Neurodivergent Cognition and AI Collaboration in Data Center Operations

Four independent evidence domains converge on a finding with significant implications for the global data center industry: neurodivergent cognitive styles — particularly those associated with ADHD, the autism spectrum, and their frequent co-occurrence — are structurally well-suited to the AI-collaborative operational model toward which the industry is already converging. The field this paper names — Divergosynnoetics, the study and design of synergistic cognitive partnerships between neurodivergent minds and artificial cognitive systems — is an emerging one. This paper presents the convergent evidence from corporate outcome data (2015–2026), human–AI collaboration research (2024–2026), infrastructure labor economics, and global policy frameworks, and proposes adaptation pathways through which community-scale data center operations can implement these findings. Evidence is drawn from Asia-Pacific, Europe, and North America, with adaptation pathways relevant to jurisdictions including China, the European Union, Singapore, South Korea, and the United States. A values-alignment mechanism is identified through which environmental integrity converts from an ethical preference into workforce infrastructure: systemizing cognition, elevated in autistic populations, demands internal logical consistency between a facility's operational reality and its public claims. The paper identifies limitations — productivity figures are corporate self-reports, strengths are trait-associated rather than universal, and the values-alignment hypothesis requires controlled testing — and proposes a measurement framework through which adopting facilities generate the evidence the field currently lacks.

The data center industry is experiencing simultaneous crises of trust and talent. Community opposition has become a structural constraint on data center siting worldwide, recurring wherever facilities consume local power, water, and land while generating value elsewhere. The pattern spans grid-constraint moratoria in Europe (Amsterdam's "no, unless" policy), land and power conflicts across Asia — and a United States where the scale is quantified in unusually concrete terms: between May 2024 and June 2025, over $162 billion worth of projects were blocked or delayed (trade-press aggregate; CNBC, Data Center Dynamics, Reuters, 2024–2025). Meanwhile, 90% of operators cite staffing shortages as a critical constraint on expansion (JLL, 2025), and 40% of current professionals plan to leave despite rising salaries, citing burnout, inflexibility, and absent development (DataX Connect via Schneider Electric; CNBC, Mar 2026).

These two crises share a structural feature: both are, at root, crises of alignment. Communities oppose facilities that extract without contributing. Workers leave industries that pay well but mean little. The solution to both — genuinely green operations that serve host communities and provide intrinsically meaningful work — requires a workforce capable of deep engagement with AI-collaborative infrastructure operations and motivated by mission alignment rather than compensation alone.

This paper presents evidence that such a workforce exists, is measurably high-performing when correctly matched, and is systematically excluded by conventional hiring machinery. The evidence converges from four independent domains across three continents, each arriving at the same conclusion this paper advances — but from its own data, methods, and institutional perspective.

A note on terminology. Throughout this paper, neurodivergent describes a mind whose cognition — how it learns, focuses, senses, and processes — diverges from prevailing societal norms (including, but not limited to, ADHD, autism, and their co-occurrence); neurotypical describes a mind that falls within those prevailing norms. Each term names a position relative to a norm, not a defect or a gift. Neurodiversity names the natural variation in cognition across all minds; the neurodiversity paradigm holds that this variation is part of normal human diversity, not pathology. Neurodivergent names the minds that diverge from the prevailing norm. ADHD is used here as a widely used clinical label for one neurodivergent cognitive profile — a diagnostic label, not a defect or moral failing. The difficulties associated with such profiles arise substantially from environments designed without people with ADHD in mind: from spaces, schedules, and expectations built around a different way of thinking. Where systems accommodate divergent cognition, those difficulties recede. That shift is underway — but nowhere near the speed this new era invites. This paper contributes to an emerging field it names Divergosynnoetics (from Latin divergere, "to diverge," + synnoetics, "thinking together"): the study and design of synergistic cognitive partnerships between neurodivergent minds and artificial cognitive systems. The term braids two independent 1960s coinages of synnoetics.¹ Its central, falsifiable thesis is that such partnerships are conditional on values alignment: absent alignment between a system and its neurodivergent partner, synergy should not obtain.

¹ Synnoetics was coined independently in two 1960s lineages: Louis Fein's (1961) structural vision of human–machine joint mental power, and Philip Phenix's (1964) account of relational, tacit knowing.

Evidence & sourcing. This paper prefers primary sources — peer-reviewed research, government documents, and organizations' own published reports — and labels corporate self-reports as such rather than as peer review. It distinguishes exemplars from evidence: organizations held up as models of this work are profiled for their structure and outcomes; any source may be cited for its findings.

2.1 Scale of the Shortage

The data center industry faces a structural, global workforce shortage that is the most-cited constraint on AI infrastructure growth:

The retention crisis is particularly revealing when read against the neurodivergent profile identified in §3. Across the industry, workers are leaving despite rising salaries, seeking flexibility, development, and purpose. For this high-systemizing, justice-sensitive subset, the pattern runs deeper: engagement is sustained by meaning and values alignment — by whether the work is real and whether the organization's practice matches its word. A worker whose cognition runs on internal consistency will catch the gap between professed and practiced values the instant it appears — and will not invest in that organization; the industry is, in this sense, pushing away the very minds it most needs to retain. For this population, retention is an alignment problem — and alignment is something the industry can address at its root.

Regional distribution is equally structural. Uptime Institute's industry-first global staffing forecast projects that most demand will be in the Asia-Pacific region, followed by North America, Europe, the Middle East, and Africa (Uptime Institute, 2020). Anchoring the shortage by region. Where a figure is a commercial survey or trade-press aggregate rather than a government statistic or industry census, it is labeled as such:

These anchors span government (gov.cn, Eurostat, METI, Thailand's Digital Economy & Society ministry, Indonesia's Kominfo), industry association (Uptime, NASSCOM, MEXDC), and commercial research (NLB, Mordor) sources. The sharpest quantitative shortfalls tell the same story: China's 25–30 million digital-talent gap, the EU's ~9.6 million ICT-specialist deficit, Japan's ~450,000 IT gap, Indonesia's 9 million digital-talent need, and Mexico's 39% engineering deficit.

2.2 Geographic Immobility of Infrastructure Talent

Skilled infrastructure trades have very low geographic mobility. A new data center can instantly exhaust local talent pools. If talent cannot be imported at scale, it must be grown locally. Industry actors are already building community pipelines: Microsoft Datacenter Academy community-college partnerships, AWS Workforce Accelerator, Google STAR, and Per Scholas 16-week technician programs represent the infrastructure the industry is scrambling toward. What distinguishes the model proposed here is the additional design-layer of cognitive-friendly environments and AI-collaborative roles that open these pathways to populations currently excluded by conventional hiring.

3.1 Corporate Outcome Evidence

A decade of measurable results from neurodiversity hiring programs at major global firms documents consistent, large productivity and retention outcomes:

Additional structured programs span continents and industries: EY (Neurodiversity Center of Excellence — data analytics and cybersecurity), Goldman Sachs (with Specialisterne), Deloitte, PwC, Dell, IBM (30+ countries), Microsoft (skills-based hiring for Azure, Windows, Office), Google, and JLL.

Calibration: These productivity figures are corporate self-reports, not peer-reviewed controlled studies. Bury et al. (2020) noted a lack of controlled empirical evidence for a general "autistic advantage" in workplaces. What is peer-reviewed: the process-inefficiency-detection finding (§3.2), self-reported strength profiles, and the LLM communication-preference findings. The field lacks controlled workplace outcome studies — a gap the measurement framework in §8 is designed to close.

3.2 Peer-Reviewed Strengths

• Detail-focused anomaly detection: Peer-reviewed and workplace literature identifies attention to detail, pattern recognition, and detail-focused processing as commonly reported strengths of autistic workers (Cope & Remington, 2022; Bury et al., 2020) — strengths that map directly onto the anomaly-detection and quality-assurance roles prevalent in data-center operations, where corporate programs report the same pattern in practice.
• Strength profiles: Cope & Remington (2022) document autistic employees' self-reported workplace strengths in creativity, focus, attention to detail, task completion, and logical reasoning.
• Commercial validation: Testworks (Seoul) employs autistic teams to correct AI training data for autonomous vehicle perception systems, where sensitivity to minute errors raises overall data quality (PR Newswire, 2020; Stanford Social Innovation Review, 2024).
• Bias mitigation: Temple University research found neurodivergent annotators produced "diverse annotations that are valuable for employers," enriching training sets and mitigating bias, with autistic annotators showing consistent logic and less susceptibility to context drift (PYMNTS, Oct 2025).

3.3 The Structural Mismatch

The talent is available because conventional hiring systematically excludes it:

• Employment among autistic adults is starkly low: only 3 in 10 autistic people of working age are in employment — meaning 7 in 10 are unable to access the independence and fulfilment that employment can bring — a gap driven by systemic and environmental barriers, not by any inherent deficit in autistic people (Buckland Review of Autism Employment, UK Government, Feb 2024).
• 51% of neurodivergent people feel they cannot disclose at work due to stigma; two-thirds report employers treat neurodivergence as a red flag (Zurich UK survey via Forbes, 2026).
• Traditional interview-heavy hiring filters out this population before skills are ever assessed — which is why every successful program above replaced interviews with multi-day, hands-on, skills-demonstration formats.

The implication is structural: a facility that designs hiring around skill demonstration rather than interview performance accesses talent that conventional operators are filtering out.

3.4 Entrepreneurship Signal

People with ADHD are estimated to be up to 500% more likely to become entrepreneurs than the general population; 40% of self-made millionaires in the UK are dyslexic (research cited by Exceptional Individuals, via Forbes, 2026). Venture investors have begun treating neurodivergence as a signal rather than a coincidence. This is consistent with intrinsic-motivation-driven work selection: when external structures do not fit, this population builds its own.

3.5 A Note on Scope

ADHD and autism involve distinct — sometimes conflicting — cognitive profiles, sensory needs, and environmental preferences. What helps one person may challenge another — and even a single person's needs are not fixed: the same person may thrive in intense sensory input at one moment and need quiet at the next, depending on task, goal, and context. Structure and variation are both resources to draw on, in whatever mix the person and the task call for — not a single fixed setting. This framework does not treat "neurodivergent" as a monolith.

ADHD and autism also frequently co-occur. Estimates of ADHD prevalence among autistic individuals range from approximately 30% to over 80%, depending on diagnostic threshold and population (Hours, Recasens, & Baleyte, 2022; Simonoff et al., 2008; Salazar et al., 2015). This co-occurrence is not additive but configurational: the same individual simultaneously navigates novelty-seeking (ADHD-associated) and routine-dependence (autism-associated) cognitive drives. In the community this co-occurrence is commonly called AuDHD — a term not yet in the official diagnostic manuals, but one the community uses for itself. The resulting profile is distinct from either condition alone, and the modular, individually-calibrated approach described below is specifically designed to accommodate it — neither forcing a single template nor assuming traits are separable in practice.

Where possible, it uses trait-based profiles rather than diagnostic labels (e.g., "workers with high sensory sensitivity," "workers who thrive on non-linear workflows"), and its workplace and AI-collaboration recommendations are designed as a modular menu, not a single prescriptive environment. The skills-demonstration hiring model is individually calibrated by design.

4.1 Government and Institutional Findings

• UK Department for Business and Trade: in an evaluation of an AI assistant pilot (Microsoft 365 Copilot), respondents self-identifying as neurodiverse were more likely than other respondents to recommend the tool (p = 0.035) and more likely to report satisfaction (p = 0.100, marginal).
• EY research: generative AI reduces barriers and supports more inclusive ways of working — significant for a workforce where an estimated 15–20% of people are neurodivergent (Microsoft Source, Jan 2026).
• McKinsey/Microsoft convergent finding: generative AI raises knowledge-task productivity by up to ~40%, but the gain is largest for workers whose bottleneck was format, not capability — AI disproportionately unlocks people for whom structuring and formatting information was the barrier (Enna Global, Jun 2026).
• Revenue signal: organizations prioritizing neurodiversity inclusion have been associated with revenue increases approaching 20% (research cited in CNBC/SAS, 2025).

4.2 Peer-Reviewed and Preprint Research

• Jang, Moharana, Carrington & Begel (2024): autistic workers seeking communication support strongly preferred LLM-generated responses over human-written messages.
• ADHD-aware human-in-the-loop frameworks (arXiv, Jul 2025): survey-grounded system design documenting that executive-function support — prioritizing, initiating, organizing — is precisely where AI scaffolding produces the largest functional gains. Voice interaction, soft adaptive nudging, and privacy-first local sensing are identified as key design features.
• Tool lineage: TwIPS (Haroon & Dogar, 2024) — LLM-based texting assistant adapting to individual user style; LaMPost (Goodman et al., 2022) — email assistance for dyslexic users.

4.3 The Division of Labor

The division is natural and evidenced. AI carries linear organization: scheduling, formatting, documentation, summarization — the tasks that have historically been the barrier to performance, not the job. The human carries non-linear association: anomaly judgment, novel problem-framing, final approval — the actual cognitive work.

This maps directly onto the industry's stated strategy for AI-augmented data center operations. The industry consensus is that bounded AI agents propose actions — monitoring, alerting, recommending — while humans retain final approval for all physical-impact decisions (Gartner Manufacturing Predicts 2026; arXiv 2602.10900, 2026). The industry has converged on this architecture without yet asking which humans are best suited to the non-routine half. The evidence presented here answers that question.

4.4 Caveats

Researchers have flagged legitimate concerns. Sinha & Sinha (2026) critique "neuronormativity" in the workplace — tools and expectations designed around neurotypical cognitive assumptions that suppress neurodivergent value. Dell'Acqua et al. (2023) found AI can diminish performance on tasks beyond its capability, arguing against universal AI overlay. There is clinician concern about conformity pressure from over-reliance on LLM-mediated communication. These caveats argue not against AI collaboration with neurodivergent workers, but for designed collaboration: deliberate division of labor, individual adaptation, and human-in-the-loop structures — the model described here, not "add a generic AI assistant to everyone's desktop."

5.1 The Cognitive Mechanism

The claim that environmental integrity functions as workforce infrastructure — not as an ethical preference but as a cognitive requirement — rests on converging evidence from multiple research traditions.

Systemizing cognition — the drive to analyze, construct, and understand rule-based systems — is a well-documented trait elevated in autistic populations (Baron-Cohen et al., multiple studies). It is not about interpersonal grievance; it is about how systems work and whether they are internally consistent. This cognitive machinery operates on any rule-governed domain: a power usage effectiveness (PUE) deviation, a cooling inefficiency, and a sustainability report that contradicts a utility bill are all detected by the same mechanism.

Justice sensitivity is empirically elevated in ADHD populations. Schäfer & Kraneburg (2015) found participants with ADHD reported significantly higher justice sensitivity on the observer and profiteer dimensions, with no significant difference on perpetrator sensitivity, while the inattentive subtype scored highest across all dimensions. The elevated dimensions are specific: observer sensitivity (distress at witnessing injustice done to others) and profiteer sensitivity (distress at benefiting from injustice) — not grievance about being wronged, but distress at witnessing wrong and at benefiting from wrong.

Autism and moral judgment: autistic participants are more likely than neurotypical participants to act morally even when behaving immorally would privately benefit them — experimentally, autistic adults rejected dishonest self-benefiting offers more often than controls (Hu et al., 2020).

For individuals with co-occurring ADHD and autism, we propose that systemizing cognition and justice sensitivity may operate jointly — theoretically creating a drive to detect systemic inconsistency compounded by heightened distress at its presence.

5.2 The Occupational Health Mechanism

The cost of values misalignment is measurable:

• A 500-nurse study found perceived organizational justice inversely related to moral distress, while moral sensitivity was directly related to moral distress — the most morally sensitive people working in the least just organizations carry the heaviest distress load, and moral distress feeds burnout (BMC Nursing, PMC10559416).
• Autistic burnout is defined in the research literature (Raymaker et al.) as pervasive exhaustion, loss of function, and withdrawal resulting from continuously expending more resources than one has coping with environments ill-suited to one's needs.
• Camouflaging (masking authentic responses to fit workplace norms) is a documented risk factor for mental-health difficulties in autistic people: a meta-analysis of 16 studies and 5,897 autistic participants finds camouflaging most strongly correlated with depression (r = 0.36), though the underlying studies were often underpowered (Khudiakova et al., 2024); burnout-exhaustion partially mediates the path from camouflage to depression (Benatov et al., 2026). An employee who privately believes the operation's green claims are false, but must perform belief, is camouflaging at the values level.

5.3 The Synthesis

Systemizing cognition demands internal logical consistency. When deep, daily engagement with optimized, data-driven operational reality conflicts with an employer's external greenwashing, the resulting cognitive dissonance is not a preference — it is a cognitive incompatibility. The same cognitive machinery that detects anomalies in operational data detects the gap between a sustainability report and a utility bill. For a workforce selected for pattern recognition, greenwashing is unsustainable as information — it will be detected.

The inverse is equally supported. In genuinely green, transparently measured, community-serving operations, the dissonance is eliminated. The same systemizing drive that would have detected inconsistency converts into meticulous care for energy integrity, honest reporting, and community outcomes. The auticon literature characterizes its consultants' defining traits as "precision, rigour and analytical depth" — when pointed at a mission the operator believes in, these compound into engagement and output quality exceeding what conventional incentive models produce.

Caveat: The extension of justice sensitivity and systemizing cognition to environmental and organizational integrity domains is a reasoned inference grounded in converging evidence, not a controlled experimental finding. Two analytically separable hypotheses follow. First, that verified environmental integrity correlates with retention and wellbeing among neurodivergent operators working in AI-paired roles. Second — anchored in an observed finding — neurodivergent operators report a measurably stronger preference for AI assistance: in the UK Department for Business and Trade's evaluation they were significantly more likely to recommend the AI assistant (p = 0.035) and reported marginally higher satisfaction (p = 0.100). We propose the mechanism is that AI pairing lowers the cost of cross-neurotype information transfer, extending Crompton et al. (2020) to an AI-mediated channel no study has yet directly tested. Because an AI intermediary could instead automate masking, calibration to the operator rather than to neurotypical norms becomes an essential design constraint. The measurement framework in §8 is designed to produce the answer.

5.4 The Logical Chain

1. Systemizing cognition is significantly elevated in autistic populations (Baron-Cohen et al., multiple studies), driven by understanding system rules and detecting internal contradictions.
2. Justice sensitivity is significantly elevated in ADHD populations (Schäfer & Kraneburg, 2015), especially the distress of witnessing injustice and of benefiting from it.
3. Systemizing cognition and justice sensitivity jointly produce an executable preference for organizational integrity: the ability to detect greenwashing combined with an unwillingness to tolerate it or profit from it.
4. When operational integrity aligns with reality, the same mechanisms convert into meticulous care: lean energy optimization, scrupulous reporting, sustained engagement.
5. When misaligned — greenwashing, resource extraction, community neglect — the mechanism inverts: cognitive dissonance → moral distress → burnout → attrition. The industry's own attrition data (40%, despite rising salaries) is highly consistent with this pattern.

5.5 Dual Resonance

Green computing's core values — sustainability, systems thinking, community wellbeing — resonate with the intrinsic motivations of many neurodivergent individuals at two levels simultaneously: cognitive and values-based. This is not a single-dimension match but a dual resonance: the way the mind works and what the heart cares about aligning at once. Finding such talent is not only about finding the right "brain" but also finding fellow travelers of the mission.

Of course, in any specific local context, economic incentives and job security remain non-negotiable baseline conditions.

6.1 China

China represents the largest-scale alignment between inclusive employment policy and computing infrastructure buildout.

Policy framework: - State Council 《促进残疾人就业三年行动方案（2025—2027年）》 (Three-Year Action Plan for Promoting Employment of Persons with Disabilities) explicitly supports 全链式融合就业服务 — full-chain integrated employment services — for autistic and cognitively diverse groups (gov.cn, June 2025). The first three-year action (2022–2024, 国办发〔2022〕6号) exceeded its 1-million target, achieving 1.648 million new urban/rural disabled jobs and raising the employment rate by 4.8 points (CDPF via People's Daily, Jul 2025). - Nationally, 8.915 million people with disability certificates were employed as of end-2025 (持证残疾人就业; CDPF《2025年残疾人事业发展统计公报》, Apr 2026) — a talent pool the technical-infrastructure career lane has yet to tap. - 《孤独症儿童关爱促进行动实施方案（2024—2028年）》 builds full-lifecycle support including vocational programs adapted to autistic students' learning characteristics and market demand (CDPF). - The 15th Five-Year Plan (十五五) lists special education and disability services among its 109 major engineering projects (Xinhua, Apr 2026). - Provincial programs: Jiangsu (2025–2028, full-lifecycle), Zhejiang, Anhui, and Shenzhen — whose implementation plan supports capable autistic individuals entering government agencies, public institutions, and state-owned enterprises (Legal Daily, Jun 2025). - Legal scholars propose a 四位一体 (four-in-one) integrated employment system with a graduated employment transition mechanism — preparatory → supported → stable employment (Legal Daily, 2025).

Infrastructure context: China's total computing power reached 962 EFlops (FP32) as of June 2025 — roughly 21% of global capacity, up 73% year-over-year — with intelligent compute at 782 EFlops (+96%) and 10.85 million standard racks in service (CAICT《先进计算暨算力发展指数蓝皮书（2025年）》, Mar 2026). Under 东数西算 (East Data West Computing), STL Partners' network-edge capacity forecast projected that by 2026 China would host ~26% of global network edge sites — the world's largest share (Osborne, STL Partners). 东数西算 has driven over ¥1 trillion in cumulative social investment (National Data Administration / CCTV, 2025), with the eight national hub nodes alone attracting ¥43.5 billion in direct and ¥200 billion+ in induced investment by mid-2024 (Xinhua, Aug 2024). The edge deployment model maps structurally onto the community-scale facilities proposed here.

The gap: Current autism employment programming is concentrated in 辅助性就业 (assisted employment) — crafts, agriculture, service roles. The policy scaffolding for high-skill technical pathways exists on paper; the technical-infrastructure career lane is essentially unbuilt. Connecting the inclusive-employment policy ambition to the computing-infrastructure buildout is implementable under current policy — no new legislation required.

Cultural context: Stigma remains a significant barrier; autism is still colloquially framed as 孤独症 ("loneliness disorder"), and most autistic children are educated outside mainstream schools (Autism Spectrum News, 2025). Advocacy organizations have shifted strategy from awareness to 社会融合 (social integration) — supporting institutions to adapt rather than asking individuals to disclose and request. Universal-design framing (environments good for everyone) aligns with this strategy direction.

6.2 European Union

Four of the jurisdictions in this section operate inside a single supranational framework. That framework gives the European entries a shared implementation floor which national law then builds upon, and it is the level at which a facility's obligations begin.

Framework: The European Commission's Union of Equality: Strategy for the Rights of Persons with Disabilities 2021–2030, adopted 3 March 2021, sets a ten-year programme spanning employment, accessibility and independent living. Its Disability Employment Package (2022) is the employment-specific instrument, pairing guidance for national employment services with a practical employer toolkit. Supporting flagships run continuously through the decade: the Disability Platform (2021), the AccessibleEU resource centre (2022), the European Disability Card and Guidance on Independent Living (2023), and a Framework for Social Services of Excellence (2024). The Commission reviewed the Strategy at its midpoint in the Communication Enhancing the strategy for the rights of persons with disabilities up to 2030, adopted 6 May 2026, which commits among other actions to strengthening implementation of the Disability Employment Package and to improving access to assistive technologies and AI.

Scale: The Commission puts the number of people in the EU living with a disability at around 90 million, and the 2021 Strategy identified 42.8 million of working age. Only half of persons with disabilities are in employment, against three in four of those without — a gap Eurostat measured at 24.2 percentage points in 2025. That population is the addressable workforce behind every member-state pathway below.

What this means for a facility: A community-scale data center sited anywhere in the Union inherits the Strategy's direction of travel before any national requirement applies. Germany's EnEfG, the Netherlands' "no, unless" test, Denmark's district-heating precedent and Bulgaria's employment quota are national expressions of a common floor — which means a methodology proven in one member state has a shorter path into the others than it would across unrelated jurisdictions.

6.3 Germany

Germany offers the world's strongest pairing of binding green data center law and proven neurodivergent-IT employment model.

Green DC law: The Energy Efficiency Act (EnEfG, in force November 2023) imposes PUE ≤ 1.2 for new data centers from July 1, 2026, staged waste-heat reuse requirements (10% → 15% → 20%), 100% renewable electricity from 2027, and mandatory public efficiency-register reporting (German Datacenter Association; TÜV Rheinland). A June 2026 cabinet-draft amendment is not yet law and is politically contested; the July 1, 2026 new-build PUE 1.2 deadline stands.

Neurodivergent employment: auticon, founded Berlin 2011 — the world's first company exclusively employing autistic adults as IT consultants. Per its 2024 impact report: 552 employees across 15 countries, 79% neurodivergent, 405 autistic people in technology and data roles, 307 global clients including BMW and Allianz. 90% of autistic hires were unemployed before joining. Current engagements include AI implementation at scale, large-scale financial-services data migration, and mission-critical data annotation.

6.4 Netherlands

The Netherlands codifies the strongest "community benefit" test for data center existence in Europe, alongside one of the continent's most mature neurodivergent IT employment pipelines — the same green-law-plus-employment pairing seen in Germany.

Green DC ("no, unless"): Amsterdam's data center policy operates on an explicit "no, unless" (nee, tenzij) basis — new data centers are refused unless they (a) benefit Amsterdam and (b) cause no additional grid congestion. After the 2019 moratorium and December 2023 restrictions, the city announced a moratorium on new data centers until 2035, and nationally hyperscale projects of ≥70 MW or ≥10 hectares are banned in most of the Netherlands (PropertyNL, Jan 2025; DCD, Jun 2026). "Community benefit" is thus already a legal test for data center existence here.

Neurodivergent employment: ITvitae (founded 2013, Amersfoort, with a new Eindhoven location) is a registered Social Enterprise (ANBI) training and placing autistic and gifted ICT specialists in software development, cybersecurity, and data science through ~9-month programs, with partners including Sogeti and Northwave. The numbers that motivate it: 54% of autistic people with higher IQ are unemployed in the Netherlands — over 20,000 unemployed autistic talents at university level or above (itvitae.nl). The 2023 auticon–Unicus merger created the world's largest autistic-majority company, operating across 15 countries including the Netherlands.

Industry-side labor gap: The Dutch Data Center Association's own labor-market study (Pb7 Research, 2021) quantifies the demand this pipeline serves: direct data center employment grows ~8% annually, from ~5,000 (2021) to more than 7,100 (2026), with over 600 open vacancies at the time of study — and only 3% of technical staff are women. The study explicitly points to untapped talent pools (veterans, status holders, migration-background youth) as the answer to the sector's structural shortage; the 2024 State of the Dutch Data Centers reports 70% of operators citing personnel shortage as a major problem, up from 45% in 2023 (DDA; AGConnect, 2024).

6.5 Denmark

Denmark is both the origin point of the modern neurodiversity-employment movement and the clearest existing proof that green data centers can benefit their actual neighbors.

Origin: Specialisterne was founded in Denmark in 2004 by Thorkil Sonne after his son's autism diagnosis. By the organization's own count, more than 10,000 autistic people have since been placed in competitive jobs globally — the corporate programs and pipelines documented throughout this paper are downstream of that one act.

Green, done right: Denmark also demonstrates green data-center benefit in its most physical, least-offsetable form. Meta's Odense campus sends server waste heat into municipal district heating for up to about 9,000 homes, free of charge. Microsoft's Høje-Taastrup site, with the utility VEKS, warms about 6,000 homes. atNorth and Vestforbrænding in Copenhagen will add more than 8,000 from 2028. This is thermal, adjacent benefit — the exact opposite of a claim that benefits some other community far away from the actual problem.

6.6 Bulgaria

Bulgaria illustrates the EU's mandatory-quota approach to disability employment alongside an early neurodivergent-IT footprint that has not yet been studied. The Persons with Disabilities Act (Закон за хората с увреждания, in force 2019, Art. 38) obliges employers of 50–99 people to hire at least one person with a permanent disability, and employers of 100+ to maintain a 2% share of average headcount (European Commission; Disability:IN). DXC's Dandelion Program — the Asia-Pacific flagship neurodiversity employment initiative — has expanded into Bulgaria, yet no peer-reviewed research yet connects Bulgarian neurodivergent employment to the country's data center sector. The finding here is the scarcity itself: Bulgaria is a named site for generating the missing evidence.

6.7 Japan

Japan's Ministry of Economy, Trade and Industry (METI) maintains an official neurodiversity promotion policy — a national economic ministry, not a welfare ministry, framing neurodivergent traits as valuable workforce attributes. The framing is economically motivated: a domestic labor shortage intensified by population aging makes neurodivergent talent a structural necessity rather than a diversity initiative. The rationale is explicit in METI's own materials: its survey projects a maximum IT talent shortage of ~790,000 by 2030 (demand ~1.92 million against supply ~1.13 million; the main scenario projects ~450,000), and the ministry states that it focuses its neurodiversity work on the digital field because "securing talent has become an urgent issue for the government" (METI, 2019; 2022). Beyond assertion, METI's commissioned field study (8 companies, 40 teams, 198 respondents — roughly half with disabilities, over 70% of them ASD or ADHD) measured productivity directly and identified three intervention points: job proficiency, physical environment, and, most influential, psychological safety — enabled in turn by humble leadership that lets team members disclose weakness and seek help (METI, 2022).

6.8 Singapore

Singapore is the one jurisdiction where formal green data center standards and neurodiversity-plus-AI research already coexist — with nothing yet connecting them.

Green DC (formalized, binding): - Green Data Centre Roadmap (IMDA, May 2024): allocates at least 300 MW of additional capacity with further capacity conditional on green energy use. - Green Mark for Data Centres 2024 (BCA-IMDA joint certification) rates operators on sustainability performance. - SS 715:2025 — among the world's first tropical-DC IT-equipment standards — targets ≥30% energy reduction and mandates safe operation at up to 35°C (IMDA, Aug 2025).

Neurodiversity + AI (research-active, employment-gapped): - Key paper: "Understanding the workplace needs of autistic adults in Singapore" (Frontiers in Psychiatry, 19 May 2026) explicitly concludes that designing LLM tools aligned with neurodivergent working styles can complement systemic inclusivity efforts. It flags the risk that off-the-shelf models may reflect neurotypical norms and reinforce masking, and notes Singapore's underrepresentation in Western-dominated literature. - Employment gap: persons with disabilities are 3.4% of Singapore's resident population but only 0.1% of the private-sector workforce (Autism Enabling Masterplan).

6.9 South Korea

Testworks (Seoul, founded 2015) — a high-tech social enterprise specializing in AI training data and software testing — represents the closest existing analogue to the model proposed here. In partnership with autonomous-vehicle firm StradVision, Testworks' autistic team corrects labeling errors in deep-learning training data for road-safety systems. Founder Dale Yoon: autistic individuals have a "unique sensitivity that allows them to notice small things that non-autistic people would simply overlook, including minute errors" — safety-critical AI data work, commercially operating in Asia, validating the anomaly-detection thesis.

Ecosystem depth: Autistar (autistic designers, Samsung Electronics), Donggubat ($10M revenue, majority-disabled workforce) demonstrate a Korean social-enterprise sector already proving the commercial model (Stanford Social Innovation Review, 2025).

6.10 Malaysia

Malaysia is the clearest case in this section of a computing sector being built and an employment mechanism being determined at the same moment — the rare condition under which design-time rather than retrofit implementation is actually available.

Buildout: Johor, adjacent to the Singapore corridor described in §6.8, held 1,110 MW of operational IT capacity in the first half of 2026 — third in Asia-Pacific behind Tokyo and Singapore, and effectively full, with colocation vacancy at 0.7% against Singapore's 4.9%. Two research houses report that operational figure independently (Cushman & Wakefield, APAC Data Centre H1 2026 Update; Knight Frank, Data Centre Atlas 2026). A further 602 MW is under construction and 2,486 MW planned. Separately — and measuring a different quantity — the Ministry of Energy Transition and Water Transformation told Parliament in July 2026 that the national utility had contracted approximately 4.5 GW of electricity supply to 36 operating data centers as of the first quarter, with 23 more under construction. That is grid supply rather than IT load, and larger for that reason: it covers cooling and facility overhead, and is contracted ahead of energisation. The ministry projects data-center electricity at 31% of national demand by 2035, against 7% in 2026. Dr Jade Vichyanond, a senior economist at the ASEAN+3 Macroeconomic Research Office writing in her personal capacity, frames the same arc in her signed opinion column "Malaysia's Data Center Boom: From Investment Surge to Sustainable Growth" — treating power and water constraints as design conditions, not retrofit burdens (2026).

Employment framework: the Persons with Disabilities Act 2008, sections 29 to 32, establishes the right of persons with disabilities to employment on an equal basis. Legal scholarship observes that the Act attaches no penalty provision for non-compliance — Md Tah (2013) names this the "rights-without-remedy paradox" — so the mechanism by which that right is realised in practice remains an open design question rather than a settled one.

The pathway: that openness is the opportunity. Every other jurisdiction in this section is fitting inclusion onto facilities and compliance systems that already exist. Malaysia is deciding both at once, at the exact moment the region's fastest buildout is under construction — the same design-time advantage this paper argues community-scale facilities hold over hyperscale retrofit, available here at national scale.

6.11 Thailand

Thailand has one of Southeast Asia's largest computing pipelines and an employment statute with an unusual structure: it offers three distinct routes to compliance rather than a single obligation.

Employment framework: the Persons with Disabilities Empowerment Act B.E. 2550 (2007), amended in 2013, requires one employee with a disability per 100 employees, and one further per additional 50 (§33). Employers may instead contribute to the Fund for the Empowerment of Persons with Disabilities (§34), or engage persons with disabilities through subcontracting or specialised employment services on annual contracts (§35). The three-route design matters for a facility. Section 35 in particular permits a structured relationship with a specialist employment provider — the model Specialisterne and Testworks operate elsewhere in this paper — rather than requiring direct hiring alone. UNDP examined implementation of the Act in 2023.

Buildout: pipeline capacity stood at approximately 2.87 GW as of September 2025, concentrated in the Eastern Economic Corridor provinces of Chonburi, Rayong and Chachoengsao. On 6 May 2026 the Board of Investment approved three data-center and data-hosting projects worth THB 913 billion, within a six-project round totalling THB 958 billion — the largest a THB 842 billion data infrastructure expansion (BOI Press Release No. 67/2569). Domestic coverage identifies workforce, alongside grid and regulation, as the constraints this buildout tests.

Evidence gap: no Thailand-specific research on autistic or neurodivergent employment was located for this paper; the statute is disability-general. Thailand is named here as a site where that evidence could be generated. A facility operating under §35 with a specialist provider would, in the ordinary course of doing so, produce exactly the data the field currently lacks.

6.12 Indonesia

Indonesia is building data center capacity under an explicit sovereignty-and-sustainability framing, alongside a disability employment statute that covers both public and private sectors and against which representation is actually being measured.

Employment framework: Law No. 8 of 2016 on Persons with Disabilities requires 2% of the workforce in government, regional government and state-owned enterprises, and at least 1% in private establishments (Art. 53). It carries reasonable-accommodation duties and penalties for non-compliance, and Indonesian courts have affirmed the work rights it establishes. Indonesian peer-reviewed scholarship has measured representation against that mandate and reports figures below target in state-owned banking. The measurement itself is notable: most jurisdictions in this section have a mandate without published measurement of how far it reaches. Indonesia has both — which makes the distance between requirement and practice a known quantity rather than an assumed one, and a known quantity is something support infrastructure can be designed against.

Buildout: national capacity was approximately 500 MW in 2025 and is projected at around 900 MW in 2026. Digital Edge is developing its USD 4.5 billion, 500 MW AI-ready CGK Campus in the GIIC industrial estate at Bekasi, east of Jakarta, designed to scale to 1 GW; three initial buildings are scheduled ready-for-service in Q4 2026, Q1 2027 and Q2 2027, targeting an annualised PUE of 1.25 with direct-to-chip liquid cooling and recycled-water systems. Komdigi positions the national data center ecosystem as a strategic digital asset, prioritising data sovereignty alongside green data center principles and the distribution of capacity beyond Java.

The pathway: that stated policy of distributing capacity beyond Java is the opening. A facility sited outside the Jakarta concentration is, by definition, a community-scale facility in a region with a labour market of its own — the configuration this paper argues is most favourable to design-time implementation.

6.13 India

India pairs one of the world's fastest data center buildouts with a statute that names autism explicitly, a population-based prevalence study of unusual rigour, and a documented shortage concentrated in exactly the roles this paper argues neurodivergent cognition suits.

Prevalence, measured: Arora and colleagues assessed 3,964 children aged 2 to 9 across five geographically and economically distinct sites — Kangra, Dhenkanal, Palwal, North Goa and Hyderabad. They found 475 children, 12.0% (95% CI 11.0%–13.0%), with at least one neurodevelopmental disorder (PLOS Medicine, 2018). The authors note that India's 2011 census recorded all disabilities at 1.1% for ages 0–4 and 1.5% for ages 5–9 — almost ten times less than the population-based estimate. The gap between administrative record and measured prevalence is itself the finding, and it is the same gap this paper describes in hiring.

Legal framework: the Rights of Persons with Disabilities Act, 2016 came into force on 19 April 2017, with the RPwD Rules notified on 15 June 2017. It recognises twenty-one categories of disability, adding autism explicitly, and reserves 4% of government-sector vacancies — of which 1% covers locomotor disability, autism, intellectual disability and multiple disabilities together. The reservation applies to government establishments; private establishments operate under incentives and tax benefits rather than a quota, which is why the Indian private-sector pathway has developed through industry rather than mandate.

The shortage, and where it sits: India's total data center capacity crossed 1,700 MW in 2025, with operational IT load at roughly 1.3–1.53 GW in early 2026 and a further 500 MW of supply expected during 2026 — an increase of about 30% year on year (CBRE; corroborated by Colliers). Mumbai holds around half of operational supply. The workforce picture is more subtle than a simple headcount gap. The NIIT India Skills Gap Report 2026 concludes that India does not face a skills shortage but a readiness and progression failure: mid-career talent with six to fifteen years of experience is the scarcest pool, cited by 38% of recruiters, and the binding constraint is access at scale rather than motivation. The Quess Corp report India's Data Centre Decade (Q3 FY26) quantifies the shortfall directly: IT Operations — the sector's largest job family, holding 44–46% of an ecosystem of 86,000 to 90,000 professionals across more than 50 providers — faces a 73% shortage, with core operations at 35% demand versus 16% supply (Supply Sufficiency Index 46) — a 19-point operational staffing gap — while AI Operations records a Supply Sufficiency Index of 47, demand (30%) outpacing supply (14%).

Existing machinery: FutureSkills Prime, a collaboration between the Ministry of Electronics and Information Technology and NASSCOM, has registered more than 3.4 million learners in emerging-technology upskilling, of whom 1.3 million have certified — and 86% come from tier-two and tier-three cities rather than the metros. That distribution matters: the skilling channel already reaches the places where community-scale facilities would be sited. India is also an origin point for the corporate movement documented throughout this paper: SAP's Autism at Work began as a volunteer-driven initiative at SAP Labs India in 2011. A 2025 qualitative study of seventeen clinically diagnosed neurodivergent professionals across Indian sectors documents the distance that remains between inclusion policy and daily experience (Career Development International).

The pathway: the skilling infrastructure, the statutory recognition and the measured need are all present. What is missing is a channel connecting them to data center operations — and the shortage figure identifies precisely which roles that channel should target first.

6.14 Mexico

Mexico pairs a severe data-center workforce shortage with a serious, peer-reviewed research base — and a front-line urgency case in Querétaro.

Workforce shortage: The Mexican Data Center Association (MEXDC) reports a 39% deficit in electrical/mechanical engineers and 32% in data-center operations, with 67% of firms unable to fill vacancies (MEXDC Data Center Profile Report, Feb 2026). Roughly 100,000 new jobs are projected over the next decade, with Querétaro the flagship hub.

Research base: Camprubí & Vega (2023, Revista Digital Universitaria 24(6), UNAM; DOI 10.22201/cuaieed.16076079e.2023.24.6.3) argue for organizational accommodation — not special treatment — as the path to unlocking neurodivergent productivity and integrity. INEGI's ENADID 2023 data place economic participation at 40.6% for people with disabilities against 68.0% without. Fombonne et al. (2016) established the first autism prevalence estimate for Mexico (0.87%). Monica Tentori's Veritas lab at CICESE (roughly 120 publications across CHI/UbiComp) anchors a world-class Mexican human-computer-interaction research lineage.

The front line: Querétaro is the urgency case — a dozen data centres already operating with plans for up to ten more, during the worst drought in a century. Seven of its eleven aquifers have no groundwater available for new concessions, a combined annual deficit of roughly 173 billion litres (CONAGUA/SEMARNAT, 2023), and roughly 96% of the state's remaining available groundwater sits in the Sierra Gorda highlands, far from the buildout. A $4.8 billion project to build six data centres by 2027 proceeds while residents receive water three days a week. Mexico has the epidemiology, the labor data, the policy critique, and the HCI capacity; the buildout is arriving faster than the test.

6.15 Chile

Chile is where the values-alignment mechanism described in Section 5 can be tested against the public record rather than inferred, because Chilean environmental permitting places operator projections and operator filings side by side.

Why verification is possible here: projects seeking approval file operational detail with Chile's Servicio de Evaluación Ambiental, permanent staffing among it, and those filings are public. A November 2025 investigation set them against the employment figures announced publicly for the same projects (Rodríguez Salamanca, Rest of World). For one operator, an announced figure of roughly 81,000 jobs — spanning direct, indirect and economy-wide effects, and derived from a consultancy economic-impact model — sits alongside an environmental permit describing 75 long-term jobs over thirty years at the facility concerned. For another, 5,520 jobs "supported" between 2021 and 2023, per a consultancy report the operator commissioned, sits alongside permits describing no more than 223 full-time operations roles. Across seventeen reviewed projects, filings account for no more than 1,547 permanent operations employees, averaging roughly ninety per facility; thirty-two further planned centers would add 909 permanent positions across a thirty-year operating phase.

What this demonstrates: the divergence is between two figures the same operators produced, measuring different things — economy-wide modelling against permanent site staffing. Section 5 argues that a workforce selected for systemizing cognition will notice precisely this kind of gap between a facility's public claims and its operational reality. Chile is where that argument stops being a prediction. The divergence is already on record, and it is on record because the filings exist.

Employment framework: Ley 21.015 requires employers of 100 or more — public bodies, the armed forces and private companies alike — to fill 1% of positions with persons with disabilities or holders of an invalidity pension, phased in from April 2018 for employers of 200 or more and April 2019 at the 100-employee threshold, with defined alternative compliance routes. This is the same threshold as Brazil's Art. 93.

The pathway: Chile already has both the disclosure infrastructure and the employment instrument. What a facility operating there can do more readily than elsewhere is publish operational staffing that a reader can check against the regulatory record — which converts transparency from a claim into a verifiable practice.

6.16 Brazil

Brazil pairs Latin America's largest computing buildout with a graduated employment statute that already covers autistic workers by operation of law, and with a domestic research lineage studying neurodivergent professionals in precisely the technical roles this paper describes.

Legal framework: Article 93 of Lei 8.213/1991 obliges companies of 100 or more employees to fill 2% to 5% of positions with qualified persons with disabilities or rehabilitated workers, on a graduated scale — 2% up to 200 employees, 3% from 201 to 500, 4% from 501 to 1,000, and 5% above that. Lei 13.146/2015, the Lei Brasileira de Inclusão, amended that article and set the wider rights framework around it. A third statute makes the connection to this paper: Lei 12.764/2012 provides that a person with autism spectrum disorder "is considered a person with disability, for all legal purposes" — which places autistic workers inside the Article 93 quota without further qualification. The 100-employee threshold is the operative detail: that is the staffing scale of a community-scale facility, not only of a hyperscale one.

Research base: two independent Brazilian groups study neurodivergent software professionals directly. At UFPE Recife, Gama and Lacerda examined how agile teams understand and support neurodiverse developers (SBES '23, pp. 497–502), work extended with researchers in Iceland and Portugal on the effect of cognitive differences on developer performance (arXiv:2411.13950). At UFPA Belém, Menezes, Rocha, Camelo and Mota interviewed nine neurodivergent practitioners on inclusion in software development teams, applying sociotechnical theory (arXiv:2503.09001, 2025). The two arXiv items are preprints and are not yet peer-reviewed. Together they are the closest existing match to the operational profile proposed here: neurodivergent people doing precision technical work, in teams.

Buildout: São Paulo is Latin America's largest data center market at 536.7 MW, holding 48 facilities and 351 MW of IT load — 51.42% of national capacity — with 39.2 MW absorbed in the first quarter of 2026, and Fortaleza the fastest-growing edge location (CBRE, Q1 2026). Brazil accounts for approximately four in ten data centers in Latin America.

The pathway: Brazil is the jurisdiction where the quota, the autism recognition and the research capacity are all already in place. What has not yet been done is to connect them to the data center sector — which makes Brazil a site where the measurement framework this paper proposes could be implemented against an existing legal obligation rather than a voluntary commitment.

6.17 South Africa

South Africa is where the geographic gap this paper names in §6.20 begins to close. It holds Africa's largest computing buildout alongside the newest employment instrument in this section, and a domestic research base already studying autistic adults in the open labour market.

Policy instrument: the Employment Equity Amendment Act 4 of 2022 took effect on 1 January 2025, proclaimed on 12 November 2024. Section 15A empowers the Minister of Employment and Labour to identify national economic sectors and set numerical targets for each — including for people with disabilities — across all occupational levels. Eighteen sectors were identified, with targets gazetted alongside the Employment Equity Regulations, 2025, on 15 April 2025. The amended definition of "people with disabilities" was expanded to include intellectual and sensory impairments. This is the only instrument in this section that operates per sector rather than per employer size, which means a data center sector can be addressed as a sector.

Buildout: Teraco's JB7, at the Isando Campus in Ekurhuleni east of Johannesburg, is a 40 MW hyperscale facility funded by an R8 billion syndicated loan and scheduled for completion in 2026 — eight data halls of 1,500 m² each, built in a single phase, with liquid-to-air and liquid-to-liquid cooling for high-density AI workloads. It brings the Isando campus to 110 MW of critical power load. Teraco, Africa Data Centres and Vantage are simultaneously developing capacity in the Samrand–Midrand corridor in Gauteng. The national colocation market is projected to grow from USD 496 million in 2025 to approximately USD 1.33 billion by 2030.

Research base: Soeker, Beversdorf and Fish (2022) examined the worker role as experienced by autistic adults across South African and central United States contexts, identifying supported employment and client-centred practice as the mechanisms that carried placement (Journal of Vocational Rehabilitation). A separate qualitative study followed ten autistic adults, with five key informants, moving from skills-training programmes into the open labour market in Cape Town. The transition pathway is documented rather than assumed.

The pathway: South Africa is the case where a sectoral target could be set for the data center sector itself, at the moment that sector is being built. Section 15A makes that administratively possible without new primary legislation.

6.18 United States

The United States contributes the largest single-country share of the evidence in this paper and has the least settled legal environment for applying it. It is a high-volume case, not a reference case.

The distinguishing feature: there is no single federal statute doing what Germany's EnEfG or the Netherlands' "no, unless" test do. The community-benefit test is being written case by case, county by county — negotiated rather than legislated. That is the current shape of the legal environment, not its absence.

What is concrete: prevalence and participation data (Centers for Disease Control: autism at 2.21% of US adults; National Autism Indicators Report: 58% of young adults on the spectrum worked for pay between high school and their early twenties), and a quantified demand signal (a projected shortfall of roughly 340,000 data-center roles by end-2026, a commercial survey). The adaptation pathway is contractual rather than statutory: fast to negotiate at one site, slow to generalize across many.

6.19 Additional Country Evidence

• Australia: DXC Dandelion Program (founded 2014) — 92% employment retention, 30–40% team productivity increase, 75% job satisfaction; methodology shared openly through Cornell University's Yang-Tan Institute; curriculum updated for the AI era (DXC, CIO, Jun 2026).
• Egypt: National IT agency co-sponsorship of neurodivergent tech careers through DXC Dandelion × ITIDA partnership.
• Iran: The research landscape is extending beyond its Western and East-Asian concentration. Iran — whose domestic compute buildout and technical-education base are substantial — shows an emerging 2026 AI+neurodiversity literature: a peer-reviewed editorial in the Medical Journal of Tabriz University of Medical Sciences positions AI as especially significant for conditions including autism and schizophrenia (Shafiee-Kandjani, Salatin & Hamidi, 2026), and an Iranian-context mixed-methods field study of AI-supported language learning for 142 neurodiverse learners reported large intervention effect sizes — vocabulary +16.3 (d=1.21), reading comprehension +13.3 (d=1.05), oral fluency +9.2 wpm (d=0.89) (Isaee et al., 2026, preprint — not yet peer-reviewed). No data-center-specific workforce figure is yet verifiable for Iran; it enters here as a research-landscape signal, with the region's community data centers as natural sites for generating the missing operational evidence.

6.20 Geographic Evidence Gaps

Published evidence is heavily concentrated in North America, Europe, East Asia, and Australia. Much of Africa, the Middle East, and Central Asia remains absent from the indexed literature — though South Africa (§6.17) and, more briefly, Egypt (§6.19) begin to close that gap, and Iran's 2026 AI+neurodiversity output (§6.19) is an early sign the concentration is beginning to loosen. Community data centers in underserved regions are natural sites for generating the missing evidence — converting absence of evidence into participation.

Community-scale data centers possess structural advantages for implementing these findings that centralized hyperscale facilities are less positioned to pursue. They are the most people-dependent segment of the industry, running staffing ratios of 2.0–3.0 staff per MW against 0.2–0.3 in automated hyperscale (iRecruit, 2026). Moderate scale permits design-time rather than retrofit implementation of cognitive-friendly environments and AI-collaborative workflows.

7.1 Workplace Environment Design

Sensory-friendly design — noise reduction, adjustable lighting, flexible workstations — can be achieved at community scale when accessibility is addressed at design time. Organizational-level supports, not awareness training alone, are what the research shows to be effective (Hartman et al., 2024, a peer-reviewed study). The W3C COGA "Content Usable" Working Group Note offers guidance for people with ADHD, people with autism, people with dyslexia, and people who are neurodivergent more broadly, and is cited as guidance by public-sector bodies such as the UK Department for Education. COGA and WCAG are not interchangeable — the W3C presents COGA as supplementary guidance whose objectives WCAG does not fully address. Interfaces should target WCAG 2.2 AA conformance and COGA Content Usable alignment jointly, plus ETSI EN 301 549 where European public-sector procurement requirements apply. The World Health Organization estimates that more than one billion people — roughly 15% of the world's population — live with some form of disability, and cognitive and learning disabilities form a substantial, often under-served subset. Accessibility barriers are situational as well as permanent, and a network operations center can be highly sensorially demanding. Publicly available industrial operations tooling often does not foreground cognitive accessibility as a distinct category. Usability validation by operators who are neurodivergent should be part of the definition of done for operational interfaces.

7.2 AI-Assisted Operations

In the proposed model, repetitive and linear tasks are handled by bounded AI agents governed by versioned, human-readable policy specifications. Agents propose physical-impact actions but never execute them unilaterally. Human operators focus on pattern recognition, anomaly triage, and final approval. Environmental variables — grid carbon intensity, water exposure — are first-class inputs to agent policy, not reporting afterthoughts. This division of labor is not a proposal; it is where the industry is heading. A GlobalLogic corporate publication (Oct 2025) describes DCIM evolving from a dashboard into an intelligent execution layer. A vendor publication (HiveMQ, Apr 2026) identifies a unified event backbone as the precondition for higher-order reasoning. Industry analyst firm Gartner projects semiautonomous agents will orchestrate 10% of key operations, quality, and maintenance use cases by 2030, with humans retaining final approval. A 2026 arXiv preprint, not yet peer reviewed, formalizes the bounded-agent pattern — agents that propose actions without executing them unilaterally, with carbon and water as first-class variables. The synthesis: the industry is converging on human-on-the-loop agentic operations and has not yet specified which operators sit in that seat or what interface they need. The loop and the human in it are one design problem, not two.

7.3 Community Talent Development

Partnering with local educational institutions to establish data center skills pathways tailored for ADHD, autistic, and co-occurring profiles. The template is proven: structured academy programs report an average of 75% of graduates securing employment within three months, across programs on three continents (Specialisterne Academy via Hg Foundation). National precedents exist for public sponsorship: Egypt's ITIDA, China's 全链式融合就业服务 funding channels, Japan's METI neurodiversity promotion, and vocational-school partnership mandates in multiple jurisdictions.

7.4 Methodology Transfer

Multi-model parallel scheduling, non-linear research workflows, and cross-domain correlation analysis — working methods developed through neurodivergent–AI cognitive collaboration — migrate directly into operations-team practice, improving overall efficiency for all staff. This is universal design applied to method: what is built for neurodivergent operators measurably serves everyone, including every operator at 3 a.m. during an incident.

The field's most significant evidence gap — lack of controlled workplace outcome data (Bury et al., 2020) — can be addressed by the facilities themselves. Adopting facilities should implement a common measurement schema across four domains, publishing anonymized aggregates annually:

The metrics track those already used by corporate neurodiversity programs (Juicebox, 2026), ensuring comparability. Each adopting facility becomes an evidence generator — converting the field's biggest weakness into a structural contribution. Facilities are encouraged to register their measurement protocols and contribute anonymized data to a common repository, building the controlled evidence base the field currently lacks.

This paper identifies the following limitations and open questions:

1. Productivity figures are corporate self-reports (§3.1). The headline numbers (90–140%, 150%) come from corporate programs, not peer-reviewed controlled studies. Peer-reviewed confirmation exists for specific strength domains (process inefficiency detection, LLM communication preference, self-reported strengths); controlled workplace outcome studies across neurodivergent and neurotypical teams in matched operational environments do not.

2. Strengths are trait-associated, not universal (§3.5). The cognitive strengths described are population-level tendencies surfaced by skills-based matching, not guarantees about any individual. The distinction between "trait-associated tendency" and "individual capability" is load-bearing.

3. AI complementarity requires design (§4.4). Providing AI tools without deliberate design for neurodivergent working styles does not produce the gains described. The 2025 neuronormativity critiques and conformity-risk warnings require that AI-collaborative workflows be individually adapted, not universally deployed.

4. The values-alignment hypothesis (§5) is a reasoned inference grounded in converging evidence — justice sensitivity research, systemizing cognition literature, moral distress and burnout mechanisms — but has not been tested in the specific configuration proposed: whether verified environmental integrity correlates with retention and wellbeing among neurodivergent operators working in AI-paired roles. A second, analytically separable question concerns mechanism rather than outcome: the observed finding that neurodivergent operators prefer and recommend AI assistance (DBT, p = 0.035) is established; the proposal that AI pairing lowers the cost of cross-neurotype information transfer (Crompton et al., 2020, human-human) is untested. Because the two may be linked, the measurement framework in §8 is designed to test them jointly via a pre-specified mediation analysis (pairing → reduced transfer cost → retention/wellbeing), rather than as independent claims.

5. Geographic evidence gaps (§6.20). Published evidence is concentrated in North America, Europe, East Asia, and Australia. Generalization to other regions requires locally generated evidence.

6. Diagnostic scope. The evidence base concentrates on ADHD, autism, and their co-occurrence. Other neurodivergent configurations (dyslexia, dyspraxia, dyscalculia, Tourette syndrome) are acknowledged but less represented in the cited literature. The framework's modular design is intended to accommodate expansion as evidence develops.

Four independent evidence domains — corporate outcomes, AI collaboration research, labor economics, and global policy — converge on the same finding: neurodivergent cognitive styles are structurally well-suited to the AI-collaborative operational model the data center industry is already adopting. A decade of corporate programs documents 90–140% productivity and ~90% retention. Peer-reviewed research confirms specific cognitive strengths — anomaly detection, systemizing, attention to detail — that map directly onto operations roles. The industry's own workforce crisis creates operational necessity for tapping systematically excluded talent pools. And policy frameworks in China, Singapore, Japan, Germany, and beyond have created the institutional scaffolding for exactly the community talent development pathways described here.

The values-alignment mechanism — through which environmental integrity converts from ethical preference to workforce infrastructure — provides the retention logic that the industry's own exit data show it urgently needs. A workforce selected for pattern recognition and driven by intrinsic meaning will not sustain engagement in operations whose claims contradict their observed reality. Genuinely green, transparently measured, community-serving operations convert that same sensitivity into meticulous care and sustained engagement.

The evidence gaps are named: productivity figures require controlled validation, the values-alignment hypothesis awaits testing, and geographic coverage is uneven. A measurement framework is proposed through which adopting facilities close these gaps — converting each community data center into a generator of the evidence the field currently lacks.

The institutional components exist. The evidence converges. The adaptation pathways are defined. Divergosynnoetics names the field within which that implementation unfolds — an emerging field, proposed here and made testable by the measurement framework in Section 8.

1. Forbes (Mar 2026). "Why AI Is Making Neurodivergent Talent the Most Valuable Hire in Tech." forbes.com
2. Juicebox (2026). "The Complete Guide to Neurodivergent Hiring Programs for 2026." juicebox.ai
3. CNBC (Nov 2025). "Why workers with ADHD, autism, dyslexia should use AI agents." cnbc.com
4. CNBC (Mar 2026). "How the red-hot AI data center boom is igniting demand for a new, lucrative career path: Trade workers." cnbc.com
5. Microsoft Source (Jan 2026). "How AI Helps Neurodivergent Professionals Showcase Their Strengths." news.microsoft.com
6. Enna Global (Jun 2026). "AI: The Most Powerful Assistive Technology for Neurodivergent Staff." enna.org
7. Jang, J., Moharana, S., Carrington, P., & Begel, A. (2024). "It's the only thing I can trust": Envisioning Large Language Model Use by Autistic Workers for Communication Assistance. In Proceedings of the CHI Conference on Human Factors in Computing Systems (CHI '24). ACM. doi.org
8. arXiv:2507.06864 (Jul 2025). "ADHD-Aware Human-in-the-Loop Frameworks." arxiv.org
9. Hartman et al. (2024). "Neurodiversity in the workplace: An agenda for research and action." Wiley Diversity & Inclusion Research. onlinelibrary.wiley.com
10. Cope, R. & Remington, A. (2022). "The Strengths and Abilities of Autistic People in the Workplace." Autism in Adulthood. doi.org
11. PYMNTS (Oct 2025). "How Neurodivergent Workers Could Provide Value in AI Training." pymnts.com
12. Stanford Social Innovation Review (Jul 2024). "Changing Our Minds About Developmental Disabilities in South Korea." ssir.org
13. PR Newswire (2020). "Testworks and StradVision Partnership." prnewswire.com
14. JLL (2025). "Data Center Outlook: Staffing as Critical Constraint." jll.com
15. iRecruit (2026). "Data Center Staffing Ratios 2026." irecruit.co
16. Per Scholas (2026). "The Data Center Staffing Crisis: The Hidden Bottleneck to AI Growth." perscholas.org
17. DataX Connect / Schneider Electric (Feb 2026). "Mind the Gap: Bridging AI Talent Shortages in Data Centers." blog.se.com
18. Uptime Institute (2024). "Global Data Center Staffing Survey." uptimeinstitute.com
19. DC Geeks (Apr 2026). "The Data Center Workforce Shortage Is Your Way In (2026)." dcgeeks.com
20. Grand View Research (2026). "Edge Data Center Market Report." grandviewresearch.com
21. STL Partners, Osborne, H. (accessed 2026). "Key Edge Computing Statistics" — network edge capacity forecast figures: "26% of network edge sites globally will be in China in 2026"; "just under 1,200 network edge data centres by 2026." stlpartners.com
22. Gartner (2026). "Manufacturing Predicts: AI-Augmented Operations."
23. Cruzes, S. (2026). "AI Infrastructure Sovereignty." arXiv:2602.10900. arxiv.org
24. State Council of China (Jun 2025). 《促进残疾人就业三年行动方案（2025—2027年）》. 国办发〔2025〕23号. gov.cn
25. China Disabled Persons' Federation (CDPF). 《孤独症儿童关爱促进行动实施方案（2024—2028年）》. cdpf.org.cn
26. Xinhua (Apr 2026). "十五五规划纲要：特殊教育与助残服务列入重大工程项目." news.cn
27. Legal Daily / Guangming (Jun 2025). "Jiangsu, Zhejiang Full-Lifecycle Disability Employment Plans." legal.gmw.cn
28. Guangming Daily (2024). "China Autism Prevalence: 13M+ and Growing." gmw.cn
29. Autism Spectrum News (2025). "Increasing Autism Awareness and Reducing Stigma in China." autismspectrumnews.org
30. Sichuan Provincial Policy Briefing (Jul 2025). "全链式融合就业服务 Implementation." jxt.sc.gov.cn
31. DXC Technology (Jun 2026). "DXC Marks Five Years of Opening Tech Careers to Neurodivergent Talent." dxc.com
32. CIO (Dec 2025). "The DXC Program Empowering Neurodivergent IT Pros." cio.com
33. Cornell University Yang-Tan Institute. "DXC Dandelion Program Methodology." ecommons.cornell.edu
34. IMDA Singapore (2024). "Green Data Centre Roadmap." imda.gov.sg
35. German EnEfG (2023). "Energieeffizienzgesetz: PUE and Heat Reuse Requirements."
36. PropertyNL (Jan 2025). "Amsterdam Data Center Policy: No, Unless." propertynl.com
37. Schäfer, T. & Kraneburg, T. (2015). "The Kind Nature Behind the Unsocial Semblance: ADHD and Justice Sensitivity—A Pilot Study." Journal of Attention Disorders, 19(8), 715–727. doi.org
38. Hu, Y., Pereira, A. M., Gao, X., Campos, B. M., Derrington, E., Corgnet, B., Zhou, X., Cendes, F., & Dreher, J.-C. (2020). "Right Temporoparietal Junction Underlies Avoidance of Moral Transgression in Autism Spectrum Disorder." Journal of Neuroscience, 40(8), 1699–1708. doi.org
39. BMC Nursing. "Organizational Justice, Moral Sensitivity, and Moral Distress Among Nurses." PMC10559416. ncbi.nlm.nih.gov
40. Frontiers in Psychiatry (19 May 2026). "Understanding Workplace Needs of Autistic Adults in Singapore: Inclusive AI Support." frontiersin.org
41. Raymaker, D.M., Teo, A.R., Steckler, N.A., et al. (2020). "Having All of Your Internal Resources Exhausted Beyond Measure and Being Left with No Clean-Up Crew": Defining Autistic Burnout. Autism in Adulthood. doi.org
42. Hours, C., Recasens, C., & Baleyte, J.-M. (2022). "ASD and ADHD Comorbidity: What Are We Talking About?" Frontiers in Psychiatry, 13, 837424. frontiersin.org
43. Simonoff, E. et al. (2008). "Psychiatric Disorders in Children with Autism Spectrum Disorders: Prevalence, Comorbidity, and Associated Factors." Journal of the American Academy of Child & Adolescent Psychiatry, 47(8), 921-929. pubmed.ncbi.nlm.nih.gov
44. Salazar, F. et al. (2015). "Co-occurring Psychiatric Disorders in Preschool and Elementary School-Aged Children with Autism Spectrum Disorder." Journal of Autism and Developmental Disorders, 45, 2283-2294. pubmed.ncbi.nlm.nih.gov
45. Baron-Cohen, S., Richler, J., Bisarya, D., Gurunathan, N., & Wheelwright, S. (2003). "The systemizing quotient: an investigation of adults with Asperger syndrome or high-functioning autism, and normal sex differences." Philosophical Transactions of the Royal Society B, 358(1430), 361–374. doi.org
46. Bury, S.M., Hedley, D., & Uljarević, M. (2020). "The autism advantage at work: A critical and systematic review of current evidence." Research in Developmental Disabilities, 105, 103750. doi.org
47. Dell'Acqua, F., McFowland, E., Mollick, E.R., et al. (2023). "Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality." Harvard Business School Working Paper 24-013. ssrn.com
48. Sinha, S. & Sinha, D. (2026). "Why Our Voices Don't Count: The Employment Experiences of Neurodivergent Employees Through a Double Empathy Lens." researchgate.net
49. SAP / amazingworkplaces (2025). "Autism at Work: Global Program Results."
50. AskEARN. "Neurodiversity Hiring Initiatives Database." askearn.org
51. Harvard Business School Working Knowledge (Jul 2016). "Neurodiversity: The Benefits of Recruiting Employees with Cognitive Disabilities." library.hbs.edu
52. Newz Hook (Apr 2021). "Autism at work programmes that help tech companies integrate inclusion with business development goals." newzhook.com
53. Specialisterne Academy / Hg Foundation. "Structured Academy Program: 75% Graduate Employment Within Three Months."
54. Alibaba / arXiv (2026). "Generative AI in Customer Service Operations: Large-Scale Field Experiment." arXiv:2603.29888.
55. Fein, L. (1961). "The Computer-Related Sciences (Synnoetics) at a University in the Year 1975." American Scientist, 49(2), 149–168; BIT Numerical Mathematics, 1, 227–255.
56. Phenix, P. H. (1964). Realms of Meaning: A Philosophy of the Curriculum for General Education. McGraw-Hill.
57. Milton, D. (2012). "On the Ontological Status of Autism: The 'Double Empathy Problem.'" Disability & Society, 27(6), 883–887.
58. Harding, S. (1991). Whose Science? Whose Knowledge? Thinking from Women's Lives. Cornell University Press.
59. Haraway, D. (1988). "Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective." Feminist Studies, 14(3), 575–599.
60. Walker, N. (2021). Neuroqueer Heresies: Notes on the Neurodiversity Paradigm, Autistic Empowerment, and Postnormal Possibilities. Autonomous Press.
61. Data Center Dynamics (DCD) (Jun 2026). "The ongoing impact of Amsterdam's data center moratorium." datacenterdynamics.com
62. State Council of China (Mar 2022). 《促进残疾人就业三年行动方案（2022—2024年）》. 国办发〔2022〕6号. gov.cn
63. CAICT (Mar 2026). 《先进计算暨算力发展指数蓝皮书（2025年）》. caict.ac.cn
64. China Disabled Persons' Federation (Apr 2026). 《2025年残疾人事业发展统计公报》. cdpf.org.cn
65. National Data Administration / CCTV (2025). 东数西算 带动社会投资超万亿元. nda.gov.cn
66. Xinhua (Aug 2024). 东数西算 八大枢纽节点直接投资超435亿元. news.cn
67. European Commission (2019). "Bulgaria — New Persons with Disabilities Act." European network of legal experts. equalitylaw.eu
68. State Council of China / MOHRSS (Apr 2024). 《加快数字人才培育支撑数字经济发展行动方案（2024—2026年）》 — digital talent gap of 25–30 million. gov.cn
69. Eurostat (2026). "Towards Digital Decade targets for Europe" — 10.4M ICT specialists, ~9.6M below the 2030 goal. europa.eu
70. METI (Apr 2019). 「IT人材需給に関する調査」— IT talent supply–demand survey (with Mizuho Information & Research Institute). meti.go.jp
71. Uptime Institute (Jan 2020). "Industry's First Global Data Center Staffing Forecast" — 2M FTE (2019) → ~2.3M (2025); Asia-Pacific leads demand. uptimeinstitute.com
72. NLB Services / VARINDIA (2026). "India data centre boom could create 1 lakh jobs by 2030." varindia.com
73. Mordor Intelligence (2026). "Malaysia Data Center Construction Market." mordorintelligence.com
74. MGR Online (Oct 2023). "ดีอีเปิดแนวทาง Global Digital Talent Visa" — Thailand needs ~100,000 digital workers/yr vs ~25,000 produced (DES Minister Prasert Jantararuangtong). mgronline.com
75. ANTARA (Mar 2022). "Indonesia needs nine million digital talents in 2030: Minister." antaranews.com
76. CNBC Indonesia (Mar 2024). "RI Butuh 9 Juta Talenta Digital Hingga 2030" (Kominfo DG Semuel Abrijani). cnbcindonesia.com
77. Expansión (Feb 2026). "México construye más centros de datos, pero tiene un problema" — MEXDC talent deficit (39% E/M engineering; 32% mid-level ops; 67% firms). expansion.mx
78. Diario de Querétaro (Feb 2026). "Data center enfrentan déficit de 39% en ingenieros eléctricos y mecánicos." oem.com.mx
79. METI (Mar 2022). 令和３年度産業経済研究委託事業「イノベーション創出加速のためのデジタル分野における『ニューロダイバーシティ』の取組可能性に関する調査」調査結果レポート — projects a maximum IT talent shortage of ~790,000 by 2030 (demand ~1.92M vs supply ~1.13M). meti.go.jp
80. METI. 「ニューロダイバーシティの推進について」— neurodiversity promotion; focuses on the digital field because securing talent is "an urgent issue for the government." meti.go.jp
81. Disability:IN. "Bulgaria Country Profile." disabilityin.org
82. Dutch Data Center Association / Pb7 Research (Jul 2021). "Datacenters & Werkgelegenheid 2021" — direct DC employment ~5,000 (2021) → >7,100 (2026); >600 open vacancies; 3% women in technical roles. dutchdatacenters.nl
83. AGConnect (Jun 2024). "State of the Dutch Data Centers" (10th ed.) — 70% of operators cite personnel shortage (up from 45% in 2023). agconnect.nl
84. Shafiee-Kandjani, A. R., Salatin, S., & Hamidi, S. (2026). "Artificial Intelligence in Neurology and Psychiatry." Medical Journal of Tabriz University of Medical Sciences, 48(1), 7–11. doi: 10.34172/mj.026.35338. mj.tbzmed.ac.ir
85. Isaee, H., Barjesteh, H., Curlie, S., & Manoochehrzadeh, M. (2026, preprint). "AI-Supported Language Learning for Neurodiverse Students in Low-Resource Contexts: Student and Teacher Perspectives and Its Impact on Learning Outcomes." Preprints.org, 202602.1568.v1. doi: 10.20944/preprints202602.1568.v1. Not peer-reviewed. preprints.org
86. Buckland Review of Autism Employment (Feb 2024). "Report and recommendations." UK Department for Work and Pensions. gov.uk
87. W3C COGA Task Force (Apr 2021). "Making Content Usable for People with Cognitive and Learning Disabilities" (Content Usable). W3C Working Group Note. w3.org
88. UK Department for Education. "Cognitive accessibility guidelines (COGA)." Accessibility and inclusive design manual. accessibility.education.gov.uk
89. ETSI EN 301 549. "Accessibility requirements for ICT products and services." European harmonized standard for public procurement. etsi.org
90. GlobalLogic (Oct 2025). "Revolutionizing Data Centers with Digital Twins and Agentic AI." globallogic.com
91. HiveMQ (Apr 2026). "Digital Twins and Agentic AI: A Data Maturity Path to Intelligence-Driven Operations." hivemq.com

92. World Health Organization & World Bank (2011). "World Report on Disability" — more than one billion people (about 15% of the world's population) live with some form of disability. who.int

93. Khudiakova, V., Russell, E., Sowden-Carvalho, S., & Surtees, A. D. R. (2024). "A systematic review and meta-analysis of mental health outcomes associated with camouflaging in autistic people." Research in Autism Spectrum Disorders, 118, 102492. doi.org

94. Benatov, J., Sarel-Mahlev, O., & Bar Yehuda, S. (2026). "Camouflage, Burnout-Exhaustion, and Depression in Autistic Adults." Autism in Adulthood, 8(4), 801–808. doi.org
95. European Commission (2021). "Union of Equality: Strategy for the Rights of Persons with Disabilities 2021–2030." ec.europa.eu
96. European Commission (2022). "Disability Employment Package." commission.europa.eu
97. European Commission (2026). "Enhancing the strategy for the rights of persons with disabilities up to 2030." COM(2026) 541 final, CELEX 52026DC0541.
98. Vichyanond, J. (2026). "Malaysia's Data Center Boom: From Investment Surge to Sustainable Growth." ASEAN+3 Macroeconomic Research Office (AMRO). amro-asia.org
99. Cushman & Wakefield (2026). "APAC Data Centre H1 2026 Update." cushmanwakefield.com
100. Knight Frank (2026). "Data Centre Atlas 2026." knightfrank.co.uk
101. Md Tah, I. H. (2013). "Rights of Persons with Disabilities: The 'Almost Forgotten' Protection under the Federal Constitution." In K. A. Mokhtar (ed.), Constitutional Law & Human Rights in Malaysia, 411–430. Sweet & Maxwell Asia. DOI 10.13140/2.1.2014.5285. researchgate.net
102. Thailand Board of Investment (2026). Press Release No. 67/2569. boi.go.th
103. Thailand (2007, amended 2013). Persons with Disabilities Empowerment Act B.E. 2550. un.org
104. UNDP (2023). "Promoting an Inclusive Workplace for Persons with Disabilities in Thailand." undp.org
105. Indonesia (2016). Law No. 8 of 2016 on Persons with Disabilities, Art. 53.
106. Anggraini, N. K., & Susetyo, H. (2024). "Fulfillment of the Constitutional Rights of Persons with Disabilities to Employment Opportunities in the Banking Industry of State-Owned Enterprises." Journal of Law, Politic and Humanities, 4(6), 2119–2132. doi.org
107. Indonesia at Melbourne, University of Melbourne. "Indonesian courts rule to protect the work rights of people with disability." unimelb.edu.au
108. Arora, N. K., et al. (2018). "Neurodevelopmental disorders in children aged 2–9 years: Population-based burden estimates across five regions in India." PLOS Medicine, 15(7), e1002615. doi.org
109. Government of India (2016). Rights of Persons with Disabilities Act, 2016. depwd.gov.in
110. NIIT (2026). India Skills Gap Report 2026.
111. Quess Corp (Q3 FY26). "India's Data Centre Decade: Capital, Cloud & Capability at Hyperscale." quesscorp.com
112. Career Development International (2025). "Expecting inclusion, encountering barriers." doi.org
113. Rodríguez Salamanca, L. (2025). "Microsoft and Google overstate job creation at Chile data centers." Rest of World. restofworld.org
114. Chile (2017). Ley 21.015 — Incentiva la inclusión de personas con discapacidad al mundo laboral. bcn.cl
115. Lei 8.213/1991, Art. 93 (Brazil). planalto.gov.br
116. Lei 13.146/2015 (Lei Brasileira de Inclusão da Pessoa com Deficiência). planalto.gov.br
117. Lei 12.764/2012 (Lei Berenice Piana). planalto.gov.br
118. Gama, K., & Lacerda, A. (2023). "Understanding and Supporting Neurodiverse Software Developers in Agile Teams." SBES '23, 497–502. doi.org
119. Menezes, Rocha, Camelo & Mota (2025). arXiv:2503.09001 (preprint, not peer-reviewed). arxiv.org
120. South Africa (2022). Employment Equity Amendment Act 4 of 2022.
121. Teraco (2026). JB7 hyperscale facility, Isando Campus. teraco.co.za
122. Soeker, M. S., Beversdorf, D., & Fish, C. (2022). "Exploring the experiences of individuals living with autism spectrum disorder about their worker role: The South African and Central United States contextual experience." Journal of Vocational Rehabilitation, 57(3). doi.org
123. Soeker, M. S. (2020). "A descriptive, qualitative study of the challenges that individuals with Autism Spectrum Disorder experience when transitioning from skills training programs into the open labor market in Cape Town, South Africa." Work, 65(4). pubmed.ncbi.nlm.nih.gov

Definition. Divergosynnoetics is the study and design of synergistic cognitive partnerships between neurodivergent minds and artificial cognitive systems, in which diverse cognitive styles working alongside machine capacities expand what either could achieve alone.

Etymology — a braid, not a lineage. The term braids two independent 1960s coinages of synnoetics (Greek synnoesis, "thinking together"): Louis Fein (1961), who framed it as the structural joint mental power of humans and machines, and Philip Phenix (1964), who framed it as relational, tacit, personal knowing. There is no evidence of influence between them; the present term deliberately reweaves the two. It is operationalized along two axes: position — divergence from dominant norms (community-coined by Kassiane Asasumasu and Judy Singer; theorized by Nick Walker), grounded in standpoint epistemology (the view that what one knows is shaped by one's social position — Harding; Haraway) — and mutuality, grounded in the double empathy problem (the finding that misunderstanding between autistic and non-autistic people runs both ways, not an autistic empathy-deficit — Milton, 2012). A note on terminology hygiene: modern usage is a deliberate synthesis, not a continuous tradition since 1961. Fein's coinage lost the naming contest to "computer science"; its revival here for human–AI collective cognition is a conscious reclamation.

Core premise (falsifiable). Synergy in neurodivergent–machine partnership is conditional on values alignment between the system and its neurodivergent partner. Absent alignment, synergy should not obtain — any demonstrated case of unaligned synergy disconfirms the premise.

Epistemic note. First-person testimony settles what it is like and audits what we conclude; instruments settle how often and how much.

Strengths formula. Position, not gift; contextual, evidence-based, self-attributed strengths; needs co-presented; worth unconditional. Divergence names a position — departure from prevailing cognitive norms — not a processing style. It asserts nothing about creativity or non-linearity; many divergent minds are strongly linear, sequential, and convergent — no less divergent for it. Sovereignty includes silence: no person is required to disclose, name, narrate, co-present, or audit their divergence; inference from population-level findings to any individual's self-description is prohibited — the audit runs one way.

Published as an open framework. Free to download, cite, and adapt with attribution. Contact: eternalharmony.ai/contact

Eternal Harmony AI | Eternal Harmony LLC · v1.1 · Finalized July 2026 · Published August 2026 · Updated August 2026 — §6 expanded from 11 to 20 sections