex enterprise software, often requiring dedicated teams to operate, maintain, and interpret them. Over time, fluency in these systems became a skill and even a career path. This created a self-reinforcing system. Complexity justified layers of coordination, and those layers in turn reinforced the need for complex tools and processes.

Yet cracks are beginning to appear. In recent years, several companies have reduced headcount, particularly in coordination and support roles, without a proportional decline in output. In some cases, productivity has remained stable or even improved. Major firms such as Amazon, Meta, Microsoft, and Oracle have all announced significant workforce cuts while continuing to scale operations and invest heavily in automation and AI-driven systems. According to Nikkei Asia, the tech sector alone cut nearly 80,000 jobs worldwide in the first quarter of 2026, with close to half of those reductions linked to automation and AI-related restructuring. 

The hierarchy pyramid was designed to manage scarcity of information, processing capacity, and coordination. As those constraints weaken, the structure begins to look heavier than necessary. 

Source: Bloomberg, Bureau of Labor statistics

The most important shift is not happening at the organisational level, but at the level of the individual contributor. Across industries and functions, a new type of worker is emerging: the “10x employee” one person equipped with AI tools who can now produce output that previously required an entire team. Evidence now shows measurable productivity gains from AI across industries. Developers using GitHub Copilot complete tasks about 55% faster, according to controlled GitHub studies, with the biggest improvements in complex, multi-step work. In finance, analysts using large language models are reducing multi-day due diligence and modelling processes to just hours, with firms like JPMorgan and BlackRock reporting major cuts in analyst time per memo. In marketing, generative AI is scaling content production by five to ten times without proportional headcount growth. Across all cases, AI does not replace judgment but removes execution friction, shifting the constraint from output capacity to quality of thinking and decision-making.

Source: McKinsey

A second major shift is structural, fundamentally reshaping how organisations operate. In traditional hierarchies, information flowed upward and decision-making authority concentrated at the top. Managers held value largely because they possessed superior context, broader visibility across data, functions, and organisational dynamics. AI impacts this informational asymmetry. With access to well-configured large language models and internal data, analysts can now perform synthesis tasks that once required senior managerial input. They can run scenarios, test assumptions, surface precedents, and generate well-reasoned recommendations without waiting for formal approval cycles or weekly review meetings.

Source: PwC

As a result, employees rely less on managers for data analysis and strategic input, two core managerial functions for decades. The role of management shifts from decision-maker to validator, with managers increasingly evaluating the quality of outputs produced faster and closer to execution. The WEF 2025 Future of Jobs report highlights this as a key transformation in knowledge work: a compression of decision latency driven not by organisational redesign, but by AI capability diffusing directly to individuals.

What replaces traditional hierarchies is a hierarchy of capability rather than position. Influence depends less on seniority or title and more on three factors: the ability to effectively leverage AI tools, the speed from question to output, and the quality of judgment used to direct and evaluate AI systems. Prompting and judgment become central. As AI grows more powerful, the only limiting factor is human directing it. A poorly framed question produces fluent but low-value output, while a precise, well-informed one generates meaningful insight. Bridging domain expertise with effective AI use is becoming a defining skill of the decade. Deloitte’s 2025 Global Human Capital Trends survey shows that organisations investing in structured AI fluency programs achieved 34% higher engagement and 22% lower attrition among top performers. For leadership, the performance cannot be measured mainly by output volume, but by the quality of judgment applied to highly productive systems. Organisations that adapt evaluation frameworks early will align incentives with this shift before outdated metrics create new inefficiencies.

Source: BCG

In a world built around AI, resilience is not just a technical issue but an evolving source of strength. By 2030, the organisations that succeed will be those that grasp how AI reshapes value creation by displacing traditional hierarchies and adjust their position to match this new reality.

The primary winners are companies that control the AI stack. Those who build foundational models, own proprietary training data, and manage delivery infrastructure capture disproportionate value. Firms like OpenAI, Google, and Microsoft act as “toll roads,” extracting ongoing economic rent through APIs, data dependence, and pricing power. Organisations relying on these platforms improve productivity but remain structurally dependent. Financial evidence reinforces this dynamic. Goldman Sachs estimates that the top AI infrastructure providers will capture around 40% of total AI economic value through 2030, leaving adopters in a highly competitive environment where gains are quickly eroded. Short-term margin improvements from AI adoption are not sustainable advantages, as competitors can access the same tools easily. Long-term financial winners are those that build proprietary data assets, develop tailored AI models, and retain partial infrastructure control. BCG (2025) shows such firms command a 30-40% valuation premium. For most companies, the strategy is not to compete with big tech, but to treat data as a balance sheet asset, embed AI deeply to create switching costs, and negotiate partnerships with clear awareness of dependency risks.

The second dimension is human capital. Management roles are shifting from supervision to orchestration, with direct financial implications. Traditional oversight tasks are being automated, effectively reducing their economic value. Firms investing in orchestration reduce operational errors and recover training costs quickly, while those that fail to redesign roles face talent loss and high replacement costs. Ultimately, strategy, judgment, and ethics remain human-driven and retaining this expertise is critical to sustaining financial resilience.