Hong Kong (China)

By the close of this century, Hong Kong tourism will be almost unrecognisable to anyone who knew the harbour in its early-2000s heyday. The city will be hotter, wetter, more storm-battered, demographically halved, deeply enmeshed in the Greater Bay Area, and pervasively mediated by artificial intelligence. The arc is set less by any single shock than by the steady compounding of climate, demography, regional integration, and technology — four slow tides that, taken together, are already lifting the foundations of every coastal attraction from the Tsim Sha Tsui Promenade to the stilt houses of Tai O.

The climate baseline shift is the most legible of these forces. Under the central trajectory used by the Hong Kong Observatory, annual mean temperatures are projected to rise about two degrees Celsius from the 1995–2014 baseline by the end of the century, and the city is already running ahead of that envelope. "Very hot days" of 33 °C or above, which averaged fourteen a year between 1995 and 2014, have lately settled into a band of roughly fifty a year — and statistically downscaled projections suggest the late-century average could approach ninety. Annual rainfall is creeping up, but the more consequential change is the rise in extreme daily rainfall, projected at around 16 percent above a 203 mm baseline. The September 2023 rainstorm that broke three Observatory records and flooded sections of the MTR was not an outlier so much as a glimpse of routine.

Sea level is the second slow knife. The Observatory's application of IPCC ranges puts the likely mean rise at between 0.37 and 0.82 metres by 2081–2100 under the intermediate scenario, with higher-emission and ice-sheet-instability ranges that climb considerably further. Today's once-in-fifty-year extreme sea level becomes nearly an annual event over much of the harbourfront. The Star Ferry piers at Central and Tsim Sha Tsui sit barely above current high tide. The Kai Tak Cruise Terminal occupies a former runway perched on the very water it serves. Hong Kong Disneyland's low-lying Penny's Bay site already shutters during the worst typhoons. By the 2070s, an elevated harbourfront promenade linking Tsim Sha Tsui through Hung Hom, Wan Chai, Central, and the Western District is not a flourish but a structural necessity, alongside seawall raising at Hung Hom, Quarry Bay, and Aberdeen, and flood-gate retrofits across the underground rail network.

Typhoons have been intensifying in ways that matter for tourism well before they make landfall. Research from regional climate centres projects that, under high-warming scenarios, late-century landfalling storms in coastal Asia will arrive faster, last longer, push further inland, and roughly double their destructive power. The string of Hato, Mangkhut, Saola, and the September 2025 Ragasa — which closed Hong Kong Disneyland for two days — is a preview rather than a peak. The most acute cultural-tourism casualty of all this is Tai O. The stilt-house fishing village, repeatedly inundated by recent storm surges and already covered by an emergency flood-response plan, has been described by one Hong Kong atmospheric scientist as a place that could become uninhabitable by century's end without major adaptation. Whatever it becomes, it is unlikely to remain the village that draws today's visitors.

Inland, the country parks that cover roughly 40 percent of Hong Kong's land area face a quieter transformation. Hiking on the MacLehose Trail in mid-summer will be a hazard rather than a pastime; shoulder-season and night-time use will dominate. Hill-fire risk shifts with the drier winters, and roughly a quarter of assessed local species are at risk of local extinction. The Mai Po Ramsar wetlands, a staging ground on the East Asian–Australasian Flyway for tens of thousands of migratory birds, sit at the sharp edge of sea-level rise: under a 1.5-metre rise paired with reduced Pearl River sediment supply, mangroves, tidal flats, and high-tide roosting habitats are largely lost. Preserving inland fishponds as a managed habitat-migration corridor is the only credible lever, and the decision is essentially political.

The demographic story is, in its way, just as drastic. United Nations medium-scenario projections show Hong Kong's population falling from roughly 7.5 million today to about 2.1 million by 2100, driven by a total fertility rate that is now the world's lowest. Government projections show the share of residents aged 65 and over rising sharply within the next two decades, and by 2100 something close to half of all Hong Kongers may be over that threshold. For tourism, this means a shrinking and ageing host population, a hospitality workforce structurally dependent on Mainland and South and Southeast Asian guest workers, and a cultural-linguistic centre of gravity that drifts away from the Cantonese vernacular which has long given Hong Kong its tourism brand its distinctive flavour.

Within the same horizon, Hong Kong's identity as a discrete destination dissolves into the Greater Bay Area. With the West Kowloon high-speed rail and the Hong Kong–Zhuhai–Macao Bridge already operating, the eventual "one-hour living circle" of a roughly 70-million-person megaregion implies a tourism geography in which visitors arrive at whichever Bay Area airport is most convenient and circulate across multiple cities in compressed trips. Hong Kong's role becomes the international wing of a larger cluster — closer in feel to what Macau is to Zhuhai today than to a standalone city-state. Mainland visitors, already over 70 percent of arrivals, will deepen their share, and same-day visitation from Shenzhen and Guangzhou will likely become the single largest tourism segment.

Artificial intelligence cuts both ways across this terrain. On the constructive side, real-time crowd management can disperse visitors away from Mong Kok and Causeway Bay choke points; AI-enhanced nowcasting can shave casualty risk during typhoons in Tai O and the outlying islands; smart-building HVAC optimisation is essential to meeting the city's commitments to carbon neutrality before 2050. Heritage digital twins of walled villages, temples, and the stilt houses themselves are quietly being built as climate insurance. On the other side, data centres in a hot, humid, water-scarce subtropical city are an awkward fit; Hong Kong already imports the majority of its fresh water, and the cooling load of regional AI growth will press on that dependency. Mainland-style facial recognition, payment-by-face, and seamless biometric entry are likely to become standard, with consequences for visitors from jurisdictions with stricter data-protection regimes. And the same AI itineraries that promise to disperse visitors tend, left to themselves, to cluster them on the same handful of sites, paradoxically deepening over-tourism unless explicitly counter-engineered.

The pollution picture, in contrast, is the rare element trending in the right direction. Local emissions controls have cut ambient PM10, PM2.5, NO₂, and SO₂ by between 45 and 88 percent over two decades. Hours of reduced visibility are down by more than 80 percent. The remaining binding constraints are regional: cross-boundary ozone from the Pearl River Delta, marine pollution and microplastics flowing south, and the slow recovery of the Chinese white dolphin, whose numbers collapsed by more than 80 percent in the 2000s and 2010s and have only marginally rebounded. By mid-century, if Mainland decarbonisation holds to schedule, views from the Peak may be clearer than at any point since the 1970s. The exception is light pollution, where Hong Kong remains the world's brightest sky — Tsim Sha Tsui has been measured at roughly 1,200 times the international dark-sky standard — and where voluntary measures have made almost no dent. The Symphony of Lights and the dark-sky Astropark at Sai Kung sit on opposite sides of that unresolved trade-off.

Underneath all of this lies a governance question that may matter more than any single climate parameter. Hong Kong's drainage and coastal infrastructure is being designed to a sea-level-rise allowance well below what the upper end of credible scenarios demands. Closing that gap, mandating binding light-pollution rules, designating a dolphin conservation area in west Lantau waters, and climate-proofing Tai O with raised walkways and partial relocation are decisions the next decade can still make. So is the choice of how privacy and identity tracking are governed in the smart-tourism systems now being designed. None of these are foregone. By 2100, the Hong Kong that visitors recognise will be the one whose authorities chose, in the 2020s and 2030s, to take the upper end of the climate envelope seriously, to keep some space between the city's tourism brand and the surveillance defaults of the surrounding megaregion, and to spend on the unglamorous infrastructure — seawalls, flood gates, cooling shelters, habitat corridors — that allows a coastal subtropical city to keep being itself.