So here's what actually happened in San Francisco the week ending June 20. Occupancy hit 84.8% (up 17.6% year-over-year). ADR spiked to $301.35 (up 53.5%). RevPAR landed at $255.45... an 80.5% year-over-year jump. The following week, RevPAR was still up 40.7%. Those are real numbers. CoStar confirmed them. The combination of FIFA World Cup matches at Levi's Stadium (43 miles south in Santa Clara, but close enough that San Francisco absorbed the hospitality demand) and the Databricks Data + AI Summit created a compression event that most markets would kill for.

But here's what I keep thinking about. I talked to a GM at a 250-key downtown property last month who told me his biggest fear during compression events isn't demand... it's bandwidth. "When we hit 85% occupancy and half the guests are tech conference attendees running video calls from their rooms, our network falls apart by 10 AM." That's not a hypothetical. That's a Tuesday during Dreamforce. Now stack a World Cup fan base on top of it... people streaming matches, uploading to social, FaceTiming family back home... and you've got a building designed for 2014 internet usage trying to handle 2026 demand patterns. The revenue looks incredible on the STR report. The one-star reviews about WiFi that drop two weeks later tell a different story.

Look, the macro numbers are genuinely encouraging. San Francisco Travel is projecting 24.2 million visitors and $9.9 billion in spending for 2026, which would beat the 2019 record. The Moscone Center has 38 events scheduled, generating an estimated 674,000 room nights. That's a real pipeline. But the city's occupancy is still roughly 10% below 2019 levels... the widest gap among major U.S. markets. So this isn't a recovery that's done. It's a recovery that's happening unevenly, and the compression weeks are masking how much ground still needs to be made up in the shoulder periods. A property that does $301 ADR during World Cup week and $180 the week after hasn't "recovered." It's riding spikes.

What actually interests me about this story is the technology stress test that nobody's writing about. When you push a hotel to 84.8% occupancy with a guest mix that's disproportionately tech-savvy (conference attendees plus international soccer fans), you're essentially load-testing every system in the building simultaneously. PMS check-in queues. Mobile key infrastructure. Payment processing throughput. Network capacity. HVAC demand. I've seen properties where the guest-facing app crashes at 70% occupancy because nobody stress-tested it above the average. Conference compression doesn't care about your average. It hits your ceiling. And if your ceiling is your 1990s-era electrical wiring creating interference on half the access points (sound familiar?), you're going to have a very expensive week where you capture the revenue and lose the reviews.

The real question for San Francisco operators isn't whether the demand is coming. It is. The convention pipeline plus the AI industry presence plus global events have created a demand floor that didn't exist three years ago. The question is whether the infrastructure... physical, technological, and staffing... can handle the peaks without breaking. Because a compression event that drives $255 RevPAR and generates 15 one-star reviews about WiFi, elevator wait times, or understaffed front desks is a net negative over the next 90 days. You made the money this week. You lost the direct bookings for the next quarter. That trade-off doesn't show up in the weekly STR report. But it shows up in your channel mix six months later.

So Hyatt is bringing Alila to Hakone, Japan. Sixty keys. Private natural hot spring bath in every room. Kengo Kuma designing the thing. Opening 2028. And every hotel tech publication is going to write about the "digital guest journey" and the "smart room experience" and whatever other buzzwords get clicks this week.

I want to talk about something else entirely. I want to talk about what happens when you try to wire a luxury technology stack into a property where the core guest experience is... water. Hot water from the earth, piped into 60 individual rooms, each one requiring its own temperature monitoring, flow management, and maintenance alert system. I consulted with a resort group in Southeast Asia last year that had individual plunge pools in every villa. Their "smart room" system looked gorgeous in the demo. In production, the pool temperature sensors threw false alerts every 90 minutes because humidity in the mechanical spaces exceeded what the hardware was rated for. The engineering team disabled the alerts within a month. So now you've got a $200K monitoring system that nobody monitors. That's hotel tech in a nutshell.

Here's what actually matters about Alila Hakone from a technology perspective. This is Hyatt's 10th brand in Japan, joining 22 existing hotels across nine brands. That means Hyatt already has a regional tech infrastructure... PMS standards, loyalty integration requirements, revenue management platforms. But Alila isn't a Hyatt Place. The operational technology for a 60-key ultra-luxury onsen resort has almost nothing in common with the tech stack running a 300-key Grand Hyatt in Tokyo. The PMS needs to handle kaiseki dining reservations with multi-course timing. The guest profile system needs to capture bathing preferences (temperature, minerals, timing) that don't exist as fields in any standard loyalty platform. The spa booking engine needs to manage gender-separated and mixed-gender thermal facilities with capacity limits that change by time of day. None of this is in the standard Hyatt tech playbook. So either they build custom (expensive, slow, maintenance-heavy) or they force-fit existing platforms (cheap, fast, terrible guest experience). I've watched this exact decision get made at four different luxury brands expanding into non-standard property types. They almost always choose force-fit first, realize it doesn't work about eight months post-opening, and then spend 2x building custom anyway.

The building itself is going to be a technology challenge that most people aren't thinking about. Hakone sits inside a national park. The Sengokuhara area has volcanic geology, dense forest cover, and infrastructure that wasn't designed for modern bandwidth requirements. You're putting a luxury resort into a location where the cellular signal might be inconsistent and the nearest fiber trunk line serves a town of maybe 4,000 people. Kengo Kuma's design philosophy is minimalist integration with nature... which is beautiful and also means the architecture probably won't accommodate the cable pathways, equipment rooms, and antenna placements that a modern hotel technology stack requires without some very creative engineering. My family's hotel has 1978 wiring that kills WiFi on the second floor. Now imagine that problem, but the building is deliberately designed to disappear into a mountainside.

Look, I'm not saying Alila Hakone won't be stunning. It probably will be. Kengo Kuma doesn't do mediocre. And the Japan luxury hotel market is projected to grow from about $7.3 billion to over $10 billion by 2034, with 42.7 million international visitors in 2025 alone... so the demand is real. Hilton is putting an LXR property in Hakone for the same reason. But the technology conversation around properties like this always focuses on the guest-facing stuff... the app, the digital key, the in-room tablet. The actual technology challenge is infrastructure. It's the monitoring systems for 60 individual hot spring feeds. It's the network architecture in a building designed to look like it has no technology in it. It's the integration between a hyper-local Japanese hospitality operation and a global loyalty platform that was built for business travelers in Chicago. The Dale Test question here is brutal: when the hot spring feed to room 215 drops below temperature at 2 AM, what does the system do, and can the one person on duty fix it without calling an engineer?

So Manglam Group just committed ₹350 crore (roughly $42 million) to build a 220-key Sheraton on the Jaipur-Ajmer Highway. That works out to about ₹1.59 crore per key... which, for context, is actually cheaper than their previous Westin project in the same city, which ran ₹2.22 crore per key for 135 rooms. The scale economics are showing. More keys, highway-adjacent land (not city center), Sheraton instead of Westin positioning. The development math, on paper, makes sense.

But here's what I keep coming back to. This is Manglam's third collaboration with Marriott, and it's structured as a management contract, not a franchise. That distinction matters enormously. Under a management contract, Marriott operates the hotel. They hire the staff. They control the PMS, the revenue management system, the loyalty integration, the tech stack... all of it. Manglam builds the building, puts up the capital, and then hands the keys (literally) to Marriott to run. For an owner whose core competency is real estate development (125 completed projects, 62 million square feet of built space), this might be the right call. You don't suddenly become a hotel operator because you poured concrete in the right shape. But the technology implications of a management contract versus a franchise are completely different, and nobody in the press coverage is talking about that.

Here's what I mean. When Marriott manages your property, you're running their systems. Period. Their PMS. Their RMS. Their loyalty platform. Their distribution stack. You don't get to shop vendors. You don't get to negotiate integration costs. You don't get to say "actually, we found a better revenue management solution for our market." The tech decisions are made in Bethesda, not in Jaipur. I talked to a hotel owner last year who was three years into a management contract with a major international brand and told me, "I own the building, but I don't own a single data point about what happens inside it." That's not a technology complaint. That's a structural power imbalance baked into the contract.

Now, Jaipur's market fundamentals are genuinely strong. Demand CAGR around 10% versus supply growth of 8%. ADR jumped 20-25% year-over-year as recently as mid-2025. UNESCO World Heritage status, the Golden Triangle tourist circuit, destination weddings, proximity to the Mahindra World City SEZ generating corporate demand... the demand drivers are real and diversified. And Marriott's India pipeline is massive... 200 hotels planned, Series by Marriott already at 75 signed with 50 operational. They're not dabbling in this market. They're flooding it. Which raises the question every owner building into a Marriott-heavy market should be asking: what happens to my ADR when three other Marriott-branded properties open within my comp set in the next five years? The brand that's filling your hotel today is also potentially diluting your rate tomorrow. That's not a conspiracy. That's just how pipeline math works.

The location choice is interesting from a technology infrastructure perspective. Highway corridor development in India means you're building on land that may not have the telecom and power infrastructure of a city-center site. I've consulted with hotel groups building in similar corridors and the WiFi and connectivity buildout alone can add 3-5% to your project cost if the local infrastructure isn't there. And for a brand like Sheraton, where Marriott Bonvoy integration, mobile check-in, and digital key are baseline expectations... your connectivity isn't optional. It's the operating system. If the building's electrical and telecom infrastructure isn't spec'd for what Marriott's tech stack demands on day one, you're retrofitting within 18 months. And retrofitting under a management contract means Marriott tells you what to fix and you write the check. Ask anyone who's been through a Marriott technology standards update mid-contract. The PIP equivalent for tech compliance is a conversation nobody has before signing and everybody has after.

So here's something that should make every hotel technology person uncomfortable: on June 6th, Visa and Mastercard transactions stopped working across Cuba. Not because of a system outage. Not because of a cyberattack. Because a banking partner walked away due to US sanctions, and there was no fallback. No redundancy. No Plan B. An entire country's hotel payment infrastructure went dark because it depended on a single relationship with a single partner, and when that partner left, the system didn't degrade gracefully... it just stopped.

I think about this stuff differently than most people covering it. Everyone's focused on the geopolitics (and fair enough, the new executive order targeting GAESA is massive... we're talking about a military conglomerate that controls roughly 80% of Cuba's foreign exchange and owns 50,000 hotel rooms through its subsidiary). But strip away the politics for a second and look at the technology architecture underneath this collapse. You've got international chains like Meliá terminating contracts on 15 of their 33 properties. Iberostar walking away from 12 hotels. Airlines suspending routes. And now the payment rails are gone. This isn't a market correction. This is a full stack failure... every layer of the technology and distribution ecosystem that makes modern hotel operations possible is being pulled out simultaneously. The Cuban government says national chains will absorb these properties and maintain "normal operations." I've consulted with hotel groups going through technology transitions far less dramatic than this, and I can tell you... there is nothing normal about ripping out an international operator's PMS, revenue management system, distribution connections, loyalty integration, and payment processing all at once and expecting the building to function the next morning.

Look, I get that Cuba's situation is extreme and most of my readers aren't operating under US sanctions. But the underlying vulnerability is universal. I talked to an independent operator last year who had his entire rate management workflow running through a single vendor. When that vendor got acquired and sunset the product with 90 days notice, he was manually updating rates on four OTA extranets for three months while he found a replacement. That was ONE system at ONE hotel. Cuba just lost everything at thousands of properties. The question I'd ask any operator reading this: how many single points of failure exist in your technology stack right now? Your payment processor. Your channel manager. Your PMS vendor. If any one of them disappeared tomorrow (not because of sanctions... because of an acquisition, a bankruptcy, a contract dispute), what's your recovery path? What does the night auditor do at 2 AM when the system that handles 100% of your transactions goes offline and there's no local fallback?

The human cost here is staggering and it shouldn't get lost in the technology analysis. Over 7,000 workers at one resort area alone lost regular employment when hotels closed. Occupancy across the island was 21.5% in the first half of 2025... before the latest sanctions hit. Tourist arrivals are down to 328,608 for the first four months of 2026, a 55.8% decline year-over-year. And the skilled workers who know how to run these properties? They're leaving for other Caribbean destinations that can actually pay them. So even if the political situation resolves tomorrow (it won't), the institutional knowledge is walking out the door. I've seen this at a much smaller scale... a property that lost its entire front office team over six months because management wouldn't address compensation. Rebuilding that team took twice as long as losing it. Cuba is experiencing that at a national level, and the technology infrastructure that could help bridge the gap (automated check-in, dynamic pricing, digital payment alternatives) doesn't exist because the same sanctions that caused the workforce exodus also killed the technology pipeline.

The part that actually keeps me up at night: Cuba's government directed 34.5% of total investment toward tourism in early 2024, up from 27.8% the year before. They're building MORE hotels while existing ones run at 21.5% occupancy. That's not a technology strategy. That's not even an investment strategy. That's building hardware with no software, no distribution, no payment rails, and increasingly no staff to operate it. It's the most expensive version of the "build it and they will come" fallacy I've ever seen... and the workers who depend on these properties for their livelihoods are the ones absorbing the consequences.