How it works
“…we have to create a system that allows AIs to learn the common sense of the physical world… and the data is quite scarce.” Jensen Huang, Nvidia, CES 2026
The Apple of cities. One OS – Orchestr8.
Not in the wave's sense, a learned model that generates and simulates 3D space. HarborX is the operating layer of a real district, with a live model of the island, its digital twin, that simulates the build-out, the crowds and the traffic to plan against. World models plug into it; the island's data is what they are short of. The comparison on the value slide is about what the wave's buyers pay for position, not about the technology.
The diligence a lead would do, on one slide. Nothing here is stronger than the paper behind it.
| Item | Where it stands | |
|---|---|---|
| The city corporation's support for the proposal, and CommonX as its lead developer and interface for the programme | In the city corporation's own text of the memorandum | In place |
| The exhibition centre: funded by the LGU and the reclamation authority, under construction, the pilot's first venue | Funded and being built now | In place |
| The data room: the financial model live, the revenue register, every stream priced, this plan, the technology programme line by line | Open to named readers | In place |
| The city corporation's participation in the venture | In the schedules proposed with the memorandum; settled in the definitive agreements | To be agreed |
| The commercialisation rights: the memorandum gives CommonX the mandate to negotiate them; the definitive agreements grant them, before the first event | Mandate in the memorandum; grant in the definitive agreements | To be agreed |
| Ticketing on the app at the venue | To be written into the venue technology agreement | To be agreed |
| The curb agreement with the LGU: parking, pickup bays, road-pricing collection | Six streams depend on it; none is counted before it is held | Open |
| The city corporation's two first-customer contracts: the city operations dashboards and the digital-twin licence, priced at arm's length | Proposed for signature before the first closing, so the revenue is contracted, not forecast | To be agreed |
| The licences the platform needs: the wallet, the network, fare-charging transit, paid advertising | Held by licensed local partners or a Filipino-majority company where the law requires it; nothing on the island depends on a licence the venture does not yet hold | Structure work |
| The lead of the first closing | A strategic technology partner in discussion | In discussion |
Every stream in the register carries the agreement it depends on.
“…unlike language models, training data is scarce for today’s [physical AI].” Fei-Fei Li, November 2025
“…market fragmentation, and interoperability bottlenecks in spatial AI prevent scalable, sustainable revenue growth.” Gartner report featuring Niantic Spatial, June 2026
▷ Try the simulator and dashboard▷ Play with the super app▸ Request the data room
Also: New York City (every ride-hail trip’s pickup and drop-off), California (robotaxi operating data), drone Remote ID in the United States and Japan. On the island the same rules would be the city’s, set through its permits, as Los Angeles does; no national law is needed.
The city, why they get the app, one day, how it earns, proof, the ground; then the financials, what it could be worth, the engine in detail, where it leads, the precedents, the round, the timeline, the people. Every stream's own working is in the data room.
Every visitor stream is priced on one crowd, so this is the first number to check.
Event visits are the venue owner's own projection in the project's approval documents (NEDA Preliminary Report, ICC submission, Table 9): 20,000 visitors an event and six events a month, 1,440,000 a year at the full programme.
The timing. Year 2 is the handover of the completed works. The venue may hold a limited number of events in Year 2, and none of them is counted: the exhibition centre's events count from Year 3. Years 3 and 4 carry 57% and 79% of the full programme, taken from a launch calendar by event type (expos, trade shows, conventions, corporate events and concerts, each as events × days × people a day) that reaches the full programme in Year 5 and grows 3% a year after it.
Everyone on the island is the city corporation's building schedule: its first buildings are delivered in Year 3, and residents, workers and visitors grow with each stage.
When a visitor pays through the wallet, the merchant pays a small processing fee. The wallet is accepted everywhere and is never the only way to pay.
When the app sends a visitor to a restaurant, a ride or a show, the business that gained the customer pays for the introduction or the booking.
Where the platform sells the product itself, such as a pass or a ticket, it keeps a published share and pays the rest to the operator or the organiser.
Every figure is what the platform keeps, never the money that moves through it. Figures are in constant US dollars at ₱58 to the dollar.
Three methods, from the same cash flows the slides run on. Projections of value before any shareholder's terms; the price of a round is set with its lead, and a fund buys a multiple on that price, not a size. No upside is in these figures: not the other cities and smart hubs in prospect, not the data. The market may price HarborX well above these methods, for its position as the engine of predictive urban intelligence and for the data only it will hold.
A spatial-intelligence company with no revenue ever disclosed was bought for a reported US$8.2 billion: a team, two models and a head start, paid in stock by a chipmaker worth close to a trillion dollars, 52% above a round seven months earlier. Its seed investors made about 6.5× after dilution. Not a multiple of revenue: the price of position to a buyer that needs it. All figures reported.
Receipts: a venue under construction with the platform designed in, the city corporation's first contracts, customised streams on Philippine precedents. Every US$100M of first-closing post-money is worth – by Year 10 on ordinary methods. That is the floor a cash business sets; strategic buyers pay for position on top of it: the island's default rail, a running reference city, the ground after the pilot.
A spatial-intelligence company with no revenue disclosed was bought for a reported US$8.2 billion. Capital is moving into world models now, and nobody has a real city's data yet. HarborX will, with a real customer signed.
A closed commercial district built around the largest exhibition and convention centre in Southeast Asia, with hotels and resorts beside it and a seawall full of attractions, where the platform is the city's default from the first event. The city corporation is the design partner and the paid first customer. The whole city follows: residents move in after the pilot island's first years, and the population curve takes off.
The shuttle, the ticket, the room, the payment, the memories, in one app.
Before you arriveShuttle, parking, hotel and ticket in one tap.
While you are hereNo queues, no cash, no keys. Tap to pay, ride, open the room, earn points.
After you leaveThirty seconds after checkout the day comes back as a story, not a receipt: your photographs, added by tap, never taken; the points from the table, spent on tomorrow's ride.Nobody pays for the HarborX app. It is the way in, the way around and the way to pay less.
01 · The ride onto the islandShuttle, transit and parking, booked in the app.
02 · Free Wi-FiAcross the island, every day.
03 · The ticket to every eventEvery ticket on the island, in the app.
04 · Member prices and pointsIn the app only, at every merchant.Exclusive from day one: never the only option, always the default one.
| Engine | What happens | What it earns |
|---|---|---|
| The Visa model | The wallet takes the payment | a share of the bill |
| The Bloomberg model | His choice joins the anonymised dining data | licensed market data, recurring |
| The Salesforce model | The restaurant reads its own dashboard | monthly SaaS per tenant |
| The App Store model | A delivery partner fulfils on our rails | revenue share per transaction |
| The SI model | The restaurant's systems run on the platform | annual recurring per tenant |
A feature earns once. A platform earns every time anything happens: customised streams, city-wide.
The data is governed with the city corporation: aggregated and anonymised, sold, never taken. The city's data is also served to agents as tools (MCP). The detailed diagram is in the appendix.
One city, no upside counted: not the other cities and smart hubs already in prospect, and not the data: the most valuable asset a city will ever produce, which this platform both consumes and generates with predictive intelligence. The model of record, before any shareholder's terms.
The city corporation is the design partner, participates in the venture and buys first: the dashboards and the digital twin. The venue's ticketing is on the app from Year 3.
▷ Try the simulator and the city operator dashboard▷ Play with the super app▸ Request access to the data room
From the teams that built the operating systems behind the MagicBand and the Ocean Medallion, and Starbucks Mobile Order & Pay.
Reported market figures: smart-city platforms, US$24.5 billion in 2025 to US$39.5 billion by 2030; physical AI, the industry world models serve, US$60 to 100 billion by 2030; world-model developers raised a reported US$2.3 billion in one half-year.
A global leader in smart city design and electronics leads, and half of it is filled. The rest is open to a co-lead and to investors on identical terms. The round re-opens in Year 3 at a higher valuation.
Ingestion and sensor fusion; the cognitive engine and the guest genome; closed-loop operations and one command centre; personalisation for businesses and individuals; the world model and future planning.
A seed round in two closings into the venture every revenue line on these slides belongs to. A global leader in smart city design and electronics leads, and half of it is filled; the rest is open to a co-lead and to investors on identical terms.
What the money does. The team and the platform's own build, and the business's share of the city's technology capital, less what the platform keeps.
| What can be checked, and where | |
|---|---|
| Every stream's price, working and Philippine precedent | the register |
| The crowd: the venue owner's projection and the city's schedule | the model · the people appendix |
| The technology programme: US$743.5M to full build-out, 236 lines | the model |
| Revenue, operating costs and capital, year by year | the financials appendix |
| The enterprise value on three methods | the model · the ENTERPRISE VALUE tab |
The business on its own, before any shareholder's terms. Time is counted in months and years from the start of the programme.
One city, no upside counted. US$ millions in constant prices: what the platform keeps, what it spends, what it builds, and the city's technology capital it pays for, less what it recovers. Other cities are planned, with prospects already in hand; none of them is in these figures.
Operating costs are the team, cloud and infrastructure, tools, office and the ground team; payroll capitalised as build is counted once, in the platform's own capital. The later years are summed in the last column.
Visits to the exhibition centre come from the venue owner's projection and its opening ramp. The daily population comes from the city corporation's building schedule: the capacity of the buildings delivered by each year. Visitors are split into five kinds by the model's mix. The venue's visitors a day are the event visits ÷ 365, counted in the headcount at the share set above.
Each bar is a stream, largest first, showing what it earns over Years 2 to 5 in US$ thousands. Select a bar to see the stream below, or a chapter name to go to its slides. Each stream's panel says in plain words what one unit earns, how many units the figure needs against the crowd, and why the figure is a cautious one: these are the researched pilot figures, priced inside the Philippine precedents on 65 of the 75 lines.