diff --git a/marketing/images/singapore-zh.jpg b/marketing/images/singapore-zh.jpg
new file mode 100644
index 0000000..959b84c
Binary files /dev/null and b/marketing/images/singapore-zh.jpg differ
diff --git a/marketing/images/singapore-zh.png b/marketing/images/singapore-zh.png
new file mode 100644
index 0000000..0d0db81
Binary files /dev/null and b/marketing/images/singapore-zh.png differ
diff --git a/marketing/pop-map-proposal.md b/marketing/pop-map-proposal.md
new file mode 100644
index 0000000..3d2d111
--- /dev/null
+++ b/marketing/pop-map-proposal.md
@@ -0,0 +1,92 @@
+# vault1984 — Proposed POP Map (Rationalized)
+
+## Methodology
+- AWS for core markets, Atal Networks for everything else
+- One POP per major region — no clustering in Northern Europe
+- Every inhabited continent covered
+- POPs don't scale vertically — you add POPs, not cores
+- Single Go binary + SQLite per node, ~20MB idle, vault under 1MB
+
+---
+
+## Proposed Map (28 POPs)
+
+### Americas (8)
+| City | Provider | Notes |
+|------|----------|-------|
+| Virginia 🇺🇸 | AWS us-east-1 | US East |
+| San Francisco 🇺🇸 | AWS us-west-1 | US West |
+| Montréal 🇨🇦 | AWS ca-central-1 | Canada — keep |
+| Mexico City 🇲🇽 | LightNode | LatAm North — keep |
+| Bogotá 🇨🇴 | LightNode | LatAm Mid — keep |
+| São Paulo 🇧🇷 | AWS sa-east-1 | LatAm largest — keep |
+| Santiago 🇨🇱 | LightNode | LatAm South — keep |
+| Buenos Aires 🇦🇷 | LightNode | LatAm South alt — keep |
+| ~~Lima~~ | LightNode | Skip — Santiago + BA cover the region |
+
+### Europe (3) — rationalized from 4
+| City | Provider | Notes |
+|------|----------|-------|
+| London 🇬🇧 | AWS eu-west-2 | UK + Western Europe — keep |
+| Zürich 🇨🇭 | AWS eu-central-2 | Central Europe + privacy brand — keep |
+| Madrid 🇪🇸 | AWS eu-south-2 | Southern Europe + Spanish market — keep |
+| ~~Stockholm~~ | AWS eu-north-1 | **DROP** — redundant with London + Zürich |
+
+### Middle East & Turkey (3)
+| City | Provider | Notes |
+|------|----------|-------|
+| Istanbul 🇹🇷 | LightNode | Bridge EU/ME, large market — keep |
+| Dubai 🇦🇪 | AWS me-central-1 | Gulf hub — keep |
+| Riyadh 🇸🇦 | LightNode | Saudi Vision 2030, distinct from Dubai — **ADD** |
+
+### Africa (4)
+| City | Provider | Notes |
+|------|----------|-------|
+| Lagos 🇳🇬 | LightNode | West Africa — keep |
+| Nairobi 🇰🇪 | LightNode | East Africa — keep |
+| Cape Town 🇿🇦 | AWS af-south-1 | South Africa — keep |
+| Cairo 🇪🇬 | LightNode | North Africa + Middle East gateway, 100M people — **ADD** |
+
+### Asia-Pacific (13)
+| City | Provider | Notes |
+|------|----------|-------|
+| Mumbai 🇮🇳 | AWS ap-south-1 | South Asia — keep |
+| Singapore 🇸🇬 | AWS ap-southeast-1 | SE Asia hub — keep |
+| Kuala Lumpur 🇲🇾 | LightNode | SE Asia second node, large market — **ADD** |
+| Jakarta 🇮🇩 | AWS ap-southeast-3 | 4th most populous country, own AWS region — **ADD** |
+| Manila 🇵🇭 | LightNode | 110M people, huge developer workforce — **ADD** |
+| Bangkok 🇹🇭 | LightNode | SE Asia third hub — **ADD** |
+| Hong Kong 🇭🇰 | AWS ap-east-1 | China gateway — keep |
+| Seoul 🇰🇷 | AWS ap-northeast-2 | Korea — keep |
+| Tokyo 🇯🇵 | AWS ap-northeast-1 | Japan — keep |
+| Sydney 🇦🇺 | AWS ap-southeast-2 | Australia — keep |
+| Auckland 🇳🇿 | LightNode | New Zealand — keep |
+| ~~Dhaka~~ | LightNode | Secondary wave — not yet |
+| ~~Karachi~~ | LightNode | Secondary wave — not yet |
+
+---
+
+## Summary
+
+| Region | Current | Proposed | Change |
+|--------|---------|----------|--------|
+| Americas | 8 | 8 | — |
+| Europe | 4 | 3 | -1 Stockholm |
+| Middle East/Turkey | 2 | 3 | +1 Riyadh |
+| Africa | 3 | 4 | +1 Cairo |
+| Asia-Pacific | 9 | 13 | +4 (KL, Jakarta, Manila, Bangkok) |
+| **Total** | **26** | **28** | **+2 net** |
+
+---
+
+## What changed vs. current list
+
+**Dropped:** Stockholm — three Northern Europe nodes is redundant. London and Zürich cover the market.
+
+**Added:**
+- Riyadh 🇸🇦 — Saudi Arabia is spending billions on AI. Different market than Dubai.
+- Cairo 🇪🇬 — 100M people, North Africa gateway, no major vault presence.
+- Jakarta 🇮🇩 — AWS has its own region here (ap-southeast-3). 280M people. Obvious gap.
+- Kuala Lumpur 🇲🇾 — SE Asia second hub. Large Muslim-majority market, underserved.
+- Manila 🇵🇭 — 110M people, massive BPO/developer workforce building agent automations.
+- Bangkok 🇹🇭 — Completes SE Asia triangle with Singapore + KL.
diff --git a/marketing/vault1984-vision.md b/marketing/vault1984-vision.md
index 3f06d18..96acd38 100644
--- a/marketing/vault1984-vision.md
+++ b/marketing/vault1984-vision.md
@@ -22,7 +22,9 @@ We open-sourced it. You don't have to trust us — you can read the proof.
 vault1984 is built to lead in ways that are irrational to chase.
 
 **Global infrastructure no competitor has built**
-22 regions. Every continent. Local providers in markets the major clouds don't reach — Lagos, Nairobi, and beyond. Low latency wherever your agents run. No competitor is close. Every month, the gap widens.
+22 regions. Every continent. Local providers in markets the major clouds don't reach — Lagos, Nairobi, and beyond. No competitor is close. Every month, the gap widens.
+
+But the global network isn't a feature. It's a consequence of the security model. vault1984 never caches credentials — not on your laptop, not in the browser extension, not in the agent. Every access is a live fetch from the vault. The credential exists in your possession only for the instant you use it. Nothing sits around to be stolen. That's a stronger guarantee than any incumbent offers — but it means latency is felt on every single credential lookup. An agent making dozens of calls across a workflow feels every millisecond. The 22 POPs exist because zero-cache architecture requires them.
 
 **Architecture the incumbents cannot copy**
 1Password, Bitwarden, Dashlane — they can read your vault. They choose not to. To give you what vault1984 gives you, they'd have to deprecate every existing vault, migrate every user, and rebuild from scratch. That's not a roadmap item. That's a company-ending decision. They are structurally, permanently trapped.