By Wang Jianfei

Senior Fellow, Tencent Research Institute | August 3, 2026

After 2025, "AI-driven restructuring" transitioned from speculative commentary into standard corporate disclosure language.

Historically, corporate layoff memos relied on polished corporate euphemisms: macroeconomic realignments, organizational optimization, portfolio streamlining, or strategic focus. Generative technology upended that playbook. Technology leaders—including Amazon, Meta, Block, and Klarna—began coupling workforce reductions directly with algorithmic efficiency gains.

By late 2025, US job reductions explicitly attributed to artificial intelligence surpassed 55,000 positions. Amazon eliminated approximately 30,000 corporate roles across two rounds, concentrating reductions within L5 to L7 middle-management tiers. Block took a more aggressive step, cutting 46% of its workforce while operating profitably.

By SOLOMOAT Editorial Team

Block is no peripheral startup; it is a mature public enterprise operating complex merchant systems, consumer payment rails, and balance-sheet lending.

Attributing headcount reductions to machine automation can also serve as corporate cover for past over-hiring or earnings management. Yet the adoption of this framing signals an institutional shift: a decade ago, replacing personnel with software invited public backlash; today, executives present it as strategic foresight.

This flattening alters corporate architecture. As CNBC noted regarding Amazon’s corporate restructuring, eliminating 14,000 management roles occurred alongside massive capital reallocation toward custom silicon, server facilities, and foundational compute clusters.

Data from payroll platform Gusto covering 8,500 small and mid-sized businesses indicates that the average manager's span of control doubled between 2019 and 2026. Across Microsoft, Alphabet, Meta, and Amazon, the "Great Flattening" compressed managerial hierarchies.

Firms did not automatically become smarter; they became leaner.

This dynamic spawned a popular narrative: AI augments the individual into a Super-Individual, while augmenting the enterprise into a Super-Organization. A single operator wields an agent fleet to match the yield of ten employees, while an enterprise coordinates agents to deliver the output of ten firms.

Industry commentators, including 360 Group founder Zhou Hongyi and Microsoft China CTO Wei Qing, have championed this vision, arguing that every professional should operate as a One-Person Company (OPC) by 2026.

This thesis overlooks a basic structural contradiction: if machine intelligence enables an individual to execute any operational workflow across an enterprise, why would that individual remain inside a corporate hierarchy?

By training personnel to operate as self-contained execution units, corporations are mass-producing their own direct competitors. The Super-Organization narrative claims that individual workers become exceptionally powerful while large corporations become more indispensable.

In reality, the forces that expand individual capability dismantle the economic rationale for large hierarchies. The two models cannibalize each other.

Displaced Talent Converts from Cost Reductions to Market Competitors

In industrial assembly lines, machinery automated physical routines, reducing interchangeable manual labor without creating immediate external competitors. A displaced line worker could not easily construct an independent automotive plant.

White-collar knowledge production operates under different unit economics. Writing code, designing interfaces, querying data, generating copy, building media assets, managing outbound marketing, and producing strategic research do not require heavy capital assets. Knowledge workers historically organized into teams because no single individual commanded the time, bandwidth, or cross-disciplinary dexterity to execute every stage alone.

AI agent frameworks provide cognitive leverage, enabling a single operator to complete an initial prototype, test scripts, and coordinate deployment in an afternoon.

Consequently, workforce reductions eliminate more than payroll liabilities; they release experienced operators who understand the incumbent's customer complaints, vendor dependencies, pricing models, and operational blind spots.

In past cycles, commercializing this tacit knowledge required raising seed capital, hiring engineers, building sales pipelines, and retaining operational staff. AI shortens this translation cycle.

A displaced middle manager can launch a focused micro-SaaS targeting a niche workflow overlooked by their former employer. The competitive threat to enterprise incumbents comes not merely from peer conglomerates, but from thousands of specialized solo operators attacking specific customer pain points.

Cursor did not rebuild Microsoft's operating system, enterprise sales network, or cloud division. Instead, it captured a critical node in developer workflows—code editing—and rapidly integrated frontier reasoning models.

Menlo Ventures’ 2025 State of Generative AI in the Enterprise noted that Cursor generated $200 million in revenue before hiring its first enterprise account executive.

Gamma executed an identical strategy against Microsoft Office, scaling to $100 million ARR with roughly 50 employees by focusing strictly on slide layout automation.

Harvey deployed 25,000 custom legal agents across more than 100,000 attorneys—including majorities across the AmLaw 100—automating document collation, contract reviews, and diligence pipelines that previously required cohorts of junior associates.

The Industrial Revolution centralized production because individual craftsmen could not fund steam engines or mechanical looms. Generative systems invert this dynamic by distributing operational capability back to the individual.

When a single operator can deliver the throughput of a traditional department, the line between an in-house employee and an external vendor dissolves.

Transaction Costs and the Shifting Boundary of the Firm

In his 1937 work The Nature of the Firm, Ronald Coase established that enterprises exist because market transactions incur search, negotiation, contracting, and enforcement costs. Hierarchies internalize these interactions, replacing continuous open-market bargaining with internal management commands.

When an operator can invoke model APIs to draft research, generate design assets, run statistical models, and route inbound support tickets, the rationale for maintaining dedicated internal departments recedes.

Capital allocations reflect this shift. Menlo Ventures reported that enterprise generative AI spending expanded from $11.5 billion in 2024 to $37.0 billion in 2025, with $19.0 billion allocated to the application layer.

Critically, 76% of enterprise AI deployments in 2025 involved purchasing external solutions rather than building in-house capabilities, up from 53% in 2024.

A parallel disruption is unfolding across Business Process Outsourcing (BPO). Valued by Andreessen Horowitz (a16z) at over $300 billion in 2024 and projected to reach $525 billion by 2030, the BPO sector organized human labor pools to handle repetitive customer support, billing reconciliations, and compliance operations.

Agent architectures unbundle these manual operations:

Decagon’s customer agents achieve resolution rates above 80% with improved customer satisfaction scores.

Camber’s healthcare revenue cycle models reduced first-time claim rejections by 80% and compressed billing administration cycles by 50%.

Enterprise capability theorists counter that corporations retain core advantages in risk absorption, institutional brand equity, and proprietary knowledge retention.

Yet modern organizations are increasingly codifying this internal knowledge into vectors, standardized prompts, and internal agent workflows. This shifts institutional memory from human employees to software systems, altering corporate dependence on internal labor.

Where AI systems exhibit reliability limits, operational retrenchment occurs. Gartner projects that through 2027, half of enterprises that cut staff in premature AI restructuring will rehire human personnel, while Forrester found 55% of employers expressed regret regarding initial workforce reductions.

Klarna re-opened hiring for human support personnel after automating 700 roles, reflecting the ongoing need for human judgment in complex customer edge cases.

Internal Fissuring: The Cellular Enterprise Model

If corporate hierarchies do not scale into super-organizations, they will likely transition into decentralized operating cells.

Rather than maintaining rigid functional divisions, large enterprises can fragment into autonomous operating units that share capital, brand equity, and compute infrastructure while competing directly in open markets.

Haier pioneered this approach through its Rendanheyi framework, splitting a 70,000-employee manufacturing corporation into over 4,000 self-governing micro-enterprises with independent decision rights and customer-funded compensation models.

Following Haier's acquisition of GE Appliances in 2016, this cellular structure helped generate 11% sales growth against a 1% broader market contraction.

In pre-AI environments, cellular decentralization carried high internal coordination and administrative overhead. Today, modular software handles accounting, CRM routing, marketing collateral, and compliance automatically, lowering the operational cost of managing independent enterprise cells.

A business unit that previously required thousands of personnel can be structured as a lean cell of domain experts directing autonomous software agents.

Attempts to govern decentralized systems purely through protocol code—such as Decentralized Autonomous Organizations (DAOs)—highlighted the difficulty of achieving high decision velocity, full decentralization, and consensus simultaneously.

As organizations flatten, administrative coordination does not vanish; it migrates to protocol design, infrastructure pricing, and algorithmic resource allocation.

Platform Centralization vs. Operational Decentralization

Decentralized market coordination requires reliable technical standards.

In software development, Git provides version control and branch reconciliation without corporate hierarchy. GitHub builds on this foundation by providing hosting, access permissions, automation, and discovery platforms.

GitHub does not employ the global open-source developer base or direct their daily hours, yet it commands platform power by managing the collaboration layer.

Future B2B interactions will likely follow this model: independent solo operators and micro-enterprises collaborating via standardized APIs, agent communication protocols, and programmatic escrow settlement.

The central nodes of this ecosystem—foundational model providers, compute hosts, payment rails, and discovery platforms—command leverage by managing infrastructure access.

This dynamic alters the cost structure of software applications. Traditional SaaS platforms enjoyed near-zero marginal costs at scale, with fixed hosting expenses amortizing as revenues grew.

By contrast, AI application providers incur marginal compute, data retrieval, and inference charges on every user interaction. As usage expands, upstream compute bills scale in tandem, leaving application margins vulnerable to changes in upstream API pricing or rate limits.

Concurrently, foundational model providers are moving downstream into user-facing applications, launching native workspace integrations, specialized reasoning agents, and enterprise tools that compete directly with third-party wrappers.

The Economic Realities of the Solo Enterprise

The rise of the solo operator presents an asymmetric tradeoff:

For experienced domain specialists, AI tools lower the cost of building software, reaching buyers, and fulfilling services.

As mechanical execution becomes commoditized, market value concentrates in non-transferable human capabilities: domain judgment, aesthetic taste, problem specification, enterprise trust, and accountability.

However, operating as an independent economic unit transfers market volatility directly onto the individual.

Inside a traditional enterprise, fixed salaries buffer employees from cyclical downturns and project failures. In an unbundled market, customer churn, algorithm revisions, API price increases, and legal liabilities fall directly on the individual founder.

Venture capital allocation is adjusting to these unit economics. Historically, institutional allocators prioritized organizational scale, hiring plans, and sales headcount.

Today, investors underwrite lean teams generating high revenue per employee: Gamma reaching $100 million ARR with 50 personnel, Cursor scaling past $200 million before deploying enterprise sales reps, and vertical platforms coordinating complex legal workflows through automated agents.

Investors are funding operational leverage rather than headcount expansion.

The Protocol-Driven Future

The narrative of the "Super-Organization" assumes that technology will simply scale legacy corporate structures.

Past technological waves—steam power, the assembly line, mainframe computing, and cloud infrastructure—initially disrupted corporate organizations before being integrated into larger, more centralized corporate forms.

Generative AI operates differently: it strengthens individual operators by externalizing execution capacity and embedding corporate processes into standardized software. It expands what an individual can build while lowering the barriers to operating outside a corporate hierarchy.

Heavy industrial manufacturing, resource extraction, physical transport, hardware engineering, and regulated utilities will continue to require centralized physical assets and formal corporate balance sheets.

Yet across digital services, software development, creative production, and specialized knowledge industries, the traditional corporate hierarchy is fragmenting.

The emerging market structure is not an ecosystem of bloated corporate giants, but an integrated network of lean operators, autonomous agent clusters, and specialized micro-firms coordinating via standardized APIs, payment rails, and foundation models.

The primary coordinating unit of the digital economy is shifting from the corporate org chart to the open protocol mesh.

Frequently Asked Questions

Does AI automatically make an organization more effective?

No. AI can increase individual output and reduce layers, but organization-wide performance still depends on coordination, decision rights, incentives, and a coherent operating model.

Why can AI empowerment challenge corporate hierarchies?

When a capable operator can independently execute more workflows, a hierarchy must offer strategic leverage and shared opportunity rather than merely controlling access to execution.