Researched by BEAST Library Curator | Verified by Harper | Quality: 8.4/10

BLUF: PropTech VC hit a record $16.7B in 2025 (+67.9% YoY) and ran at a $1.7B-per-month pace into 2026 — but the headline number hides a structural shift. The largest owners are no longer buying proptech; they're building it. Anthropic's $1.5B joint venture with Blackstone and Goldman, and OpenAI's $4B "Development Company" raise backed by Brookfield, move foundation-model engineers directly inside the asset owners. For everyone who isn't a top-10 landlord, this is a build-vs-buy decision you should make on purpose — not by default.

The number everyone quotes, and the one that matters

The number you'll see in every deck: PropTech venture funding reached $16.7 billion in 2025, a 67.9% jump that pushed the sector past its pre-pandemic peak. January 2026 alone drew roughly $1.7 billion, up 176% year-over-year. AI-native proptech grew at a 42% annualized clip — nearly double the 24% for non-AI peers.

The number that actually changes your 2026 roadmap is smaller and harder to find: $300 million. That's the individual check Anthropic, Blackstone, and Hellman & Friedman each wrote into a new $1.5B enterprise-AI venture that embeds engineers inside portfolio companies to rebuild core workflows — not sell them a SaaS seat. OpenAI is doing the structurally identical thing through "The Development Company," raising $4B at a $10B valuation from 19 investors including Brookfield, TPG, Advent, and Bain.

Read those two facts together and the thesis is unavoidable: the biggest owners in commercial real estate have stopped evaluating proptech vendors and started growing their own. The capital that used to flow to proptech startups as customers is now flowing around them.

The 2026 capital map

Here's where the marquee money actually went, and what each deal signals for an operator:

Deal Amount / Valuation Date Backers Operator signal
Anthropic × Blackstone, Goldman, H&F $1.5B JV ($300M each) May 2026 Apollo, General Atlantic Owners building custom AI in-house
OpenAI "The Development Company" $4B raised @ $10B val May 2026 Brookfield, TPG, Advent, Bain Same move, larger scale
Bedrock Robotics (Series B) $270M @ $1.75B Feb 2026 — Autonomous construction is fundable
Basis AI (Series B) $100M @ $1.15B Feb 2026 Accel, GV, Khosla, Blankfein Agentic accounting = back-office first
Metiundo (Series A) $47M 2025 — Smart-building / energy is a category
Propsoch — Bengaluru (Seed) $2M Jun 2026 Athera, Sparrow, Vakil Group APAC advisory-layer still seed-stage

Four proptech startups crossed unicorn status since mid-2025, and every one is an AI automation play — property management at scale, investor-relations admin, construction automation, HOA operations. Notice what they have in common: they automate operations a CRE owner already pays humans to do. That's the wedge. It's also exactly the wedge Blackstone and Brookfield just decided to keep for themselves.

What this means if you're not Blackstone

The vast majority of the $12 trillion in APAC commercial building value — JLL counts 260,000+ individual properties worth over $10M each — is held by owners who will never get an embedded Anthropic engineer. The APAC proptech market is forecast to grow 17%+ annually to $51.8B by 2035, yet AI adoption in 2024 sat at just 26% of Asian firms, 27% in Australia, 35% in New Zealand. The capability gap between the giants and everyone else is widening faster than the adoption curve is closing.

For a Taipei, Singapore, or Tokyo asset manager, the giants' "build it ourselves" path is not available — and that's fine, because it's also not necessary. The custom-AI JVs exist to justify $300M checks against portfolios with thousands of assets. At a 5–50 building portfolio, the math inverts: bespoke engineering is the wrong tool, and the region's structural fragmentation (Singapore, Japan, China, and India markets behave nothing alike) defeats any single platform trying to scale across all of them anyway.

Here's what I'd do if this were my portfolio

  1. Stop waiting for the perfect platform. The "one suite to rule them all" never arrives in APAC because the markets are too different. Pick the single operational pain that costs you the most labor — usually energy M&V, lease abstraction, or work-order triage — and solve that one thing first.
  2. Buy outcomes, not lock-in. The lesson from the giants going custom is that workflow ownership is the prize. You can't afford to build, but you can refuse to rent a black box. Favor vendors whose data and models stay portable — if you can't export your building's data and history, you're renting your own operations back.
  3. Run the back office first, like the unicorns did. Basis AI ($1.15B) and the IR-admin unicorns prove the fundable wedge is automating the boring repeatable work. Start where errors are countable and ROI is auditable — accounting, reporting, M&V — before you touch tenant-facing experience.
  4. Treat "AI-native" as a procurement filter, not a feature. A vendor retrofitting AI onto a 2018 codebase is a different risk than one built on it. The 42%-vs-24% growth gap is the market pricing that difference. Ask where the intelligence lives: in the product, or in a slide.

The 90-day move

Do not respond to a record-funding headline by shopping for more software. Respond by auditing one workflow. Inside 90 days, a facility GM can: (1) pick the single highest-labor operational process in the portfolio; (2) baseline its current cost in staff-hours and error rate; (3) pilot one AI-native tool against that one baseline with a clean before/after; (4) make portability — data export, model transparency — a hard contract term. That sequence gives you the giants' discipline (own the workflow, measure the outcome) without their balance sheet.

The reckoning isn't that proptech is dying. It's that the plug-and-play decade is over, and undifferentiated middleware is getting squeezed from both ends — by owners building their own, and by AI-native challengers automating the core. The owners who win the next cycle are the ones who decide, deliberately, which problem is worth solving and which vendor earns the right to keep their data.

For a deeper look at sizing the financial case, see our companion analysis on retrofit and incentive math in the Library, or browse the full PropTech & Capital Markets archive.

Have a question about this topic? Ask our CRE AI Agent →