QTS Hong Kong and the Scarcity Price Behind a Gateway Data-Centre City

The Procurement Meeting Starts With Three Cities, Not One Rack

The first question in the procurement room is not whether QTS Hong Kong has a good data-centre sales deck. It is whether a financial platform should keep a material piece of its Asia execution stack in Hong Kong at all. The platform clears payments, runs market-data feeds, serves broker APIs and keeps a growing crypto and tokenised-asset product line close to banks, custodians and exchange partners. Its risk team has three columns on the screen: Hong Kong, Singapore and Tokyo. Each column has a different answer to the same problem.

Singapore has policy discipline, a deep regional headquarters base and a curated new-capacity regime after years of restriction. Its Green Data Centre Roadmap says the city-state aims to provide at least 300 MW of additional data-centre capacity, with more possible where operators bring green energy solutions, at https://www.imda.gov.sg/how-we-can-help/green-dc-roadmap. That makes Singapore credible, but not loose. A buyer who wants Singapore capacity is still competing inside a policy-filtered market that values energy efficiency, national planning and water discipline. Tokyo has scale, a vast domestic enterprise market and Japan's legal and engineering credibility. JLL's Japan note at https://www.jll.com/en-jp/insights/japan-data-centre-market-opportunities says rising generative AI use is raising power-shortage concerns, while data centres and semiconductor factories could drive power demand to 7.15 million kW by 2034. Tokyo can be large, but it is not magically unconstrained.

Hong Kong is the uncomfortable middle case. It is smaller than Tokyo, more politically exposed than Singapore, and still hard to replace for institutions that need proximity to mainland China, offshore capital, securities markets, cross-border treasury, regional staff and low-latency connectivity into a dense financial city. The Digital Policy Office's July 2026 fact sheet at https://www.datacentre.gov.hk/en/downloads/HK_as_DC_prime_location.pdf describes Hong Kong as an international financial, trading and logistics hub where high-performance and secure data-centre services support global corporations. It also says Hong Kong has a fully liberalised telecommunications regime, high-speed fibre availability throughout the territory, 3,651 broadband-service licensees as of May 2026, household broadband penetration of 100.3% and mobile subscriber penetration of 442.6% as of February 2026. Those are not decorative statistics. They explain why a trading or payments platform does not treat Hong Kong as merely another Asia-Pacific pin on a map.

The procurement team therefore does not ask whether Hong Kong is cheap. It asks what scarcity buys. Hong Kong capacity is valuable when it reduces latency to counterparties, keeps operational teams near financial customers, supports direct cloud access, provides a credible regional resilience point, and gives China-adjacent customers a jurisdiction they can explain to both international and mainland stakeholders. It becomes less valuable when the facility is old, power-dense expansion is unavailable, cross-connects are weak, customer concentration turns one tenant into the economic landlord, or political and compliance uncertainty makes the buyer overpay for optionality that cannot be used.

That is the right frame for QTS Hong Kong. The public record for the Hong Kong-specific entry is thin, but the owner context is unusually large. PeeringDB lists QTS Hong Kong at https://www.peeringdb.com/net/14383 under QTS Realty Trust, Inc., with ASN 58797, open general peering policy, no disclosed traffic level, zero IPv4 prefixes, zero IPv6 prefixes, and no listed public exchange points or interconnection facilities. Data Center Catalog lists a QTS Hong Kong Data Center in its Hong Kong directory at https://datacentercatalog.com/hong-kong-gridview, but without detailed features. QTS itself now presents a global platform backed by Blackstone, and its leadership page at https://q.com/our-leadership/ says QTS has 75 data centers in operation or under development across the United States and Europe, more than $3 billion in annual leasing revenue and total contracted power capacity eclipsing 3 GW.

The procurement decision, then, is not a simple "does QTS have Hong Kong?" box. It is an underwriting exercise around a public trace, a global capital sponsor and a constrained city. The valuable answer would be a Hong Kong facility with bank-grade resilience, power expansion rights, a clear customer book, serious cloud on-ramp reach and cross-border demand that is sticky rather than episodic. The dangerous answer would be a legacy record, a thin network footprint and a facility claim that cannot be tied to power, interconnection, contractual customer demand or current operating control. In Hong Kong, the spread between those two answers is the scarcity price.

Hong Kong Sells Milliseconds and Legal Comfort, Then Bills for Scarcity

Hong Kong's data-centre appeal begins with an unusual bundle: finance, telecom density, China proximity, information flow and operational habit. The government's "Why Hong Kong" data-centre page at https://www.datacentre.gov.hk/en/accommodating_data_centres/why_hk.html says financial services and insurance plus trading and logistics accounted for about half of total data-centre space. It links demand to high-frequency stock trading, e-commerce and cloud computing. That matters because data-centre demand is not just compute demand. It is a geographic reflection of the industries that need to settle, reconcile, route, authenticate and recover in the same city where customers, regulators and banks already operate.

The physical bill arrives quickly. Hong Kong is dense, land is expensive, and data centres do not fit neatly into a city that also needs housing, logistics, offices, ports, community uses and transport corridors. The "Find a Site" page at https://www.datacentre.gov.hk/en/accommodating_data_centres/find_a_site.html shows how much of Hong Kong's data-centre supply logic depends on zoning, industrial buildings, "Other Specified Uses" business areas, commercial buildings and open-market premises. That is not how a hyperscale developer thinks about a greenfield campus in a large land market. Hong Kong's starting point is conversion, redevelopment and site-by-site permission. The result is a market where the building shell can be as decisive as the server spec.

The 2024 Legislative Council reply at https://www.info.gov.hk/gia/general/202406/05/P2024060500513.htm gives the numbers behind that friction. It says Hong Kong had about 970,000 square metres of data-centre floor area at the time, expected total floor area could reach 1.5 million square metres by 2026, and operators surveyed in 2023 indicated short-to-medium-term demand for about 300,000 square metres. It also says 44 waiver applications for converting industrial-building floor space had been approved by mid-May 2024, around 70% of those applications involved high-tier data centres, and waiver plus lease-modification approvals were estimated to provide about 300,000 square metres of total floor area, including more than 240,000 square metres for high-tier use.

Those figures cut both ways. They show the government is not ignoring supply. They also show why new Hong Kong capacity is a real-estate and power-approval exercise before it is a sales exercise. A buyer deciding between Hong Kong, Singapore and Tokyo sees not only capacity today but the probability of expansion tomorrow. Singapore's roadmap turns expansion into a national allocation question. Tokyo's expansion story depends on Japanese power and regional-diversification policy. Hong Kong's expansion story runs through industrial conversion, established clusters, future Northern Metropolis supply and the ability to turn older stock into modern, power-dense rooms without destroying the economics.

CBRE's 2026 Hong Kong release at https://www.cbre.com.hk/press-releases/apac-data-centre-report-2026-hong-kong-remains-resilient sharpens the point. It says Hong Kong leasing demand continues to strengthen, supported by hyperscale cloud service providers, mainland Chinese technology and e-commerce companies, financial institutions, multinational banks and international financial institutions. It also says demand is polarised: occupiers either become early tenants in new high-specification facilities or secure short-term capacity in older assets as an interim solution. Vacancy is expected to decline over 12 to 18 months as the supply cycle matures, and longer-term supply is anchored by the Sandy Ridge Data Facility Cluster, expected to deliver about 250,000 square metres of gross floor area from 2029.

That is the scarcity essay in one market paragraph. Hong Kong is not simply short of buildings. It is short of the right kind of powered, resilient, connected, bank-comfortable buildings at the right moment. QTS Hong Kong's value, if the local operating reality is strong, would come from sitting on the right side of that divide: not merely another data-centre name, but scarce high-specification capacity in a gateway city where customers may use older assets temporarily while waiting for better ones.

QTS Hong Kong Is a Thin Public Record Inside a Much Larger Capital Story

The evidence for QTS Hong Kong must be read in two layers. The first layer is the Hong Kong-specific public trace. PeeringDB's record at https://www.peeringdb.com/net/14383 identifies "QTS Hong Kong" and QTS Realty Trust, Inc., but it does not show visible public peering, listed interconnection facilities, disclosed traffic levels or originated IPv4 and IPv6 prefix counts. It records an open general peering policy and an RIR status marked okay, with last update in 2022 and RIR status update in 2024. That is evidence of a registered network presence or historical interconnection intent, not proof of a live, revenue-generating Hong Kong facility with active customer traffic.

The third-party facility layer is also incomplete. Data Center Catalog's Hong Kong grid at https://datacentercatalog.com/hong-kong-gridview lists "QTS Hong Kong Data Center" among a crowded set of Hong Kong facilities, including PCCW, Equinix, Telehouse, MEGA-i, MEGA Plus, Global Switch and others. The QTS entry has no detailed features in the listing. That is useful as a market signal because it suggests the QTS Hong Kong name has circulated in facility directories. It is not sufficient to underwrite power capacity, operational control, live tenancy, tier standard, cross-connect density, leasehold position or expansion rights.

The second layer is the global QTS and Blackstone story. QTS is no longer a public REIT reporting like it did before Blackstone took it private. Blackstone's 2021 transaction announcement at https://www.blackstone.com/news/press/qts-realty-trust-to-be-acquired-by-blackstone-funds-in-10-billion-transaction/ valued QTS at approximately $10 billion including assumed debt. QTS's own announcement at https://q.com/news/qts-realty-trust-to-be-acquired-by-blackstone-funds-in-10-billion-transaction/ framed Blackstone's capital access as support for expanding data-centre solutions. Blackstone's AirTrunk discussion at https://www.blackstone.com/insights/article/behind-the-deal-airtrunk/ says QTS grew by more than 900% in size in the first three and a half years under Blackstone ownership and that market rents rose 84% against tighter supply constraints. The same piece says AirTrunk has assets in Australia, Japan, Singapore, Malaysia and Hong Kong.

This matters because a Hong Kong data-centre buyer is not only buying local walls. It is buying sponsor credibility. A buyer choosing a small independent Hong Kong operator has to ask whether the company can fund power upgrades, absorb delay, manage large customers, procure equipment and survive a bad utilisation period. A buyer looking at a Blackstone-backed QTS context asks a different question: whether the sponsor's capital and global customer relationships can be translated into a specific Hong Kong asset, or whether the local record is too thin to benefit from the platform story.

The difference is important. Capital cannot conjure a substation where one is not available. A global lease book cannot make a building suitable for high-density loads if floor loading, risers, chillers, fire systems or utility feeds are wrong. A strong sponsor can, however, change the probability of execution when the site is otherwise viable. It can sign longer utility commitments, accept longer payback, pre-develop land, bring in anchor tenants and negotiate with customers who want capacity across multiple regions. QTS's "Access to Infrastructure" page at https://q.com/why-qts/access-to-infrastructure/ explicitly presents land, capital, supply chain, utilities and labour as core resources. That is exactly the list a Hong Kong facility would need.

The QTS Hong Kong thesis is therefore conditional. If QTS has direct operating control, expansion-capable power, modern facility standards, strong cross-connect reach and a bankable customer book, Hong Kong scarcity could make the asset more valuable than its megawatt count suggests. If the public record is mostly a legacy network entry and a sparse directory listing, the QTS brand and Blackstone ownership are not enough. The market value turns on documents not visible in the public record: site rights, power commitments, customer contracts, interconnection maps, occupancy, renewal terms and who actually controls the facility.

Power Is Reliable; Expansion Is the Harder Variable

Hong Kong's official power story is strong. The Digital Policy Office fact sheet at https://www.datacentre.gov.hk/en/downloads/HK_as_DC_prime_location.pdf says power supply reliability exceeds 99.999%, the two power companies have extra capacity and emergency support arrangements, and Hong Kong enjoys comparatively low electricity tariffs in the Asia-Pacific region. The data-centre FAQ at https://www.datacentre.gov.hk/en/useful_info/faq.html says dual utility feeds are theoretically possible but practically not required because CLP and HK Electric have robust generation, transmission and distribution networks, and because high-tier data centres can use two feeds from different substations of the same power company.

Reliability, however, is not the same as expandable power at a specific building. A financial platform does not ask only whether Hong Kong's grid is reliable. It asks whether the facility can deliver another 1 MW, 5 MW or 10 MW when workloads move from ordinary colocation to dense compute, database replication, analytics, artificial intelligence inference and security telemetry. It asks whether committed power is backed by utility letters, whether cooling can handle higher rack densities, whether diesel storage and generator runtime match the service commitment, and whether power costs are passed through in a way finance can forecast.

The arithmetic is blunt. CLP Power's November 2025 tariff announcement at https://www.clp.com.hk/content/dam/clp-group/channels/media/document/2025/20251118_en.pdf says the average net tariff for 2026 is HK$140.6 cents per unit of electricity, effective 1 January 2026. One megawatt running flat for a 30-day month consumes 720,000 kWh before cooling overhead and electrical losses. At HK$1.406 per kWh, the raw energy input is about HK$1.01 million a month, or roughly US$130,000 at a 7.8 exchange rate, before demand charges, cooling, UPS losses, generators, maintenance, security, rent, financing, remote hands and margin. Ten megawatts turns that into roughly HK$10.1 million a month of raw electricity input. The calculation is simple enough for the procurement team to understand and large enough to change a lease negotiation.

This is where QTS's global operating model becomes relevant. QTS's sustainability page at https://q.com/2024-sustainability-report/ says its owned operational portfolio uses closed-loop cooling that does not consume water for cooling once operational and saved nearly 1.5 billion gallons of water in 2024. That does not prove the same design applies to any Hong Kong site. It does show the global platform's preferred answer to a world where water, power and community scrutiny have become part of data-centre economics. A Hong Kong buyer would want to know whether the local facility follows equivalent cooling and energy discipline or relies on older building systems that may not support dense deployments economically.

CBRE's 2026 Hong Kong release makes power the determinant rather than the footnote. Samuel Lai, CBRE Hong Kong's head of industrial and logistics, is quoted at https://www.cbre.com.hk/press-releases/apac-data-centre-report-2026-hong-kong-remains-resilient saying power availability is now a key determinant of where data-centre capacity can be deployed across Asia Pacific. For Hong Kong, future growth depends on the city's ability to support higher-density facilities and unlock new development opportunities, including in the Northern Metropolis. Cushman & Wakefield's global 2026 comparison at https://www.cushmanwakefield.com/en/insights/global-data-center-market-comparison makes the same global point: power availability and delivery timelines are fundamentally reshaping site selection.

That is why a QTS Hong Kong underwriting file cannot stop at "Hong Kong has reliable power." Reliability supports the uptime argument. Expansion power supports the valuation argument. If a QTS Hong Kong facility has secured megawatts in an established cluster, the scarcity value is real. If it has only nominal power, old cooling and no utility path for density, it becomes interim capacity in CBRE's polarised market, useful for short-term demand but not for the next generation of financial and cloud workloads.

Land Scarcity Turns Facility Evidence Into the Main Diligence Question

Land scarcity in Hong Kong is not an abstract macro condition. It decides what kind of facility can exist, how fast it can expand, and whether a buyer should treat an operator as a long-term platform or a tactical stopgap. The government's 2020 concessionary-measures review at https://www.datacentre.gov.hk/en/downloads/Data_Centre_Study_Report_2020_Eng.pdf says data-centre installations were clustered mainly in Tseung Kwan O, Kwai Chung and Tsuen Wan, and estimated about 743,000 square metres of data-centre floor space in Hong Kong as of the end of 2019. It also says 12 data centres in Tseung Kwan O and Tai Po industrial estates occupied about 356,000 square metres, with Tseung Kwan O forming the largest cluster.

The same review explains the conversion logic. Industrial buildings are a flexible supply source, but old industrial shells can be hard to adapt because high-tier data centres have stringent power and building-structure requirements. The review says executed waiver applications contributed about 141,000 square metres of internal floor area by June 2020, while dedicated sites and lease-modification routes added further supply. It projected floor-space supply rising from about 743,000 square metres at the end of 2019 to 1,169,000 square metres by 2025, with more long-run supply potentially coming from Hung Shui Kiu and strategic cavern areas.

The 2024 Legislative Council reply updates that progression at https://www.info.gov.hk/gia/general/202406/05/P2024060500513.htm. The public-sector view is that total data-centre floor area could reach 1.5 million square metres by 2026 and meet short-to-medium-term demand. A buyer should not read that as "supply is easy." The same reply exists because legislators were asking whether existing industrial buildings are unfit for higher-tier facilities and whether building laws, fire safety, security, sewage, parking, lifts and toilets need review for data-centre and supercomputing use. In other words, supply exists on paper, but conversion quality is the market's fault line.

QTS Hong Kong's public facility evidence has to be tested against that fault line. A facility listing with no features is much less valuable in Hong Kong than in a land-abundant market, because the hard questions are building-specific. What is the floor loading? What is the ceiling height? How many independent risers exist? How close is the substation? Is the building in Kwai Chung, Tseung Kwan O, Tsuen Wan, Fo Tan, Fanling, Cyberport or another cluster? Does the landlord allow heavy mechanical upgrades? Can rooftop or plant-room space support cooling? Are there rights to install additional generators, fuel tanks, chillers or switchgear? Is there room for new meet-me infrastructure? Does a lease allow enough term to amortize fit-out?

The scarcity value is highest when the answers are already secured. A shell that can be converted is not the same as a powered data hall. A data hall with low-density racks is not the same as a high-density cloud or financial-services facility. A cluster address is not the same as a cross-connect-rich building. This is why Equinix's Hong Kong page at https://www.equinix.com/data-centers/asia-pacific-colocation/china-colocation/hong-kong-data-centers sells not only uptime but ecosystem density: 4,900-plus enterprises, 2,000-plus network service providers, direct links to mainland China and Asia-Pacific corridors, and AI-ready data centres. It is trying to show that its buildings are not generic real estate.

For QTS Hong Kong, the absence of comparable public local detail is the central weakness. It does not mean the asset is weak. It means the public evidence cannot yet tell whether QTS Hong Kong sits in the high-specification capacity bucket or the interim older-asset bucket. In a normal market, that might be a small disclosure gap. In Hong Kong, it is the underwriting question.

Interconnection Is the Rent, Not the Garnish

Data-centre scarcity in Hong Kong is not only about megawatts and square metres. It is also about who can be reached without leaving the building, the campus or the metro. For a financial platform, the difference between a rack and a useful rack is the reach to exchanges, banks, brokers, cloud regions, payment networks, security providers, content platforms and counterparties. Cross-connect density can turn the same power bill into either commodity hosting or a strategic operating surface.

Hong Kong has a deep interconnection base. HKIX says at https://www.hkix.net/hkix/whatishkix.htm that, owing to its large scale and traffic volume, it is widely considered one of the largest internet exchange points in Asia Pacific. Internet Society Pulse records at https://pulse.internetsociety.org/en/ixp-tracker/country/HK/ that Hong Kong had 16 active internet exchange points and 603 combined members as of July 2026, with 79% of active networks either IXP members themselves or customers of IXP members. Those numbers do not tell a buyer whether one QTS Hong Kong hall is well connected. They tell the buyer why a well-connected Hong Kong hall can carry a premium.

Cloud reach adds another layer. AWS lists Equinix HK1 in Tsuen Wan as an AWS Direct Connect location for the Asia Pacific Hong Kong region at https://aws.amazon.com/directconnect/locations/, available in 1G, 10G and 100G modes. AWS's earlier Hong Kong region post at https://aws.amazon.com/blogs/aws/now-open-aws-asia-pacific-hong-kong-region/ noted Direct Connect locations at iAdvantage Mega-i and Equinix HK1. Equinix explains cloud on-ramps at https://blog.equinix.com/blog/2024/01/30/what-is-a-cloud-on-ramp/ as direct private connections offered by cloud providers inside vendor-neutral colocation data centres. For a platform that runs both local infrastructure and hyperscale services, the practical question is whether its Hong Kong colocation point can reach cloud providers privately, predictably and without adding operational complexity.

This is where the PeeringDB QTS Hong Kong record is both useful and sobering. The record at https://www.peeringdb.com/net/14383 does not list public exchange points or interconnection facilities. It shows open general peering policy but no public peering rows. A buyer should not overinterpret that; PeeringDB records can be incomplete and private interconnection does not always appear as public peering. But the absence of visible local exchange and facility rows is a real public-data signal. It means a QTS Hong Kong sales claim would need to be supported by private cross-connect maps, carrier lists, meet-me-room diagrams, cloud partner availability, latency data and proof of live routes.

The Hong Kong data-centre market already has strong interconnection incumbents. Data Center Catalog's Hong Kong list at https://datacentercatalog.com/hong-kong-gridview shows major assets such as MEGA-i, MEGA Plus, MEGA Two, PCCW facilities, Telehouse, Equinix and Global Switch. The procurement team will know that moving into a less connected facility can be expensive even if the rack price is better. Every missing carrier becomes a metro circuit. Every absent cloud on-ramp becomes another network design. Every extra hop becomes latency, fault domain and operational burden.

That does not mean QTS Hong Kong cannot compete. A capital-backed operator can lease or build transport, bring customers to a new site, and use anchor demand to create an ecosystem. But the economics are path dependent. Interconnection becomes easier after the first major customers and carriers arrive. It is harder before that point because each buyer asks why it should be the one to create the network gravity. In Hong Kong's scarce market, a facility with power but weak interconnection may still lease to tenants that need space urgently. A facility with power and dense interconnection can charge for a different product: the ability to keep infrastructure close to the institutions that make Hong Kong valuable.

The China Gateway Is Valuable Because It Is Not Fully Solved

Hong Kong's China gateway role is often described too cleanly. The actual value comes from a tension: companies want proximity to mainland China, but they also want international legal comfort, financial-market access, global cloud relationships, and operational separation from mainland domestic infrastructure where appropriate. Hong Kong data-centre capacity is valuable because that tension has not disappeared. It is not valuable because it solves every China problem.

The Digital Policy Office fact sheet at https://www.datacentre.gov.hk/en/downloads/HK_as_DC_prime_location.pdf names proximity to the Chinese Mainland market as one of Hong Kong's advantages. The "Why Hong Kong" page at https://www.datacentre.gov.hk/en/accommodating_data_centres/why_hk.html says the Mainland is a significant market and provider of ICT products and services, with policies driving cloud computing, e-business, Internet of Things and logistics. It says these create substantial demand for high-security and high-serviceability data-centre services. This is the official version of the demand story. The commercial version is sharper: Hong Kong lets a platform serve China-adjacent customers while keeping finance, contracts, support staff and international connectivity in a city global institutions already use.

The risk is that "gateway" becomes a vague premium. A buyer should separate four things: latency to Hong Kong institutions, connectivity to mainland networks, legal and compliance comfort, and the ability to expand capacity. A facility can be strong on one and weak on another. It may have a great Hong Kong location but poor cloud on-ramps. It may have strong telecom access but limited power density. It may have enough racks for a current platform but no expansion right for the next product line. It may be legally and operationally comfortable but too expensive for workloads that could sit in Singapore, Tokyo, Osaka, Seoul, Taipei or Johor.

This is where QTS's Blackstone context matters again. Blackstone's AirTrunk acquisition discussion at https://www.blackstone.com/insights/article/behind-the-deal-airtrunk/ says AirTrunk has assets in Hong Kong along with Australia, Japan, Singapore and Malaysia, and positions Blackstone as the largest data-centre provider globally after the deal. A sponsor with QTS in North America and Europe plus AirTrunk in Asia-Pacific can think about customer demand across multiple markets. A financial-platform customer might not want only Hong Kong. It may want Hong Kong plus Tokyo, Singapore and Sydney resilience under a sponsor that understands hyperscale procurement and enterprise risk.

But QTS Hong Kong itself still needs local proof. Blackstone owning AirTrunk's Hong Kong exposure does not automatically make the QTS Hong Kong record operationally rich. It does, however, change how the market should read strategic optionality. If QTS Hong Kong is a live asset or can be tied to a broader Blackstone data-centre customer strategy, it may benefit from portfolio demand. If it is a legacy or thin record, the practical Hong Kong exposure may sit elsewhere in the sponsor's Asia-Pacific holdings. That distinction would change valuation, sales strategy and customer confidence.

The China gateway value is therefore best treated as a call option with carrying costs. It pays off when customers need Hong Kong specifically and cannot get enough high-quality capacity elsewhere. It decays when workloads can move to Singapore, Tokyo, Osaka, Johor or other regional markets without hurting product performance. QTS Hong Kong's job is to make the option exercisable: powered, connected, compliant, supported and expandable.

Blackstone Changes the Balance-Sheet Question

Before Blackstone, QTS could be read through public REIT filings. Its 2020 annual report on the SEC site at https://www.sec.gov/Archives/edgar/data/1577368/000162828021003456/qts-20201231.htm said the company had more than 1,200 customers, with the largest accounting for about 13.1% of monthly recurring revenue and no other customer above 5.4%. It reported 2020 revenue of $539.4 million and adjusted EBITDA of $299.3 million in the corresponding results release at https://www.prnewswire.com/news-releases/qts-reports-fourth-quarter-and-full-year-2020-operating-results-301229310.html. Those numbers are dated because QTS is now private and much larger, but they remain useful for understanding the old platform: a data-centre operator with meaningful scale, real customer diversity and a history of hyperscale plus hybrid colocation demand.

Blackstone changed the capital base. The 2021 transaction at https://www.blackstone.com/news/press/qts-realty-trust-to-be-acquired-by-blackstone-funds-in-10-billion-transaction/ put QTS inside long-duration infrastructure and real-estate capital rather than public-market quarterly reporting. Blackstone later described QTS as the fastest-growing data-centre company in the world and said its size grew by more than 900% under Blackstone ownership in the first three and a half years, at https://www.blackstone.com/insights/article/behind-the-deal-airtrunk/. QTS's current leadership page at https://q.com/our-leadership/ now states more than $3 billion in annual leasing revenue and contracted power capacity above 3 GW. That is an entirely different balance-sheet context from a small regional Hong Kong data-centre operator.

The practical effect is not that QTS can ignore unit economics. It is that QTS can underwrite a longer, larger and more complex path to capacity than many private operators. Data-centre development is now a competition for land, utility access, equipment, construction labour, customer pre-commitments and permission. QTS's own positioning at https://q.com/why-qts/access-to-infrastructure/ names land, capital, supply chain, utilities and labour as the resources that create customer solutions. Blackstone's Pennsylvania announcement at https://www.blackstone.com/news/press/blackstone-to-invest-more-than-25-billion-in-pennsylvanias-digital-and-energy-infrastructure-plus-catalyze-an-additional-60-billion-investment/ shows the same logic at large scale: QTS development tied to energy infrastructure, utility partnership and new generation.

Hong Kong is not Pennsylvania. It does not offer the same land or generation canvas. But the strategic lesson is transferable. In constrained markets, the operator with capital, utility relationships, customer commitments and patience can sometimes turn a difficult site into a valuable one. The operator without those advantages may win an early lease and then run into the expansion wall. QTS Hong Kong should therefore be judged not simply by whether the public record lists a facility, but by whether the local project can draw on the QTS/Blackstone machinery where it matters.

The balance-sheet question also changes customer concentration risk. A small Hong Kong facility can look stable if one large tenant fills it. That same tenant can become the asset's weakness if renewal, power pricing or expansion depends on a single relationship. In the public REIT era, QTS disclosed that its largest customer was 13.1% of MRR. A Hong Kong-specific site could have a much higher concentration because the local market rewards anchor tenants and power pre-commitments. A bank, cloud provider, exchange-adjacent platform or mainland technology company could make a facility viable while also making it fragile. The underwriting question is whether concentration is paid for through longer terms, stronger credit, expansion commitments and power pass-throughs, or whether the operator is simply renting scarcity to one customer at a price that leaves little residual flexibility.

The Customer Book Matters More Than the Name on the Door

For QTS Hong Kong, customer evidence would matter more than almost any public label. A data-centre name can be valuable, but a customer book defines cash flow, renewal risk, operating complexity and strategic leverage. Hong Kong's strongest demand pockets are clear enough from public market commentary: financial institutions, hyperscale cloud providers, mainland Chinese technology and e-commerce firms, multinational banks, trading/logistics companies, and customers using Hong Kong as a regional resilience point. CBRE names many of those demand sources directly at https://www.cbre.com.hk/press-releases/apac-data-centre-report-2026-hong-kong-remains-resilient.

Each customer type has a different margin profile. A hyperscale cloud or large technology tenant may take substantial power and sign longer commitments, but it will negotiate hard on price, expansion rights, service levels and capital contributions. A multinational bank may value resilience and compliance, but it may require audits, strong documentation, access controls, incident procedures and operational transparency. A financial platform may need smaller capacity but pay for latency, cross-connects and support. Mainland e-commerce or technology demand may be valuable but politically and commercially more variable. Colocation resellers and smaller infrastructure customers can fill space, but they may bring churn, support load and network-abuse handling.

Customer concentration becomes the hidden variable. In a scarce market, an operator can be tempted to sign the biggest anchor tenant and call the facility derisked. That may be true for debt service and initial occupancy. It may be less true for long-run pricing power. If one tenant controls most of the capacity, that tenant has renewal leverage. If the tenant uses the facility for a narrow workload, the operator may inherit stranded fit-out when the workload moves. If the tenant is a cloud provider, the operator may be exposed to the provider's regional capacity strategy rather than local end-user demand. If the tenant is a financial institution, the operator may earn trust but face demanding service obligations.

The switching-cost story is more favourable for Hong Kong than for many markets. A financial platform that has cross-connects to local banks, brokers, custodians, exchange services, cloud on-ramps and security providers does not move casually. Migration means retesting routes, latency, failover, data replication, audit evidence, access procedures and disaster-recovery playbooks. If the facility is good, the customer may renew because moving is riskier than paying a premium. If the facility is weak, the same switching cost becomes resentment: the customer stays only until a better high-specification site opens.

This is why facility quality and interconnection quality reinforce each other. A well-powered facility with weak cross-connect options must compete on space, power and service. A connected facility with limited power competes on latency and ecosystem. A facility with both can make itself the default answer for certain workloads. Hong Kong's scarcity premium accrues most strongly to the third category. QTS Hong Kong's public evidence does not yet put it there, but it tells us exactly what to look for.

The customer book would also reveal whether Hong Kong is strategic or residual for QTS. If customers are using QTS Hong Kong as part of multi-region QTS or Blackstone-supported relationships, the site may be more valuable than the public footprint suggests. If customers are mostly local, legacy or short-term, the facility may be a useful Hong Kong node without being a strategic platform. If there are no material live customers tied to QTS Hong Kong, then the public name is mostly an option, and the article's investment case would shift from operating economics to whether the sponsor will redeploy capital into Hong Kong through QTS, AirTrunk or another vehicle.

What Singapore and Tokyo Do to the Quote

The Hong Kong quote is never negotiated in isolation. The buyer has Singapore and Tokyo on the same spreadsheet because each city disciplines the other two. Singapore tells the buyer what policy-filtered scarcity costs. Tokyo tells the buyer what a larger, engineering-heavy, domestic-market hub costs. Hong Kong tells the buyer what China-adjacent financial latency costs. The final decision may include all three, but the workload split depends on the price of keeping each function close.

Singapore's role is the cleanest comparator. IMDA's Green Data Centre Roadmap at https://www.imda.gov.sg/how-we-can-help/green-dc-roadmap says Singapore aims to provide at least 300 MW of additional capacity and supports further capacity through green energy. The Singapore Economic Development Board's summary at https://www.edb.gov.sg/en/business-insights/insights/singapore-to-expand-data-centre-capacity-by-at-least-one-third-pushes-for-green-energy-use.html says at least 300 MW will be added in the next few years, with another 200 MW allocated only for operators using green energy options. Singapore offers legal clarity and cloud density, but it is reliable, not frictionless.

Tokyo disciplines the Hong Kong quote from the other side. JLL's Japan note at https://www.jll.com/en-jp/insights/japan-data-centre-market-opportunities points to power concerns as generative AI usage rises and says Japan's AI system market is projected to reach JPY 4.2 trillion by 2029. Tokyo can absorb workloads that need Japan's domestic market, Japanese enterprise customers or North Asia resilience, but it brings its own power, land, earthquake, construction and latency trade-offs.

Hong Kong earns the workloads that cannot be moved without damaging the business case: latency-sensitive financial systems, local market operations, China-adjacent customer routes, Hong Kong cloud-region interconnect and resilience designs that need to sit near Hong Kong staff and institutions. QTS Hong Kong's quote, if presented to this procurement team, would be tested on whether it provides those Hong Kong-specific benefits or only QTS global comfort. A high-quality powered hall with private cloud access, dense local carriers and credible cloud on-ramps can stand beside Singapore and Tokyo. A sparse facility with limited connectivity evidence becomes a small edge footprint while larger workloads sit elsewhere.

The buyer's final architecture is likely to be layered: Hong Kong for low-latency financial and China-adjacent operations, Singapore for regional governance and Southeast Asia reach, Tokyo for Japan and North Asia resilience, and cloud regions across all three where managed services make sense. QTS Hong Kong's opportunity is to own the Hong Kong layer with enough proof that the buyer does not treat it as merely temporary.

The Margin Test Under Each Rack

The margin in a Hong Kong data-centre facility begins with scarce powered space, but it is earned through disciplined pass-throughs, tenancy, support load and expansion timing. A rack price can look high and still be unprofitable if power, cooling, rent, financing and labour have been mispriced. A rack price can look expensive to the customer and still be rational if it prevents latency loss, operational risk and migration complexity.

Power is the first line. At the CLP 2026 average net tariff of HK$1.406 per kWh from https://www.clp.com.hk/content/dam/clp-group/channels/media/document/2025/20251118_en.pdf, a 20 kW cabinet running continuously consumes 14,400 kWh in a 30-day month before overhead. That is about HK$20,246 in raw electricity input before cooling, UPS losses, power distribution, generator fuel, maintenance and margin. A 10 kW cabinet is about half that. A 40 kW high-density cabinet doubles it. If the operator sells power-inclusive pricing without tight usage controls or escalation clauses, power volatility can move from customer problem to operator problem.

Cooling, rent and labour are the next lines. Hong Kong's climate makes heat rejection and humidity control central to operating cost, especially in converted industrial shells with limits around shafts, rooftop plant, chilled-water systems and maintenance access. The operator must also amortize electrical work, cooling plant, security, fire suppression, monitoring, generators, batteries and networking over customer terms that may be shorter than the asset life. Financial and cloud customers add labour intensity: remote hands, incident communication, access control, maintenance windows, documented procedures and quick escalation.

Interconnection is the fifth line and sometimes the best margin. Cross-connects, carrier rooms, private cloud access and network services can produce recurring fees with less power intensity than compute space. But they require ecosystem density. A facility cannot charge much for cross-connects if customers still need metro circuits to reach the real meet-me points. That is why the absence of visible public peering or facility rows on https://www.peeringdb.com/net/14383 is economically meaningful. It does not settle the case, but it puts the burden on the private sales file.

The QTS/Blackstone advantage is that the global platform understands these variables at scale. QTS's own leadership profile at https://q.com/our-leadership/ talks about contracted power above 3 GW, and QTS's access-to-infrastructure page at https://q.com/why-qts/access-to-infrastructure/ makes utility access and supply chain part of the proposition. The Hong Kong question is whether those capabilities are locally active. If they are, the margin test can work: scarce capacity, disciplined power pricing, sticky financial and cloud customers, and portfolio-backed execution. If they are not, the facility can be caught between high Hong Kong costs and customers that compare the quote against Singapore, Tokyo and public-cloud alternatives.

The Documents That Would Change the Underwriting

The power file would matter first. It should show committed utility capacity, current load, reserved expansion, substation source, redundancy design, tariff treatment, pass-through clauses, generator capacity, fuel contracts, UPS topology, cooling capacity and utility correspondence. A single megawatt number is not enough. The buyer needs to know how much power is usable today, how much is sold, how much is reserved and how much can be delivered without a multi-year infrastructure project.

The site-control file would come next: lease or ownership, remaining term, renewal options, landlord consents, zoning, permits, building-use approvals, plant rights, riser rights, floor-loading certificates, roof and plant-room rights, and reinstatement obligations. In Hong Kong, a site can look attractive until lease term, fragmented ownership, building constraints or permissions make expansion uneconomic. A QTS or Blackstone balance sheet helps only if the site rights are strong enough to receive capital.

The interconnection and customer files would decide whether the asset is strategic. The interconnection map should identify carriers, meet-me rooms, cloud on-ramps, metro transport, cross-connect counts, latency to key Hong Kong financial and cloud locations, and routes to Singapore and Tokyo. The customer schedule should show tenant credit categories, contract terms, renewal dates, power commitments, pricing escalators, service-level credits, expansion options, churn history and concentration. The old QTS public filing at https://www.sec.gov/Archives/edgar/data/1577368/000162828021003456/qts-20201231.htm showed a diversified customer base at company level in 2020; a Hong Kong site could still be much more concentrated.

The operating and sponsor files would close the case. Operating evidence should show uptime, incident reports, maintenance history, security audits, access logs, staff coverage, remote-hands response times, cooling events and customer credits. Sponsor evidence should explain whether QTS Hong Kong is standalone, part of a QTS customer relationship, linked to Blackstone's wider data-centre portfolio, or potentially rationalised against AirTrunk's Hong Kong presence. The demand book should show live quotes, lost deals, requested power densities, cloud-on-ramp demand, financial customer requirements, mainland customer demand and price comparisons against Singapore and Tokyo.

If those files are strong, QTS Hong Kong is more than a sparse public entry. It is a scarce gateway-city capacity position backed by a sponsor that can finance difficult infrastructure. If those files are weak, the QTS Hong Kong name should be treated carefully: real enough to research, not strong enough to underwrite without proof.

Sources and Signals

The public record for this article rests on three evidence groups. Hong Kong market evidence comes from the Digital Policy Office fact sheet at https://www.datacentre.gov.hk/en/downloads/HK_as_DC_prime_location.pdf, the "Why Hong Kong" page at https://www.datacentre.gov.hk/en/accommodating_data_centres/why_hk.html, the site-finding page at https://www.datacentre.gov.hk/en/accommodating_data_centres/find_a_site.html, the 2024 Legislative Council reply at https://www.info.gov.hk/gia/general/202406/05/P2024060500513.htm, the 2020 concessionary-measures review at https://www.datacentre.gov.hk/en/downloads/Data_Centre_Study_Report_2020_Eng.pdf, CLP's 2026 tariff release at https://www.clp.com.hk/content/dam/clp-group/channels/media/document/2025/20251118_en.pdf, CBRE's 2026 Hong Kong release at https://www.cbre.com.hk/press-releases/apac-data-centre-report-2026-hong-kong-remains-resilient and Cushman & Wakefield's global comparison at https://www.cushmanwakefield.com/en/insights/global-data-center-market-comparison.

QTS and Blackstone evidence comes from PeeringDB's QTS Hong Kong record at https://www.peeringdb.com/net/14383, Data Center Catalog's Hong Kong list at https://datacentercatalog.com/hong-kong-gridview, QTS leadership at https://q.com/our-leadership/, QTS access-to-infrastructure material at https://q.com/why-qts/access-to-infrastructure/, QTS sustainability material at https://q.com/2024-sustainability-report/, Blackstone's QTS acquisition release at https://www.blackstone.com/news/press/qts-realty-trust-to-be-acquired-by-blackstone-funds-in-10-billion-transaction/, QTS's parallel release at https://q.com/news/qts-realty-trust-to-be-acquired-by-blackstone-funds-in-10-billion-transaction/, Blackstone's AirTrunk discussion at https://www.blackstone.com/insights/article/behind-the-deal-airtrunk/, QTS's 2020 annual report at https://www.sec.gov/Archives/edgar/data/1577368/000162828021003456/qts-20201231.htm and QTS's 2020 results release at https://www.prnewswire.com/news-releases/qts-reports-fourth-quarter-and-full-year-2020-operating-results-301229310.html.

Interconnection and comparator evidence comes from AWS Direct Connect locations at https://aws.amazon.com/directconnect/locations/, the AWS Hong Kong region post at https://aws.amazon.com/blogs/aws/now-open-aws-asia-pacific-hong-kong-region/, Equinix's cloud-on-ramp explainer at https://blog.equinix.com/blog/2024/01/30/what-is-a-cloud-on-ramp/, Equinix Hong Kong at https://www.equinix.com/data-centers/asia-pacific-colocation/china-colocation/hong-kong-data-centers, HKIX at https://www.hkix.net/hkix/whatishkix.htm, Internet Society Pulse at https://pulse.internetsociety.org/en/ixp-tracker/country/HK/, Singapore IMDA at https://www.imda.gov.sg/how-we-can-help/green-dc-roadmap, Singapore EDB at https://www.edb.gov.sg/en/business-insights/insights/singapore-to-expand-data-centre-capacity-by-at-least-one-third-pushes-for-green-energy-use.html and JLL Japan at https://www.jll.com/en-jp/insights/japan-data-centre-market-opportunities.

The unresolved facts are material. Public sources do not disclose QTS Hong Kong's current site control, live customer book, power commitments, facility tier, expansion rights, occupancy, pricing, direct cloud-on-ramp availability or local operating staff. Those missing facts do not weaken the Hong Kong scarcity thesis. They define the underwriting boundary around QTS Hong Kong itself.