How the World’s Biggest Hotel Chains Are Actually Using AI

For the past two years, the major hotel groups have all said roughly the same thing about artificial intelligence: it will transform booking, personalisation, distribution and operations. What is genuinely useful for hotel owners, asset managers and operating leaders is to look past the slogans and ask a sharper question. What has each chain actually built, what has it cost, what is it delivering, and what is coming next?

This piece works through eight global players – Marriott International, Hilton, IHG Hotels & Resorts, Wyndham Hotels & Resorts, Hyatt, Accor, Choice Hotels International and H World Group – using two lenses. First, a functional view: how AI is being used across operations, commercial, corporate and development. Second, a stakeholder view: what these deployments mean for owners, franchisees, GMs, commercial leaders, finance teams and guests. The sources are limited to company filings, earnings calls, official press releases and reporting from Skift, PhocusWire, Hotel Dive, Hospitality Net, Hotel Management and similar trade titles.

A few patterns are worth setting out before going chain by chain. The first is that almost every group is now positioning generative AI at the front of the funnel, with conversational search and direct integrations into ChatGPT, Gemini and Claude. The second is that the more durable value, in dollars, is still coming from older AI use cases – revenue management, energy management and guest engagement messaging. The third is that owner economics, not guest delight, has become the dominant talking point on earnings calls. AI is being sold to franchisees as a labour-saver and a revenue tool.

Marriott has set the pace on capital commitment. The company guided to technology spend of US$1 billion to US$1.2 billion in 2024 – its highest tech outlay ever – and confirmed on its Q4 2025 earnings call that 2026 capex of US$1 billion to US$1.1 billion would direct up to 40 per cent to digital and corporate systems. CIO and CTO Naveen Manga has described the architecture as a model-agnostic, cloud-native platform with an agentic layer, with around ten enterprise AI use cases planned for 2025 (CIO Dive, January 2026).

On the commercial side, the most consumer-visible product is the generative AI search built for Homes & Villas by Marriott Bonvoy, developed with Publicis Sapient on Microsoft’s Azure OpenAI Service and launched in test in March 2024. Publicis Sapient has reported that users engaging with the AI search were twice as likely to save properties to favourites, and that gen AI product rollout times for other Marriott brands fell from one year to three months. CEO Anthony Capuano told analysts on the Q4 2025 call that natural language search will be deployed across Marriott.com and the Bonvoy app in the first half of 2026, and that Marriott is in early-stage work with Google on a hotel search experience for Google’s forthcoming AI Mode, plus an ad pilot with OpenAI (PhocusWire; Hotel Dive, February 2026).

The operations example most relevant to GMs is the Automated Complimentary Upgrade (ACU) system, rolled out in July 2025. The system algorithmically assigns elite room upgrades from 3pm the day before arrival, replacing what was previously several hours of daily manual work for front desk and rooms control teams. Marriott has positioned ACU explicitly as an owner-cost saving and a way to redirect labour to in-person service (View from the Wing; LoyaltyLobby, July 2025). On the corporate side, Marriott has automated its Scope 3 emissions data process. According to Smart Energy Decisions (November 2024), the move from Excel-based workflows to AI-led classification of general ledger spend cut both labour hours and reported Scope 3 emissions by between 5 and 10 per cent, an unusual case of AI directly affecting an ESG number.

For development, Marriott has been quieter publicly about AI in site selection, but the broader story is the same: it is building proprietary data scale through Bonvoy – 271 million members by year-end 2025, with 43 million added during the year (Marriott Q4 2025 earnings) – and using that data as the moat behind every other AI initiative.

The gap is real. Capuano himself framed agentic commerce on the Q4 2025 call by saying, “We’re pulling into the players’ parking lot. We’re not even in uniform or on the field.” Owners should read that as a warning that the customer acquisition pay-off from Marriott’s AI spend is still ahead, not behind.

Hilton’s AI story is older than most people remember and, on the operational side, considerably more proven. The LightStay platform, originally built in 2008 with IoT specialist ei3, now uses machine-learning algorithms to forecast every property’s energy, water and waste consumption and to track actual use against those predictive models. LightStay is a brand standard at every Hilton hotel. The company has reported that the system has delivered more than US$1 billion in cumulative cost savings since launch, with around 20 per cent reductions in water and energy and roughly 30 per cent reductions in carbon emissions and waste output across the portfolio (Hilton ESG disclosures; ei3 case study; edie). For owners, this is one of the few AI numbers in hospitality that has been third-party verified, in this case by KEMA and DEKRA.

On the guest side, Hilton’s earlier experiments – including the Connie robot concierge with IBM Watson in 2016 – have given way to a more focused generative AI play. In March 2026, Hilton launched the Hilton AI Planner, a generative AI digital concierge in beta on hilton.com. The Planner uses conversational language to help guests pick destinations, compare properties and explore amenities. Hilton has framed this as a “test-and-learn” deployment rather than a full rollout, and the timing puts it noticeably behind Marriott, IHG and Hyatt’s equivalents (Hilton press release, 10 March 2026; Hotel Dive).

For commercial leaders, Hilton’s strongest AI lever continues to be Connected Room and its accompanying IoT and machine-learning stack. The company has said publicly that AI and machine learning are used to optimise energy use when guests are not in the room – roughly 70 per cent of the time – while preserving comfort. Modernisation of loyalty program billing-cycle processing has, according to internal materials cited in third-party AI strategy reviews, cut billing-cycle processing time by around 85 per cent. Those numbers should be treated as company-reported rather than independently verified.

For GMs and franchisees, the more strategically interesting development is structural: in March 2026, alongside the AI Planner launch, Hilton confirmed leadership changes including the creation of a new chief technology officer role (SEC filing; Hotel Dive). That signals a heavier internal weight on technology delivery in the next two years. The gap to watch is the same as Marriott’s: Hilton has invested for two decades in the plumbing – converged networks, IoT, data – but its consumer-facing generative AI tools are arriving later than competitors’. Owners with Hilton-branded properties should expect the next 18 months to be about catching up on the front end while continuing to monetise the back end.

IHG’s AI strategy sits on top of a longer revenue management pedigree than any of its peers. Holiday Inn launched HIRO – its first automated revenue optimisation system – in the early 1990s, and IHG introduced price optimisation into the descendant PERFORM system in 2008, which the company has previously credited with roughly a 3 per cent revenue lift, or about US$300 million per year at the time (SAS conference paper, 2018, citing Revenue Analytics). That heritage is now wrapped into IHG Concerto, the cloud-based hotel management platform that combines the Guest Reservation System, the proprietary Revenue Management System and, over time, property management and point of sale.

In April 2024, IHG announced an expanded partnership with Google Cloud to build a generative AI travel planner inside the IHG One Rewards mobile app, using Vertex AI and Gemini. The product launched in the second half of 2024 with examples ranging from “Give me four late-night dinner options near the InterContinental London Park Lane” to “Does the Kimpton Shinjuku Hotel in Tokyo allow pets?” (IHG press release; Skift, April 2024).

The 2025 results filed with the SEC make the trajectory explicit. The Annual Report and Form 20-F 2025 describes “game-changing tech” and AI advantages across the enterprise, including a revenue management system “incorporating data science, machine learning and forecasting tools” and AI-driven enhancements to the IHG One Rewards mobile app, which reached nine million downloads in 2025. CFO Michael Maalouf told analysts on the Q4 2025 earnings call that the new RMS is fully rolled out, an AI-enhanced CRM is planned for 2026, and that AI is being used to automate routine internal tasks. Skift reported in February 2026 that IHG is restructuring its hotel data into modular, machine-readable formats and launching a new digital content platform specifically designed to be consumed by AI agents. In January 2026, IHG named Wei Manfredi senior vice president of AI and architecture – a newly created role (Hotel Technology News).

For owners and franchisees, the most relevant promise is that Concerto’s attribute-based pricing will eventually let guests choose – and pay for – specific room features such as floor, view or layout, with prices adjusting accordingly. IHG has said this will be available only to IHG One Rewards members booking direct, strengthening the direct channel. The honest caveat is that IHG declined to give RevPAR uplift guidance when first describing this functionality, and has continued to talk about Concerto in terms of capability rather than quantified return (Hospitality Technology, 2019; IHG Annual Report 2025). For revenue managers, the proof point will be in 2026–2027 pricing and conversion data.

Wyndham has been the most aggressive of the franchisor-led groups in pushing AI through to franchisee P&Ls, and the metrics it has put on the record are unusually specific. Since 2018, the company has invested more than US$450 million in technology (Wyndham investor materials, May 2026). Its flagship guest-facing platform, Wyndham Connect – powered by Canary Technologies – sends proactive guest messages, handles mobile check-in, runs AI-generated responses to inbound questions and supports voice assistance and upsell. Wyndham Connect PLUS, launched at the 2025 Global Conference, adds voice-based check-in and dynamic upsell.

The reported outcomes are substantial. Wyndham has stated that more than 5,000 hotels in North America and around 100 internationally are using Wyndham Connect, handling 4.6 million mobile check-ins, US$9 million in upsell revenue and 12 million AI messaging engagements, with a 300 basis points improvement in direct booking for hotels using AI voice (Fortune; Wyndham press releases). On the Q3 2025 earnings call, CEO Geoffrey Ballotti said the broader AI suite had handled more than half a million customer interactions, cut average handle time by 25 per cent and improved direct contribution by nearly 300 basis points for hotels using the tools fully (Hotel Technology News, November 2025). The most engaged hotels on Wyndham Connect generated more than US$60,000 in incremental revenue last year, with the highest-performing property exceeding US$200,000.

On distribution, Wyndham was the first major hotel company to go live on Anthropic’s Claude in 2025 and, in May 2026, launched a native ChatGPT app covering its roughly 8,400 hotels (PR Newswire; Hotel Dive). For franchisees, Wyndham has used PwC and Salesforce’s Agentforce to deploy agentic AI for brand standards management, cutting standard change review times from around 30 days to roughly one twentieth, according to PwC’s published case study.

Wyndham’s owner research is also worth flagging for industry leaders. The second annual Wyndham Owner Trends Report (December 2025, conducted by Wakefield Research) found that hoteliers cite data privacy and security (46 per cent), AI investment cost (42 per cent) and integration with legacy systems (40 per cent) as the three biggest barriers to wider adoption. Only 40 per cent of hoteliers are comfortable with AI making operating decisions without human oversight; 57 per cent want human oversight retained. That is a useful temperature check for any owner thinking about how far to push automation in 2026.

Hyatt has taken a deliberately enterprise-led approach. CEO Mark Hoplamazian told analysts on the Q4 2025 earnings call that Hyatt has been working on “AI enablement” for two years, has built a private-cloud infrastructure into which it in-licenses multiple LLMs from OpenAI, Microsoft, Google and Anthropic, and has identified use cases of which four have already been “executed as large-scale agentic platforms” (Investing.com transcript; Yahoo Finance summary).

On commercial, Hyatt has rebuilt the hyatt.com search engine using natural language and intent-based logic. Hoplamazian said multi-quarter data shows that the native and intent-driven search is improving booking conversion rates and revenue. The company also has a Hyatt app live inside ChatGPT and is studying user behaviour to refine search relevancy ahead of agent-to-agent booking (Constellation Research, February 2026). For revenue managers and commercial leaders, the implication is that Hyatt views agentic distribution as a near-term operating reality and is preparing its direct platforms to compete with – and ride – it.

On corporate, Hyatt has rolled out ChatGPT Enterprise across finance, marketing, operations and customer experience, used for content creation, owner and operator communications and accelerating digital platform development (Skift; PhocusWire, April 2026). Skift’s analysis noted that the rollout is unusual in hospitality precisely because hotel data is so fragmented; Hyatt spent two years building a unified data and AI layer first, before adding an employee-facing interface.

On the financial side, Hyatt undertook a corporate restructuring in late 2025 partly powered by AI-driven efficiency. The company cut about 5 per cent of its corporate workforce in October 2025 as it reorganised around five brand groups (Skift, November 2025). At the same time, an expanded Chase co-brand credit card agreement, announced alongside Q3 earnings, is expected to more than double EBITDA contribution from the credit card programmes by 2027. For owners and finance leaders, this is one of the cleanest examples of AI being directly tied to a headcount and cost-base decision at the brand level.

World of Hyatt membership reached 61 million by Q3 2025, up 20 per cent year on year, providing the data foundation for further personalisation. RevPAR grew 4 per cent in Q4 and 2.9 per cent for the full year (Hyatt earnings).

Accor’s CEO Sébastien Bazin made the boldest public claim of any major hotelier this cycle. Speaking at Viva Technology 2025 in Paris alongside Expedia CEO Ariane Gorin, Bazin said: “Before you come into the hotel, after you leave the hotel, you will be 90 per cent AI-driven” (PYMNTS, June 2025). Whether or not that figure proves accurate, it is a useful signal of where Accor is steering investment.

The most concrete recent deployment is the ALL Accor app in ChatGPT, launched in early 2026. The integration lets travellers use natural language inside ChatGPT to search Accor’s hotel network, view both public and ALL loyalty-member rates, see property details and complete bookings via redirection to the ALL Accor platform, in more than 20 languages (Accor press release; Hotel Technology News, February 2026). Hotel Technology News noted that, unlike Marriott or Hilton’s proprietary in-app concierges, Accor’s approach places its full brand portfolio and pricing directly into a third-party AI platform – a more aggressive distribution stance.

On customer service, Accor and Deloitte have implemented a global cloud telephony platform combined with an AI service layer for the luxury and lifestyle brands across more than 110 countries, focused on streamlining and automating customer service while retaining a human touch (Deloitte case study). Accor also runs a Travel Assistant built on Amazon Web Services to help guests through the booking process (Research and Markets briefing via Business Wire, July 2024).

Accor’s Group Chief Technology Architect, Khang Nguyen Trieu, has spoken candidly about the underlying constraint. In a 2024 interview with CIO.inc, he stressed that generative AI integrated with company knowledge is “only as good as the quality of the data at the source,” and that reconciling legacy systems and creating a single source of truth has been essential. That kind of public acknowledgement of legacy debt is more honest than most peers have been, and it reflects how much of Accor’s 2024–2025 spend has gone into the underlying data plumbing rather than glossy front-end products.

For commercial and finance leaders within Accor-affiliated hotels, the practical takeaway is that direct-channel AI products are arriving fast, but the underlying personalisation engine depends on the unified CRM platform Accor has been building, supported by big-data investments made before the gen AI cycle began. RevPAR uplift figures attributable to AI have not been disclosed.

Choice has the longest cloud-first track record among US franchisors. choiceEDGE, its proprietary cloud-based central reservation system, was first launched in 2018; the company committed to being 100 per cent on AWS in 2019 and completed its final data centre migration in 2024 (Choice press materials). ChoiceMAX, an AI-driven, mobile-first revenue management platform, has been live since 2021.

In April 2026, Choice announced an enterprise-wide AI deployment with AWS – the first major US hospitality provider to standardise on Amazon’s AgentCore platform for intelligent agents. The release describes AI applications across the value chain: guest discovery and booking, franchisee operations (including revenue management, maintenance and guest communications), and distribution and channel management (PR Newswire; Hotel Technology News, April 2026).

Owners and franchisees received a more tactical set of tools in May 2026. Choice Hotels Business Direct is a self-service digital booking platform for small and medium-sized businesses; EasyBid is an AI-enhanced group RFP tool; CHARLIE is described by the company as an AI-powered virtual “teammate” for staff working across core operating platforms; and RAISE is a rate management tool aimed at streamlining how owners manage pricing, rates and inventory (Hotel Dive, May 2026). For franchisees in the economy and midscale segments, where labour is the dominant cost line, these tools are pitched as productivity rather than personalisation plays.

For commercial leaders, Choice’s strongest published statement is its 10-year head start on cloud-native architecture, which the company argues lets it deploy AI improvements without forcing every property through a disruptive system overhaul. That is the most defensible part of the Choice narrative. The honest gap, as Hotel Technology News noted, is that the headline AI rollout follows on the heels of similar announcements from larger peers; Choice is competing for franchisee mindshare against Wyndham at the franchisor level and against Marriott and Hilton at the brand-equity level.

Choice has also strengthened its international franchise stack by selecting Mews as a cloud-based PMS option, integrated with choiceEDGE, to boost non-room revenue at its 1,258 international franchised properties (Skift, May 2025).

H World Group – formerly Huazhu – belongs in any global peer set, not as a regional footnote. As of December 2025, its worldwide network reached 12,858 hotels and 1,264,419 rooms, with full-year 2025 revenue of RMB25.3 billion (US$3.6 billion) and 2,887 hotels in the legacy Huazhu pipeline (H World Q4 2025 results, NASDAQ filing). That makes H World larger than Wyndham by room count and significantly larger than most Western competitors by loyalty membership, with the H Rewards programme accounting for roughly 73 per cent of room nights sold under legacy Huazhu in 2025 and 77 per cent of room nights sold through the group’s own channels (H World 2025 Annual Report, HKEX).

The technology stance is structural. Founder Ji Qi has an engineering and computer science background, and H World develops all core operating systems – PMS, RMS, CRM, procurement, staff management – in-house. Chief Strategy Officer Jihong He has described H World publicly as a “technology-enabled” hotel company; around 86 per cent of its rooms sit in limited-service brands such as Hanting, Ji Hotel and Orange, which is precisely the segment that lends itself most cleanly to automation (TTG Asia, October 2023; Skift, April 2025).

On operations, the group has built a digital system that compresses the gap between check-out and check-in, and a mobile application that automatically assigns cleaning or maintenance staff to specific rooms as they become available (ChinaTravelNews). The “Intelligent Laundry” function lets guests reserve and track laundry through the H World app. “Hello Huazhu”, an in-room voice assistant, controls lights, TV, air conditioning and window shades. The group has described its AI guest assistant as the first such tool in the China hotel industry. The 2025 Annual Report confirms the group runs an RMS and Cloud-PMS used for pricing management across all hotels except franchised properties.

For owners, the H World playbook is the most clearly industrialised AI deployment of any major chain. Modular pre-fabricated construction, paired with a centralised digital stack, lets the group open hotels faster and at lower cost than Western peers, which the company has tied to a stated ambition of doubling to roughly 20,000 hotels within five years (Skift, April 2025). For finance leaders, the metric that matters is direct distribution share: 77 per cent of legacy Huazhu room nights sold through owned channels, with a high-margin fee-based revenue mix.

The notable gap is disclosure. H World provides less granular AI ROI commentary than its US peers, and its Steigenberger / Legacy-DH international segment is on an earlier technology curve. Legacy-DH RevPAR rose 8.2 per cent in 2025, but the group’s stated 2026 priority is cost reduction and asset structure refinement in that segment – AI deployment internationally is, at this point, a forward commitment rather than a delivered result.

Reading across the eight chains, a clearer picture emerges of what AI is actually doing for each constituency in the hotel value chain. The patterns matter more than any single product launch.

Hotel owners and asset managers. The most defensible AI ROI today still comes from operational systems – Hilton’s LightStay (more than US$1 billion in verified savings since 2008), Marriott’s automated Scope 3 emissions process (5 to 10 per cent reductions in reported emissions and labour hours), Wyndham’s incremental Connect upsell revenue (averaging US$60,000+ at the most engaged hotels, with a top performer above US$200,000) and IHG’s revenue management lineage. Asset managers should be sceptical of vendor numbers that have not been third-party verified and should ask brands directly for portfolio-level RevPAR or GOP impact, not just adoption metrics.

Franchisees. Wyndham, Choice and IHG are the three groups making the loudest franchisee-facing pitch. Wyndham’s average handle time reduction of 25 per cent in call centres, its 300 basis point direct-contribution lift for fully engaged Wyndham Connect users, and Choice’s CHARLIE, EasyBid and RAISE tools are designed explicitly to reduce labour intensity for limited-service operators. The Wakefield-conducted Wyndham Owner Trends Report (December 2025) is a useful temperature check: 46 per cent cite data privacy as the top barrier, and 57 per cent want human oversight on AI operating decisions.

Hotel GMs and operations leaders. The most direct day-to-day impact is in front-desk automation (Marriott’s ACU, Wyndham’s voice agents, Hyatt’s agentic platforms), housekeeping orchestration (H World’s automated room assignment) and energy management (Hilton LightStay). The cultural challenge raised by Accor’s Khang Nguyen Trieu – that data quality and a single source of truth must come first – is the practical constraint most GMs will recognise.

Commercial leadership. Revenue managers and commercial directors should plan for a two-track 2026. Track one is the in-platform AI search rollout: Marriott, Hilton, IHG and Hyatt are all deploying conversational search on owned channels in 2025–2026. Track two is the agentic distribution layer, where Wyndham (ChatGPT, Claude, Gemini), Accor (ALL Accor in ChatGPT) and Hyatt (ChatGPT) have built native integrations into third-party LLMs. The unanswered question, raised by Skift’s December 2025 analysis, is whether agentic AI will succeed where social media largely failed at converting attention into bookings.

Financial leadership. CFOs and controllers face two distinct AI conversations. The first is corporate productivity: Hyatt explicitly tied a roughly 5 per cent corporate workforce reduction to AI-enabled efficiency. The second is capex governance: Marriott’s US$1 billion-plus annual technology spend, Wyndham’s US$450 million-plus cumulative spend since 2018 and IHG’s renewed AI investment under a new SVP of AI and architecture all imply multi-year amortisation. Marriott’s Capuano was explicit on the Q4 2025 earnings call that customer acquisition costs are the metric to watch – if AI distribution materially reduces OTA dependence, the operating leverage will be substantial.

Guests and bookers. The clearest guest-side wins are still in friction reduction: digital key adoption (around 80 per cent guest preference per Hilton reporting), faster booking conversion (Hyatt’s natural language search), better matching for non-standard requests (Marriott Homes & Villas saw a doubling of property-save rates with AI search) and automated upgrades. The risk is also real: the View from the Wing and LoyaltyLobby coverage of Marriott’s ACU rollout in 2025 captured an early elite-member backlash about transparency, suggesting that loyalty politics may be the underestimated AI risk for premium brands.

Three honest gaps deserve naming. First, almost no chain has disclosed direct, audited RevPAR uplift attributable to generative AI. The numbers cited above are real, but most relate to cost savings, handling times, conversion rates or engagement, not topline RevPAR growth. Owners should ask brands for specific RevPAR figures and treat softer “engagement” metrics with caution.

Second, the loudest AI announcements – ChatGPT apps, AI Mode integrations, AI Planners – are about distribution and discovery rather than operations. As Skift’s December 2025 piece on social media’s failure to crack travel booking pointed out, conversational AI may end up as a facilitator rather than a primary booking channel. Hyatt’s Mark Hoplamazian was careful to caveat on his Q4 2025 call that “everybody in the world is at the table with Google and everything else.” Owners would be wise not to over-index on any one LLM partnership.

Third, AI talent and governance are running ahead of AI deployment at most chains. IHG only named a senior vice president of AI and architecture in January 2026. Hilton confirmed a search for a new CTO in March 2026. Marriott is still expanding its AI governance framework, which CIO Naveen Manga has said will “continuously iterate”. The hard work of integrating these tools across thousands of franchised properties – with appropriate data, brand standard and privacy controls – is just beginning.

Looking ahead to 2026 and 2027, four near-term deployments will define the competitive landscape. Marriott will deploy natural language search across Marriott.com and the Bonvoy app in H1 2026, supported by US$1 billion-plus in capex. IHG will deploy an AI-enhanced CRM and continue restructuring its hotel content into machine-readable formats for AI agents. Hyatt has signalled further agentic platform rollouts and expects credit card-driven EBITDA to more than double by 2027. Wyndham, Accor and Hyatt will be the most aggressive in third-party LLM integration. Hilton will scale its AI Planner from beta to general availability.

For owners, asset managers and commercial leaders making capital allocation decisions over the next 24 months, the practical conclusion is simple. Pick brand partners less on the loudness of their AI announcements and more on the underlying architecture, the verifiable cost savings, the maturity of their data foundation and the clarity with which they have tied AI deployment to franchisee economics. The chains that win will not be those with the flashiest demos; they will be those that turn AI into a measurable line on the P&L within the franchise agreement period.

Sources:

Sources drawn from Marriott, Hilton, IHG, Wyndham, Hyatt, Accor, Choice Hotels and H World investor releases, SEC and HKEX filings, earnings call transcripts (via Investing.com, Yahoo Finance, Constellation Research), Skift, PhocusWire, Hotel Dive, Hotel Technology News, Hospitality Net, Hospitality Technology, Hotel Management International, Smart Energy Decisions, edie, BSR, ei3, PwC, Publicis Sapient, Fortune, CIO Dive, CIO.inc, ChinaTravelNews, TTG Asia, PYMNTS and PR Newswire / Business Wire press distributions, covering material through Q4 2025 and early 2026 announcements.