Philippines: Computing Sovereignty Between USA and China

A precise precedent: Penang, 1972

In 1972, Intel opened its first assembly plant outside the United States in Penang, Malaysia. The mechanism was straightforward: labor, neutrality, access. The country offered intermediate ground between Cold War blocs, and American capital found a low-friction pathway to production.

The structure lasted for decades. Malaysia became a node in the global chip assembly network, capturing value that neighbors pursued for a generation.

A medium-sized country gains weight when it becomes the place where two powers prefer to operate instead of clash.

Penang was exactly this. Today the script returns, with compute replacing assembly.

The numbers remain documented. Intel maintained Penang as a base for decades, and Malaysia built a testing and inspection industry around that nucleus.

The current pattern: Manila, September 2026

On September 8, 2026, the Philippines unveiled a seven-year masterplan worth $34.4 billion for artificial intelligence infrastructure, with the goal of attracting $21 billion in private capital by 2033, according to Nikkei Asia^[1].

The same figure appears in BusinessWorld^[2], which documents it as a formal government initiative. Two independent sources converge on the same number.

Manila describes the move as a turning point. It shifts from a soft-skills strategy (call centers, services) toward high-powered computing. The official framing speaks of a regional hub, competing with Southeast Asian neighbors.

The plan spans seven years and aims to redirect the economy toward compute. The scale, $34.4 billion, signals state ambition, far beyond experimentation.

This reading remains incomplete. The substance lies elsewhere.

The mechanism: reducing exit costs

Emerging economies build alternative pathways to access semiconductors and compute to reduce the cost of abandoning single-bloc dependence. Every data center, every GPU supply agreement becomes a negotiating lever.

The mechanism works like this. A country outside the blocs invites capital from both the American and Chinese spheres. Competition between the two suppliers lowers prices and raises the bargaining power of the host country.

The US dollar's share of global foreign exchange reserves has fallen from 71.4% in 2001 to 58% in 2024, according to International Monetary Fund data. The same diversification principle now operates on computing infrastructure.

Three precedents suffice to call it a pattern: Penang 1972, Singapore in the 1980s, India during the Cold War. In each case the host country gained leverage by selling neutrality.

This is a regime change, not a cycle.

The capital financing compute

Global AI capex rests on debt and credit. Central banks, from the Fed to the ECB to the Bank of Japan, set the cost of that money.

When rates stay high, data center financing becomes expensive. Emerging countries compete for scarce capital. The Philippines offer returns and geographic position to capture a share of it.

The 2020-2026 rate cycle has closed the thirty-year bull market in bonds. The subsequent regime penalizes long-duration, capital-intensive projects, a category where compute sits entirely.

Manila enters the game at the most expensive century-point to finance it.

This desk's position

My position: the Philippines are negotiating a reallocation of computing sovereignty between USA-China arrangements. They build regional capacity as a secondary effect; the true objective is positioning.

The market reads the news as local technology capex. It prices data centers, energy, real estate. It misses the structural significance for global control of critical infrastructure.

Consensus sees a regional race between Manila and its neighbors. That frame measures the wrong variable. The real game concerns who controls pathways to compute access.

What would change my reading? If the plan remained confined to generic cloud services, supplied by a single bloc with zero data sovereignty clauses, the thesis would weaken.

Until then, the pattern holds.

Three implications for capital

Three implications for those managing geopolitical exposure over the next 36 months.

1. Family offices and sovereign wealth funds: compute infrastructure in off-bloc countries becomes a separate asset class with geopolitical premium. Horizon: 24-36 months.
2. CEOs and boards: GPU and data center supply in emerging markets carries dual-dependency risk. Horizon: 18 months.
3. Chief Risk Officers: the scenario of fragmented compute access is missing from VAR models. It should be added. Horizon: 12-24 months.

The capital financing AI capex flows from debt and credit that central banks shape through the cost of money. Manila enters this chain as a peripheral node. Its positioning shifts, at the margins, the geography of sovereign risk.

Each item shares the same core. Control of infrastructure matters more than model capacity.

The forecast

The forecast is explicit. By the fourth quarter of 2027, the Philippines will announce at least one compute infrastructure agreement with capital from both American and Chinese spheres, within the masterplan's perimeter.

Confidence: Medium (58%). Horizon: December 31, 2027. Verification: public announcement of two suppliers from distinct geopolitical blocs.

Kill signal: by December 31, 2027, the plan is financed by a single geopolitical bloc, with zero counterparts from the other sphere.

What to watch

Three indicators will tell us how well the thesis holds.

• Private investment rounds toward the $21 billion target: pace and origin of capital.
• Data sovereignty clauses in data center contracts.
• Moves by neighbors (Malaysia, Indonesia, Vietnam) on the same compute ground.

These signals precede outcomes by many quarters. They are signal, not noise.