The State of AI in Monaco — 2026

Monaco has more AI activity in 2026 than at any point in its history — a sovereign cloud, a Digital 2030 vision, a data-protection regime with real teeth, active public funding, and a growing bench of firms shipping real production systems. It also has more AI theatre than at any point in its history: vendor decks with invented percentages, “audits” that sell what they were built to sell, and a quietly growing pile of pilots that died in month three.

This paper is our attempt to separate the two. It is written for the founder, COO, or compliance officer who needs to make an AI decision this quarter and is tired of white papers that answer none of the real questions.

The evidence rules we hold ourselves to: every specific number is either sourced to a public document or dropped; every sector observation is either attributable or labelled as our field view; where we could not verify something, we say so in section 11. The approach that shapes the whole paper is described in The Monaco AI Method.

This is a field report, not legal advice. Where laws or regulatory instruments are cited, they are described in general terms; specific matters require qualified Monaco counsel.

Signed, Guillaume Delachet — BSS, Monaco.

• If you’re a founder or COO — start at section 9 (what’s actually shipping) and work backwards.
• If you’re a compliance officer or DPO — start at section 3 (the regulatory perimeter) and read the sector chapters after.
• If you want the whole map — read straight through. It’s about a 20-minute read.

Monaco is small, dense, and lopsided in ways that matter for AI. Roughly 39,000 residents on about 2 km² of ground, and an economy that concentrates more revenue per square metre than almost any comparable jurisdiction. The Principality’s own statistics office, IMSEE, publishes a sector breakdown each year in its Observatoire Économique — and it is that breakdown, not any global benchmark, that determines which AI use cases are actually worth building here.

Financial and insurance activities are the largest single contributor to Monaco’s value added, sitting alongside a concentrated real-estate and construction cluster, the accommodation and food-service sector that carries the luxury hospitality economy, and a wholesale-and-retail category that in Monaco skews to high-end rather than volume commerce. Yachting does not get its own IMSEE line, but its footprint shows up across transportation, retail, and specialised services, and the Monaco Yacht Show remains the single largest sector event in the calendar. Employment follows the same shape: private-sector jobs cluster in finance, hospitality, retail, construction, and business services, and the labour force is majority non-resident, commuting daily from France and Italy.

Two facts flow from that shape and set up the rest of the paper. First, there is no meaningful Monaco industrial base to automate. No manufacturing floor, no logistics fleet, no call-centre estate. The AI conversation in Monaco is almost entirely about knowledge work — reading documents, writing to clients, checking compliance, personalising service, managing relationships. That narrows the field of viable pilots sharply and rules out large swathes of the generic “AI transformation” playbook that gets pitched to mid-market European firms.

Second, the client base for most Monaco businesses lives outside Monaco. A Monaco wealth manager’s book is international. A Monaco real-estate agency’s buyers arrive from a dozen countries. A Monaco hotel’s guest list is majority foreign. Cross-border data flows are the default, not the exception — which puts every AI project in scope of the regulatory stack we cover in section 3 — and discovery is increasingly happening in AI-mediated channels aimed at international audiences, which is why section 8 exists at all. For a longer-form companion to this report, our Monaco AI business guide sits alongside it.

The sector concentration also explains what you won’t find much of in Monaco AI in 2026. Very little agricultural AI, very little industrial IoT, very little logistics optimisation, and — despite the geography — surprisingly little maritime automation outside of the yachting services layer. There is a lot of document ingestion, a lot of multilingual client communication, a lot of KYC and AML augmentation, a lot of concierge and guest-experience work, and a growing amount of internal-tooling automation built on top of the standard SaaS stack most Monaco firms already run.

Monaco’s public AI infrastructure is real, but unevenly shipped. Some pieces are operational and load-bearing. Others are still slideware — a strategy document, a working group, an announcement. If you’re making a decision this quarter about where to put your data or which vendors to shortlist, that distinction matters more than the marketing.

The Extended Monaco programme, steered by the government’s Direction du Développement Numérique, is the umbrella under which most of what you hear announced sits — digital identity, e-government, sovereign infrastructure, and now an AI workstream. What has genuinely shipped: the MConnect digital identity, a sizeable share of state services online, and a Monaco-hosted cloud stack. What is aspirational: a coordinated national AI strategy of the kind France, the UAE, or Singapore have published. “AI” appears frequently in government communications; a full-fat strategy document with budgeted milestones does not yet exist in the public record at the time of writing.

Where the state has moved decisively is on infrastructure. Monaco Telecom, in partnership with Amazon Web Services, launched a sovereign cloud offering that keeps data physically within the Principality’s territory and under Monégasque jurisdiction — the mechanics are in Monaco’s sovereign cloud, explained and what it opens up for AI workloads specifically in sovereign AI comes to Monaco. For regulated verticals — banks, family offices, notaries, healthcare providers — this is not a marketing point, it’s a design constraint that changes which model providers, which storage architectures, and which SaaS vendors are legally viable. It is the single most important thing the state has shipped for AI.

The Digital 2030 vision names AI as a priority pillar alongside cybersecurity, connectivity, and digital inclusion. Read it as commitment, not calendar. And do not wait for a national AI strategy to make your own: firms that treated Extended Monaco as an enabling environment rather than a blueprint, and moved on their own workflows in 2024–2025, are the ones running production systems now.

Three regimes stack on top of any AI system a Monaco business deploys, and none of them can be ignored on the grounds that the others exist. Read this section in the sense of “here is the shape of the compliance floor” — the exact article-level scoping of any specific matter belongs with qualified Monaco counsel.

One — Monégasque data protection (Loi n° 1.565). The Principality’s modern data-protection statute, adopted in December 2022 and in force with staged provisions through 2023, is enforced by the Autorité de Protection des Données Personnelles (APDP). Its scope covers processing of personal data of individuals in Monaco; lawful bases (consent, contract, legal obligation, legitimate interest and the equivalents) determine whether a client dataset can be fed into a model at all; data-subject rights (access, rectification, erasure, and objection to solely-automated decisioning) must be honoured by any AI-facing workflow; and cross-border transfers carry their own regime, which is where most Monaco AI projects meet their first real friction. For a practitioner-level explainer of how the Law lands in day-to-day terms, see Law n° 1.565 in practice and data transfers from Monaco in 2026. Plain-English takeaway: you need a lawful basis to process, a documented map of what the model sees, and a defensible answer to “where does this data go and who can touch it.”

Two — the EU AI Act, extraterritorially. Monaco is not in the EU and not in the EEA. Regulation (EU) 2024/1689 — the AI Act — nonetheless reaches Monaco-established providers and deployers when the output produced by their AI system is used in the Union. A Monaco family office serving EU-resident clients, a Monaco law firm advising a French corporate, a Monaco hotel booking guests through EU platforms — all can fall inside scope. The Act’s “high-risk” classification triggers the most demanding obligations (risk management, data governance, human oversight, technical documentation, conformity assessment). The general-purpose model obligations began applying during 2025; the high-risk system obligations phase in during 2026 and 2027. Full guide: EU AI Act — what Monaco firms need and AI and data protection in Monaco. Plain-English takeaway: the EU AI Act reaches you through your customers and your vendors even though Monaco itself is outside the Union.

Three — sector overlays. Monaco’s financial intelligence and supervisory body — the Autorité Monégasque de Sécurité Financière (AMSF, the authority formerly identified as SICCFIN) — expects regulated entities to be able to explain any automated monitoring decision, which rules out opaque architectures for AML triage without a human-in-the-loop step. Professional secrecy binds lawyers, notaries, and doctors, and does not bend to a vendor’s assurance that a US-hosted model “doesn’t train on your data.” Labour law limits solely-algorithmic decisioning in employment contexts. And Monaco is a Council of Europe member state; the Council’s Framework Convention on Artificial Intelligence, opened for signature in September 2024, will layer further obligations as national signature and ratification proceed — signatory status should be verified against the Council of Europe treaty office register on the date you need it.

What actually bites today, versus what firms are preparing for. Today: the APDP is measurably more active than in the two years after Loi n° 1.565 took effect. Cross-border-transfer scrutiny is the live front. What firms are preparing for in 2027: the second wave of EU AI Act obligations, formal APDP practice specific to AI processing, and — likely — the first Monaco enforcement action citing both regimes together. The compliance floor is not optional and it is not a bolt-on; it is the design constraint that shapes which AI architectures survive contact with a regulator. If you want that constraint mapped against your specific stack, that is what our AI compliance and data protection engagements exist for. Law and guidance in this section reflect the state of public sources at the time of writing; APDP practice is young and EU AI Act obligations phase in through 2027, so this chapter carries a shorter shelf life than the sector chapters that follow.

Public funding in Monaco is not a marketing story. It is a small set of concrete programmes with concrete terms, and the question that matters for a founder or COO is which of them you can actually claim in the next 90 days, for what kind of AI work, and at what co-funding ratio. The honest answer is: more than you probably think, but each has terms that decide whether it applies to you.

The Fonds Bleu is the headline instrument for digital and environmental transformation projects. Verify the current-year co-funding ratio, eligible-expense definitions, and caps directly on the programme page before you scope, because they are revised annually and past-year figures circulate in vendor decks that are already out of date. The practical filter: your project has to be a real transformation, not a licence renewal dressed up as one. AI projects that fit tend to look like process redesigns with software, not “we bought Copilot.”

Monaco Boost and MonacoTech sit at the earlier end of the funnel — incubation and acceleration for startups, including AI-native ones. If you’re building an AI product rather than adopting one, this is where the cohort structure, workspace, and mentor access live. The eligibility is founder-shaped, not enterprise-shaped; a 40-person law firm buying an AI tool is not a Boost candidate, but a Monaco-registered founder building the tool the law firm buys is.

On the workforce side, the free AI training programmes under the Extended Monaco umbrella cover foundational literacy — what the tools are, how to prompt, where the risks sit — and are genuinely free for eligible Monaco businesses and residents. Useful for onboarding non-technical staff and defusing the “our people won’t use it” objection cheaply. Going deeper, LabImpact workshops are hands-on sessions structured around real use cases, closer to a practitioner format than a lecture — usually more per hour than a broad literacy course if you’re trying to build internal champions before a pilot.

Three practical notes before you apply to anything. First, the co-funding ratios and eligible-expense definitions are only meaningful if you read the current-year programme document. Do not build a business case on a percentage a consultant quoted you. Second, application timelines are real; Fonds Bleu in particular is not a same-week turnaround, and your project scope may need to be locked before you file. Third, these instruments compose — a training budget covered by the free programmes, plus a transformation project part-funded by Fonds Bleu, plus a MonacoTech-based vendor, is a stronger stack than any one alone, but the composition needs to be planned before you sign the vendor contract, not after.

What we could not verify at time of writing: current-year exact co-funding percentages and 2026-specific eligibility changes to Fonds Bleu. Those are set by the programme office and the linked page carries the live figures. We list this openly rather than guess. Same discipline runs through the rest of the paper.

Finance and wealth management is where Monaco’s AI adoption looks most active on the surface and most cautious underneath. The sector is a significant share of the Principality’s value added and is regulated end-to-end — banking supervision under the Franco-Monégasque monetary framework (ACPR exercises delegated prudential supervision), anti-money-laundering oversight by AMSF, and Loi n° 1.565 governing every client file that moves through a model. In practice this means firms are shipping AI in the places where the compliance envelope is clearest, and holding back where it isn’t.

Our field view: the three live workflows we see most often in Monaco banks, multi-family firms, and independent asset managers are portfolio reporting, KYC/AML augmentation, and client-communication drafting.

Portfolio reporting is the highest-leverage of the three: ingesting PDFs from custodians, normalising positions across banking relationships, and producing a first-draft quarterly report that a portfolio manager edits rather than writes. KYC/AML augmentation is narrower — models classifying adverse-media hits, summarising beneficial-ownership chains, and prioritising AMSF-relevant flags for a human reviewer to confirm. Client-communication drafting is the least glamorous and often the most valuable: multilingual first drafts for market commentaries, meeting recaps, and prospect follow-ups, in the four or five languages a Monaco relationship manager typically works in.

The workflow to avoid, and the one we see failing most often, is putting AI anywhere near investment decisioning. Not because the models can’t produce a view — they can — but because the audit trail a Monaco regulated firm needs for a discretionary decision is not something today’s model outputs give you. The recurring pattern we see: LLM-assisted allocation tooling piloted, then quietly parked when the firm’s compliance committee cannot sign off on the reproducibility of its recommendations. That failure mode is not technical. It is evidentiary.

The compliance gotcha, and it is genuinely a trap: over-automated AML triage. Guidance on automated monitoring is unambiguous that the regulated entity remains responsible for the substantive review of flagged transactions — a model can prioritise, a human must decide. Configuring the AI layer to auto-close low-score adverse-media hits without human review is exactly the surface an inspection will land on. Design the workflow so a human clicks “reviewed” on every flag the model touches, or you are building the wrong system.

The right shape for a Monaco finance pilot in 2026 looks boring: sovereign or EU-hosted deployment, no US model provider in the data path unless a specific derogation applies, an explicit register of which client data the model sees, and a human sign-off on every output that leaves the firm. Boring is what passes an APDP notification and an AMSF visit. For the operator-level pattern, our AI for finance in Monaco page sets out the workflow shortlist we run with wealth managers and banks; family offices share most of the architecture but push harder on document ingestion, which we cover in the next section.

Monaco’s family offices, real-estate agencies, and yachting firms share a workload that AI is genuinely good at: high-volume document ingestion, multilingual client communication, and pattern-matching across unstructured records. They also share the same objection — client confidentiality is the business — which is why sovereign or EU-hosted deployments have quietly become the default architecture across all three. The firms that solved the hosting question early are shipping. The firms still waiting for a vendor to solve it for them are stalled.

Family offices. Our field view: the two workflows worth piloting first are consolidated reporting across custodians (ingesting statements from ten to thirty banks into a single quarterly view for the principal) and inbound-document triage (contracts, KYC packs, tax notices routed and summarised). The workflow to avoid, still, is any form of investment recommendation — the liability exposure under both Monégasque professional duty and the EU AI Act’s high-risk classifications for financial services is not worth the marginal productivity. The compliance gotcha: family-office data is often not the family office’s data. Beneficial-owner information, trust documents, and underlying-client records carry contractual confidentiality obligations that flow through to any AI vendor as a sub-processor — read the DPA before the demo, not after. Deep-link: AI for family offices in Monaco.

Real estate. Monaco’s real-estate market runs on a small number of high-value transactions across a multilingual buyer pool — French, Italian, English, Russian, Arabic, increasingly Mandarin. Our field view: the two live workflows are multilingual listing generation and translation (a listing produced once, adapted cleanly across five languages in an afternoon) and client-correspondence drafting (initial replies to serious enquiries, drafted for an agent to review and send). The workflow to avoid is automated buyer qualification — the market is too small and too relationship-driven for a scoring model to add value it doesn’t first destroy. The compliance gotcha: property transactions in Monaco fall under AMSF AML scope, and any AI touching KYC evidence needs the same treatment as bank-facing tooling. Deep-link: AI for Monaco real estate.

Yachting. The Monaco Yacht Show economy sits on top of a year-round operational base — brokerage, management, crew, charter — that is drowning in documents: MLC certifications, flag-state paperwork, charter contracts, insurance renewals, provisioning lists in three languages. Our field view: the two workflows to pilot are charter-enquiry triage (matching an inbound request to available yachts and drafting the first-round proposal) and compliance-document extraction (pulling structured data from certificates and contracts into a management system). The workflow to avoid is anything that touches HNW client identity data without a signed DPA naming a Monaco or EU sub-processor — a captain’s contact book is not training data. The compliance gotcha: yachts move across jurisdictions weekly, and the applicable data-protection regime shifts with them; a system built for Monaco alone will fail its first Mediterranean charter. Deep-link: AI for Monaco yachting.

Across all three verticals, the pattern is the same. The winning firms scoped narrow, hosted local, and put a human on the last mile. The firms that lost time bought horizontal tools built for a US SaaS company and spent six months trying to make them respect a Monégasque confidentiality obligation they were never designed to see.

Monaco’s hospitality economy sits on a small number of high-value guests who expect to be recognised — by name, by preference, by language, by past complaint — the moment they walk in. That is the workflow AI is quietly reshaping in 2026. The palaces, boutique hotels, and Michelin-starred rooms that anchor the Principality don’t need more marketing automation; they need better guest memory, faster multilingual response, and back-office capacity that doesn’t collapse during Grand Prix week or the Yacht Show.

Our field view on where AI is actually shipping in Monaco hospitality: pre-arrival guest research pulled from CRM and public sources into a one-page briefing for the concierge team; multilingual chat for booking, room-service, and destination questions in French, English, Italian, Russian, and increasingly Arabic and Mandarin; and back-office automation on the least glamorous parts of the operation — housekeeping schedules against real-time occupancy, procurement reconciliation, group-booking paperwork, revenue management. None of it is exotic. All of it recovers hours per week that the property either bills at concierge rates or reinvests in guests who notice.

The overspend to name honestly: fully autonomous concierge chatbots deployed guest-facing without a human handover path. Properties that have tried this walked it back — the failure mode is a single high-net-worth guest getting an unhelpful or off-brand reply and telling five peers about it at dinner. In Monaco’s guest network, one bad interaction propagates faster than any recovery campaign can catch it. The pattern that works is AI-drafted, human-approved for anything touching a named guest, and fully autonomous only for anonymous top-of-funnel queries. Read the fuller argument on our luxury hospitality and retail and hospitality pages.

Luxury retail in Monaco sits at an unusual intersection. Foot traffic is a small share of revenue; a growing share comes from clients who researched the piece on their phone in Dubai, Shanghai, or São Paulo before ever walking into the store. That research is now happening inside AI Overviews, ChatGPT, and Gemini as much as inside Google’s blue links (we cover the mechanics in section 8). The pilot pattern that works: unify the client book with post-visit follow-up drafting, translate and enrich product copy at catalogue level rather than page-by-page, and instrument what AI-driven traffic is actually arriving on the site. The overspend to avoid: replatforming an entire e-commerce stack on the promise of “AI-native” commerce before the underlying product data is clean. AI amplifies the data you feed it, including the mess.

Search still routes most of Monaco’s international demand — the buyer researching a Monaco family office from London, the charter client comparing yachts from Dubai, the couple planning a Monte-Carlo weekend from Milan. What’s changed in 2026 is that “search” no longer means ten blue links. Google’s AI Overviews now sit above the results for a growing share of commercial queries, ChatGPT and Gemini are used as first-pass discovery tools for high-consideration purchases, and Perplexity is quietly becoming the default for research-heavy briefs. For a Monaco brand whose customers are almost all outside the 2 km², this is a structural shift, not a marketing wrinkle.

The mechanics matter. AI Overviews are generated from a synthesis pass over indexed pages — your site still needs to be crawlable, structured, and factually clean, but the way it is read has changed. Long, well-sourced, entity-rich pages tend to be pulled into overviews; thin brochureware tends not to. Google’s own Search Central documentation on AI features is explicit that there is no separate “AI SEO” — the same technical fundamentals apply, but the ceiling on winning is now set by clarity of information, not by keyword density. For a Monaco luxury brand, that means the “about the property,” “about the fleet,” “about the family office” pages used to be treated as afterthoughts are now the pages that decide whether an AI names you.

Search Console gives you the signal, if you know where to look. Impressions from AI Overviews are folded into standard reporting rather than broken out cleanly — meaning the useful measurement isn’t “AI traffic” as a line item, it’s the shift in click-through rate for informational queries where an overview now sits above your result. If your CTR on high-intent queries has quietly halved over the last twelve months while impressions held, an overview is almost certainly eating the click. The measurement approach: Google AI visibility in Search Console.

In luxury retail and hospitality, the queries that matter are increasingly conversational and comparative — “best Monaco hotel for a young family in August,” “quiet Monte-Carlo restaurants with a terrace” — and AI systems answer them directly, so brand mentions in reputable third-party sources now matter more than they did when a Google result page listed everything. For product discovery, AI-driven surfaces reward complete, accurate product feeds and structured data more than they reward paid placement — a shift that favours operators with clean catalogues over those with big budgets. Our field view: most Monaco luxury sites are currently under-indexed by AI systems relative to their actual authority, because their content was written for humans skimming, not for machines synthesising. The fix is not more content. It is better-structured content, and less of it. The entry point for the work is AI SEO for Monaco.

Three categories dominate real Monaco buys in 2026: assistants (chat interfaces wrapping your data), automations (workflows that run without a human clicking through), and agents (systems that decide and act across steps). Everything else is a variant of these three. If a vendor pitches you a fourth category, ask which of the three it actually is.

Assistants are the safest starting point and the easiest to overspend on. A well-scoped internal assistant — grounded on your policies, client files, or product catalogue — is typically an eight-to-fourteen week build, with monthly hosting and model costs sitting in the low hundreds to low thousands of euros depending on volume. The failure mode is scope creep: teams try to build “one assistant for the whole company” instead of one assistant for one job. The ones that survive month six answer a single question well — “what’s our policy on X?”, “summarise this client file,” “draft a response to this enquiry” — and are expanded only after they’re used daily. See AI assistants and chatbots for scoping patterns and AI chatbots for customer service for the customer-facing variant.

Automations are where the real ROI is, and where Monaco firms consistently underinvest. Document ingestion, KYC packet assembly, multilingual email drafting, reporting pipelines, meeting-note routing — these are unglamorous, high-frequency workflows where a well-built automation pays back in weeks, not quarters. The failure mode is automating a broken process: if the underlying workflow is inconsistent, you will ship an automation that produces inconsistent output faster. Standardise first, automate second. See AI automation and custom AI tools for the patterns we ship most.

Agents are the category with the widest gap between marketing and reality. A true agent — one that plans, calls tools, handles failure, and reports back — is a real engineering commitment, typically a three-to-six-month build for a production deployment with observability and guardrails. Voice agents (concierge, phone triage, appointment handling) are a distinct sub-category with their own latency and telephony constraints; see AI voice agents. The failure mode we’ve seen most often: shipping an agent without an audit log, then being unable to explain to a compliance officer or a client what the agent did on their file. If you can’t reconstruct the decision, you can’t defend it. For text agents, AI agents and the trend piece AI agents for Monaco businesses — plus agentic commerce comes to Monaco if you sell online — are the relevant reads. For anything touching sovereign data or bespoke architecture, custom AI is the natural entry point.

The antidote to vaporware pilots is structure. Our default cadence is 90 days: weeks 1–2 to scope (one workflow, one owner, one success metric agreed in writing), weeks 3–6 to build, weeks 7–10 to run in production alongside the existing process, weeks 11–13 to decide — keep, iterate, or kill. The kill option is not a failure; a pilot that produces a clean “no” in 90 days is worth more than one that limps for a year. If you can’t name the success metric before week 3, don’t start the build. Before any of this, four minutes with our AI Readiness Quick-Check will tell you whether the workflow you have in mind is pilot-ready or dependency-blocked.

Every report like this either hedges its forecasts into mush or fires off predictions it will never be held to. We are going to do neither. Here are five bets we are willing to be wrong about on record, dated 2026, to be re-scored in the 2027 edition of this paper.

Bet one — the APDP will notice AI. Our bet: at least one Monaco financial firm will receive a formal APDP notice tied to an AI-adjacent process before the end of 2027. Not a headline sanction — a formal mise en demeure or equivalent, published or leaked. The APDP has been building doctrine quietly since Loi n° 1.565 took effect, and the pattern in comparable jurisdictions (CNIL in France, Garante in Italy) is that the first AI-adjacent enforcement action lands within roughly three years of a modernised regime taking hold. Monaco is on that clock now.

Bet two — sovereign hosting becomes a checkbox. Right now “hosted in Monaco or the EU” wins deals in finance and family offices. By late 2027 every serious vendor will offer it, and the differentiator will move one layer up — to model provenance, audit logs, and human-in-the-loop guarantees. Firms that built their AI story on hosting alone will need a second act.

Bet three — voice overtakes chat. Voice agents will overtake chatbots as the highest-ticket AI category sold to Monaco hospitality and concierge operators. Chat is a solved problem and the buyer knows it. Voice — multilingual, always-on, integrated with reservation and CRM systems — is where the willingness to pay actually lives in a market whose guests expect to be spoken to, not ticketed.

Bet four — cutting against our own interest. Half of the “AI readiness audits” sold in Monaco in 2026 — ours included, on our worst days — will turn out to have been unnecessary. The audit product exists because the market cannot yet self-diagnose. As tooling matures and reference architectures become common, the diagnostic gap narrows. By 2027 the honest engagement for many mid-sized firms will be a two-week scoped pilot, not a discovery phase. This is a bet we hope loses us revenue.

Bet five — the cautionary tale. At least one high-profile Monaco AI pilot will fail publicly enough that it enters local business folklore. Every ecosystem needs its cautionary tale — the deployment that got shut off, the vendor that got sued, the compliance letter that made the news. Monaco has been spared this so far because volume has been low. Volume is no longer low. Somebody will be the example, and the market will get more disciplined the day after.

We will re-score all five in the 2027 edition of this report — win, lose, or unresolved — with the same discipline we’ve tried to bring to the rest of the paper. If you think we’re wrong about any of them, we would rather hear it now than read it in a competitor’s white paper next year.