The race takes off in the next big arenas of competition — MGI 2026 update

In the arenas, investment cycles are accelerating, value pools are shifting, and a new type of competitor is scaling across industries.

(Cover line — McKinsey Global Institute, March 2026, 127 pp.)

The 2026 update of MGI’s The next big arenas of competition (October 2024 original). Six authors led by Kevin Russell and Chris Bradley, with Naveen Sastry, Suhayl Chettih, Kweilin Ellingrud, and Natalya Goryunova. PDF metadata records a creation date of 25 March 2026; cover names “March 2026”. Confidential-and-proprietary stamped but distributed freely on McKinsey.com. Companion live event on the McKinsey YouTube channel (2026-05-12-mgi-virtual-event-race-takes-off-next-big-arenas) presented the findings on 12 May 2026 — Ellingrud and Russell presenting; Bradley moderating; panel discussion with Brendan Gaffey and Gayatri Shenai (neither in the PDF author list) plus PDF-co-author Naveen Sastry. Suhayl Chettih is a PDF co-author but does not appear on the live-event panel. The video transcript was re-fetched 2026-05-15 at --timeout 180000 after an initial 90s timeout failed; it carries substantive Q&A material not in the PDF (the Apollo-program-vs-omniscaler-capex comparison, the Anthropic/xAI infrastructure deal as live data point, Naveen Sastry’s omniscaler-vs-conglomerate framing).

TL;DR

The wiki’s first MGI economy-mapper anchor on the AI/competition wave — a McKinsey Global Institute longitudinal frame that quantifies the AI investment surge in firm-and-industry terms, complementary to AI Index 2026 (capability layer) and to Sternfels 2026 (McKinsey-firm self-narrative). Five substantive contributions:

1. The 18 future arenas have decisively outpaced the rest of the economy since 2022. From 2022 to 2025, the 18 future arenas added $18 trillion in market cap and $1.4 trillion in revenue. Market-cap CAGR: 29% for arenas vs 8% for other industries (nearly 4× faster). Revenue CAGR: 11% for arenas vs 1% for other industries (10× faster). Capex+R&D CAGR: 14% vs 4%. Total 2025 footprint: arenas $33T market cap on $5T revenue; other industries $93T market cap on $50T revenue. The 18 arenas drove ~half of total market-cap growth and revenue growth of large global companies over those three years.

2. The five-theme decomposition of the 18 arenas, with the AI foundation theme dominating value accretion:

   • AI foundation (3 arenas: semiconductors, cloud services, AI software & services) — +$10.77T market cap, +$490B revenue since 2022. Inside this: Semiconductors +$7.39T (~60% of the AI-foundation market-cap gain); Cloud +$1.83T; AI software & services +$1.55T. Theme share of arena market cap grew from 15% (2022) to 26% (2025).
   • Digitization (5 arenas: e-commerce, digital advertising, video games, streaming video, cybersecurity) — +$4.90T market cap, +$680B revenue (the largest revenue gain of any theme).
   • Electrification (3 arenas: EVs, batteries, nuclear fission) — +$930B market cap, +$190B revenue.
   • Hard tech (5 arenas: robotics, shared autonomous vehicles, future air mobility, space, modular construction) — +$560B market cap, +$30B revenue. Physical AI — defined as systems that “sense, think, and act in the real world” — is a named subset of hard tech (robotics + SAVs + future air mobility), distinct from space and modular construction.
   • New bio-frontiers (2 arenas: obesity drugs, non-medical biotech) — +$630B market cap, +$50B revenue. Obesity drugs are now ~6 of every 100 US prescriptions.

3. The omniscaler thesis — nine firms competing across multiple arenas at once. MGI coins “omniscalers” as a new category alongside hyperscalers. Definition: top-30 spenders by R&D + capital expenditures in 2024 AND earning revenue in three or more future arenas. Nine firms qualify: Alibaba, Alphabet (cluster), Amazon (cluster), Apple, Huawei, Meta, Microsoft (cluster — incl. Tesla/SpaceX-aligned investments), Samsung, with Tesla/X treated as its own cluster spanning EVs / robotics / AI / space. Together they generated ~$700B in operating cash flow and invested ~$800B in R&D + capex in 2025 alone. Combined revenues ~$2.7T — larger than Italy’s GDP. The largest, Alphabet cluster, plays in 9 future arenas; Amazon cluster in 8; Tesla/X in 5; Apple, Meta, Samsung in 5–7; Alibaba, Microsoft cluster, and Huawei in 5 each. 6 of the 9 are US-headquartered; three (Alibaba, Huawei, Samsung) are Asian.

   • The omniscaler advantage is structural: “reusable infrastructure, data network effects, high risk appetites, and top talent attraction among their interlocking strengths.”
   • Capability compounds across arenas: revenue per omniscaler-in-arenas ~$200B vs ~$10B for other arena players — a ~20× gap.
   • This concept is a candidate new wiki concept page once a second source corroborates; deferred for now.

4. Regional concentration: 90% of arena market value is US or Greater-China headquartered. US firms lead 14 of the 18 arenas in market cap and 10 in revenue, on deep capital markets and on capitalising on past-arena (cloud / consumer-internet / biopharma / EV) success. Greater China is gaining ground on revenue, especially in electrification (scale + vertical integration + manufacturing intensity in EVs / batteries / nuclear). Japan + South Korea added share via industrial and consumer electronics. Rest-of-world stands by — Europe’s venture-capital pool is materially smaller and arena-leading firms are sparse. Time to scale — permitting speed, grid and energy availability, predictable rules — is becoming a decisive differentiator alongside technological advances.

5. The “arena-creation potion” — three ingredients that, together, mark an industry as arena-prone:

   • A technology or business-model step change (e.g., transformer-based foundation models in 2022; GLP-1 receptor agonists for weight loss).
   • An escalatory investment pattern (capital outlays climbing faster than revenue; arms-race dynamics).
   • A large or expanding addressable market (one with room to absorb the investment at global scale).
   • When all three are present, arena dynamics tend to follow — high growth, high market-share shuffle, high R&D intensity, high profitability. This is the report’s reusable foresight heuristic; the wiki should track it on its strategic-foresight concept page.

Plus the strategic-implications chapter (Chapter 5) for decision-makers:

• Three concentric zones for any company: in an arena (compressed decision windows; commit early, reallocate at speed, reshape governance for agility) / near an arena (supplier / infrastructure / distribution exposure; engage early before value pools settle) / on the fringe of an arena (underperformance tomorrow, not disruption today; reallocate resources to get ready).
• Arena radar diagnostic (Exhibit 17) — plot proximity (distance from centre) × side-of-business-impact (production vs revenue) for each arena that could affect the firm; bubble size = revenue of the opportunity.
• Escalatory investment is not an automatic imperative. McKinsey’s How strategy champions win research: above-median capital expenditure helps move companies up the Power Curve only when matched with a credible path to competitive advantage. Entering or expanding within an arena requires a clear view of whether the company can earn returns exceeding its cost of capital and defend its position as competition intensifies.
• Three swing factors that will determine which of the 18 future arenas actually realise their potential: geopolitics (technology sovereignty + supply-chain resilience policies — particularly in semis and battery supply chains); AI development and adoption pace (will valuations be supported by sustained ROIC above cost of capital?); electrification pace (EVs, batteries, nuclear fission, plus second-order opportunities in adjacent industries).

What was actually ingested

This wiki page summarises pp. 1–18 (front matter, At-a-glance, Introduction, full Executive Summary, exhibits E1–E5), pp. 32–43 (AI-foundation + digitization theme deep-dives, exhibits 5–6), and pp. 78–82 (Chapter 5 implications + Exhibit 17 arenas radar). The future-arenas compendium (pp. 83+) with per-arena deep-dives across AI foundation / digitization / electrification / hard tech / new bio-frontiers is deferred; future ingests can expand specific arenas as needed. The regional-dynamics chapter (Chapter 4, pp. 66–77) is partially summarised through the executive-summary regional findings (Exhibit E6 not directly read). The technical appendix (pp. 119+) is deferred.

The PDF identity was verified per §Verifying sources before ingest: the cover names six authors, March 2026, McKinsey Global Institute. The PDF metadata’s CreationDate Wed Mar 25 16:21:27 2026 is the cited date_published. 127 pages, complete (no excerpt / sample / preview); TOC references resolve within the page count. No scope-mismatch flag.

The five themes — detailed positioning of each arena

The five-theme grouping is the report’s analytical organising frame across Chapters 2 and the compendium. The 18 arenas split as:

Theme	Arenas	2025 status vs 2040 scenario (upper / middle / lower bound)	2022–25 revenue CAGR (actual)	2022–25 market-cap CAGR
AI foundation	Semiconductors / Cloud services / AI software & services	All three tracking near or above upper bound — AI software at 55% revenue CAGR, 142% market-cap CAGR	17–25% (AI sw&s baseline)	32–142% range
Digitization	E-commerce / Digital advertising / Video games / Streaming video / Cybersecurity	Mostly middle-to-upper bound; digital ads at 14% (above scenario)	7–24% per arena	—
Electrification	EVs / Batteries / Nuclear fission	EVs at 18% (above scenario); nuclear at 13%; batteries at 14%	12–18% (EVs lead)	—
Hard tech	Robotics / Shared autonomous vehicles / Future air mobility / Space / Modular construction	Shared autonomous vehicles at 34% revenue CAGR, 69% market-cap CAGR; space at 7–10%; robotics 13–23%	Highly variable	Mostly upper-bound
New bio-frontiers	Obesity drugs / Non-medical biotech	Obesity drugs at 38% revenue CAGR, 35% market-cap CAGR (above scenario); non-medical biotech still nascent	9–38%	—

Eight of the 18 arenas are tracking near or above the upper bound of MGI’s 2040 trajectory: AI software & services, shared autonomous vehicles, cloud services, EVs, digital advertising, cybersecurity, semiconductors, and space. The other 10 are tracking middle-bound or are in earlier S-curve stages.

Operational mechanics — the omniscaler thesis in detail

The omniscaler concept is the report’s most novel contribution. MGI distinguishes them from hyperscalers (cloud-infrastructure providers) by their cross-arena footprint and mutual reinforcement of capability bundles:

What an omniscaler is (definition)

• Top-30 global spender on R&D + capital expenditures in 2024.
• Earning revenue in three or more future arenas in 2024.

Both filters together yield nine firms (clustered): Amazon (cluster — Amazon + Blue Origin + Prometheus), Tesla/X (cluster — Tesla + SpaceX), Alphabet cluster, Microsoft cluster, Meta, Apple, Samsung, Alibaba, Huawei.

Arena footprint by omniscaler

Omniscaler	# arenas	Arena revenue 2025 ($B)	Total revenue 2025 ($B)	Arena revenue share (%)	Capex + R&D 2025 ($B)
Alphabet cluster	9	380	400	95	150
Amazon cluster	8	720	720	100	240
Tesla/X	5	150	300	50	100
Microsoft cluster	5	110	110	100	20
Meta	4	200	200	100	130
Apple	3	25	500	5	50
Samsung	7	75	375	20	70
Alibaba	5	130	144	90	25
Huawei	5	10	100	10	25
Total	—	~1,800	~2,850	—	~810

Why the omniscaler advantage compounds

• Reusable infrastructure — a single capex-bet on data-centre capacity, custom silicon, or supply-chain logistics serves multiple arenas simultaneously.
• Data network effects — first-mover data accumulated in one arena (e.g., Alphabet’s search/YouTube engagement data) becomes training/feedback data for another (e.g., AI software, autonomous-driving).
• High risk appetites — the operating-cash-flow base of $700B/year in aggregate lets these firms run multi-year unprofitable bets in robotics, space, future air mobility.
• Top-talent attraction — the cross-arena employment surface offers the most ambitious technical roles in any single labour market.

Revenue gap to other arena players

• Omniscaler revenue per arena: ~$200B average.
• Other arena player revenue per arena: ~$10B average.
• ~20× per-arena revenue advantage.

This is the structural cliff that the report names as the new competitive reality: arenas are no longer level playing fields where a vertical specialist can win; they are increasingly fields contested by horizontal multi-arena platforms.

Convergence with the wiki’s existing corpus

Source	What it adds relative to this MGI report
AI Index 2026	AI-side empirical anchor at the model / research / policy / capability layer. MGI Arenas is the economy-side empirical anchor at the firm / industry / capital-flow layer. Both released spring 2026; together they triangulate the AI wave from two complementary vantages. AI Index reports e.g. SWE-bench Verified saturated and Gemini Deep Think IMO gold; MGI reports the same wave as $11T market-cap accretion in the AI foundation cluster
Sternfels 2026 (HBR IdeaCast)	Firm-as-vendor self-narrative. McKinsey-the-consultancy adapting (60K-strong = 40K humans + 20K agents; outcome-underwriting business model). This MGI report is firm-as-economy-mapper external account — same firm, two different self-presentations within a 6-week window
Levin)	Practitioner playbook for client companies — how to deploy AI inside the firm to capture 20% EBITDA uplift / $3 of EBITDA per $1 invested. This MGI report is the macro justification for the playbook’s urgency — clients sit in or near arenas where escalatory investment compresses decision windows and removes incumbent options
Future Today Strategy Group)	Process lens to MGI’s outcome lens. FTSG describes how convergence happens (four rules + seven enabling conditions + Webb’s 10-step methodology); MGI quantifies where convergence has materialised in the 2022–25 firm-level data. MGI’s three-ingredient arena-creation potion ⊂ FTSG’s seven enabling conditions — the four remaining FTSG conditions (cost-of-testing falling; economic systems slow to reorganise; energy capacity rising; legitimacy eroding-at-pace) surface in MGI as swing factors / time-to-build constraints rather than as arena-creation ingredients (i.e., uncertainty layer rather than creation layer). MGI’s omniscaler thesis is the empirical operationalisation of FTSG’s third rule (“convergences redistribute power and value”) — nine firms, ~$700B operating cash flow, ~20× per-arena revenue advantage. The two reports compose into a foresight-to-capital-allocation pipeline: FTSG-output (convergence scenarios) → MGI-output (arena-radar capital allocation). See strategic-foresight for the full side-by-side bridge with section-level alignment
enterprise-ai-adoption concept page	Adds $11T market-cap accretion in AI foundation cluster since 2022 as the firm-side scale of the AI adoption race the concept tracks qualitatively; adds the omniscaler thesis as a candidate new sub-concept
generative-ai concept page	Adds $1.5T–$4.6T revenue projection by 2040 for AI software & services (17–25% CAGR); adds the AI foundation cluster as MGI’s framing for the three-arena bundle (semiconductors + cloud + AI software)
strategic-foresight concept page	Adds the three-ingredient arena-creation potion as a reusable foresight heuristic (tech/business-model step change + escalatory investment + large addressable market); adds the arenas radar diagnostic (Exhibit 17) as a foresight-into-strategy translation tool
ai-employment-effects concept page	Adds the omniscaler top-talent attraction as a concrete labour-market reshaping vehicle; the arena/non-arena revenue divergence implies talent reallocation across the economy on a scale the wiki has tracked qualitatively

Linked entities and concepts

• McKinsey Global Institute — the publisher; the research arm of McKinsey & Company established 1990. MGI directors include Shubham Singhal (chair), Chris Bradley (this report’s co-author), Tanguy Catlin, Kweilin Ellingrud (this report’s co-author), Sylvain Johansson, Nick Leung, Olivia White. MGI partners include Arvind Govindarajan, Mekala Krishnan, Anu Madgavkar, Jan Mischke, Jeongmin Seong. The wiki’s McKinsey & Company entity page should be updated with an MGI sub-section linking to this source.
• Kevin Russell — lead co-author. First wiki appearance; entity page deferred (single source).
• Chris Bradley — co-author and MGI director; moderated the virtual-event panel. First wiki appearance; entity page deferred.
• Naveen Sastry — co-author. First wiki appearance; deferred.
• Suhayl Chettih — co-author. First wiki appearance; deferred. Not a panelist on the live-event companion (earlier wiki framing has been corrected).
• Kweilin Ellingrud — co-author, MGI director. First wiki appearance; deferred.
• Natalya Goryunova — co-author. First wiki appearance; deferred.
• Brendan Gaffey — live-event panelist (not in the PDF author list). McKinsey TMT practice lead; PhD electrical engineering. First wiki appearance; deferred.
• Gayatri Shenai — live-event panelist (not in the PDF author list). McKinsey New York senior partner; data-center / cloud authority; women-in-tech advocate. First wiki appearance; deferred. ASR-rendering variability: Shennai / Shanai / Guyry / guy tree across the transcript; YouTube description’s Shenai is canonical.

Omniscaler firms — all named in the report; most are first-wiki mentions or strengthened mentions:

• Alphabet / Google — already have entities; this source strengthens their omniscaler-cluster framing.
• Amazon — already has an entity; this source adds the Amazon cluster framing (Amazon + Blue Origin + Prometheus) — largest by 2025 arena revenue at $720B.
• Microsoft cluster — first wiki entity-relevant mention; deferred.
• Meta — first wiki entity-relevant mention; deferred.
• Apple — first wiki entity-relevant mention; deferred.
• Tesla / SpaceX (Tesla/X cluster) — first wiki entity-relevant mention; deferred.
• Samsung — first wiki entity-relevant mention; deferred.
• Alibaba — first wiki entity-relevant mention; deferred.
• Huawei — first wiki entity-relevant mention; deferred.

Candidate new concept pages (deferred until second source):

• omniscalers — MGI’s coined term. Would need a second wiki source independently using or operationalising the construct before promotion.
• physical AI — defined in the report as systems that “sense, think, and act in the real world” and named as a subset of hard tech (robotics + SAVs + future air mobility). Single-source; defer.
• agentic commerce — named in the at-a-glance summary as “creating new competitive fronts”. Single-source; defer.
• technology sovereignty — referenced in the introduction as a geopolitical force. Single-source; defer.
• arena (industry-as-arena) — MGI’s central analytical construct. Would benefit from a wiki concept page once a second corroborating source lands.

Dangling (single-source mentions, deferred): Kevin Russell, Chris Bradley, Naveen Sastry, Suhayl Chettih, Kweilin Ellingrud, Natalya Goryunova, Brendan Gaffey, Gayatri Shenai, Shubham Singhal, omniscaler concept, physical AI concept, agentic commerce concept, technology sovereignty concept, arena concept.

Body-wikilink coverage for all relationships: targets is satisfied through the Convergence table and the TL;DR.

Open questions raised

• Will AI foundation valuations be supported by sustained ROIC? The report names this as “one of the biggest open questions in business today.” — the AI-foundation market-cap accretion runs at 142% CAGR for AI software & services while invested capital is climbing fast; whether the ROIC eventually exceeds cost of capital is unsettled. The wiki should track this alongside the ai-benchmarks capability-reliability-gap thesis from HF Agentic Evals — the two are mutually informative (capability without reliability won’t unlock the enterprise spend that justifies the market caps).
• Omniscaler concept stability. A single-source MGI coinage; the construct may or may not propagate across the wiki’s other strategic/economic sources. Second-source watch.
• Chinese omniscaler trajectory. The report names three Asian omniscalers (Alibaba, Huawei, Samsung) and frames China as gaining ground on revenue; whether the US-China share gap closes by the end of the 2020s is open and politically charged.
• The chatbot disintermediation thesis for the open web — “chatbots that can return detailed, useful responses to complex user queries are disintermediating the open-web journey, reducing click-through traffic from searches”. The wiki has adjacent material in Spiegel 2026 and Nika 2025 — the convergence is “AI is climbing up the product-development funnel” (Spiegel) and reducing open-web traffic (MGI). Worth a synthesis when a third corroborating source lands.
• Physical AI as a sub-arena. The report’s physical-AI subset of hard tech (robotics + SAVs + future air mobility) is a candidate concept that would let the wiki separate “AI in software” (foundation models, agents, evals) from “AI in physical systems” (sense-think-act with sensors and actuators). The wiki currently treats both under ai-agents / generative-ai; this could warrant a future split.
• Modular construction is still nearly zero growth. The report shows modular construction at 0% revenue CAGR against a 6–10% scenario — the only arena where the 2022–25 actual missed the lower bound substantively. Worth tracking whether this corrects.
• Permitting speed / time-to-build / grid access as a competitive variable. The report names this as “a decisive factor, alongside technological advances” — a non-technology-side competitive lever the wiki hasn’t yet tracked. Candidate for the enterprise-ai-adoption concept page or a new infrastructure/logistics-side concept.

Related sources

• 2026-04-30-ai-index-report-2026 — AI-side empirical anchor; pair with this for the full AI-economy picture.
• 2026-02-09-sternfels-mckinsey-survive-ai-and-reinvent-consulting — McKinsey-firm self-narrative; pair with this for the firm-as-analyst-of-its-own-customer-base view.
• 2026-05-03-rewired-second-edition-sample — McKinsey practitioner playbook; this MGI report is the macro context that justifies its urgency.
• 2026-05-12-mgi-virtual-event-race-takes-off-next-big-arenas — companion live-event video (transcript not captured; metadata only).
• enterprise-ai-adoption — concept page extended with the AI-foundation $11T market-cap accretion and omniscaler thesis.
• strategic-foresight — concept page extended with the arena-creation potion + arenas-radar diagnostic.
• generative-ai — concept page extended with the AI-foundation cluster framing and 2040 revenue projection.