China's net investment advantage forms macro foundation underlying every AI chip competitive story in 2026.

McKinsey's June 2026 Global Institute report *Catalyzing Competitiveness: Where Investment Happens and Why* delivers a stark finding: China is now the world's largest investor across every metric tracked. On gross productive investment, China logged $5.9 trillion in 2024 (31% of GDP), the United States $5.1 trillion (17%), and the EU-27 $3.1 trillion (16%). In raw gross terms, the US and EU combined still outspend China.

But gross investment measures today's competitiveness. The forward-looking metric is net productive investment—what remains after depreciation. Capital stock deteriorates annually: factories age, grids wear, first-generation data centers lose productive value. Net investment builds tomorrow's capacity.

Here the gap is structural. China posted $4.4 trillion net (23% of GDP) versus the United States' $1.1 trillion (4%) and the EU's roughly $0.7 trillion. China invests roughly four times the US on the measure that compounds into future competitive capacity. The US and EU combined net figure of approximately $1.8 trillion is still less than half of China's.

One additional data point merits attention: India's net productive investment registers at $2.6 trillion, placing India second globally on this forward-looking metric and signaling a capital formation story that Western analysts systematically underweight.

The key insight: China's $4.4 trillion net annual investment funds the physical layer of AI—chips, fabrication capacity, grid upgrades, and data center buildout. This is not an abstract macro statistic; it is the capex pipeline. Baidu's Kunlunxin IPO signals domestic silicon scaling under export controls. China's net investment capacity is what makes homegrown chip supply chains financially viable long-term. Samsung, SK Hynix, and Micron dominate HBM supply today, but China's investment trajectory points toward domestic memory capacity that could reshape this dynamic within a decade.

The net investment gap extends beyond semiconductors. Alibaba's move to restrict Claude Code access is a permission-layer event rooted in capacity. China's investment scale means it can afford to build domestic model alternatives—the net investment gap gives that strategy a credible physical runway.

Most US-China AI commentary focuses on models, benchmarks, and export controls, missing the physical layer. AI runs on chips fabbed in fabrication plants, cooled by energy grids, connected by fiber, and housed in data centers—all requiring sustained capital formation. Net productive investment is the clearest proxy for who is building that physical layer at scale.

The McKinsey figures reveal a structural asymmetry that benchmark comparisons obscure. The United States operates at 4% of GDP in net productive investment; China at 23%. The intensity differential is dramatic—intensity determines how fast an economy's productive base grows. A country investing 23% of GDP net replaces and expands its productive base at nearly six times the rate of one investing 4%. Over a decade, that is not a gap but a divergence.

Baidu building domestic chips, China developing DRAM alternatives, Alibaba hardening its AI stack against Western model dependency—these are not isolated corporate bets. They are expressions of an investment-funded industrial strategy that the net figures make legible.

The Map of AI framework identifies nine layers: infrastructure, energy, semiconductors, memory, data, model training, inference, applications, and distribution. China's net investment lead is most acute in the bottom three—infrastructure, energy, and semiconductor manufacturing. These are precisely the layers export controls target. But export controls slow acquisition rates; they do not substitute for domestic capital formation. A country investing $4.4 trillion net annually has the financial runway to build around almost any constraint, given time.

Chip export restrictions slow China's access to leading-edge Western silicon but do not slow China's capital formation rate. At $4.4 trillion net annually, China has the financial depth to fund domestic alternatives across a sustained multi-year horizon. US policy addresses the symptom (access to specific chips) while the underlying driver (investment capacity) continues compounding.

The United States' gross investment of $5.1 trillion looks competitive. The net figure of $1.1 trillion reveals how much simply replaces deteriorating capital stock rather than expanding it. US AI infrastructure investment—hyperscaler capex, data center construction, grid upgrades—occurs within a system simultaneously aging out its existing base. That squeeze does not appear in gross figures. It appears in net.