Chinese technology giants are rapidly increasing spending on artificial intelligence infrastructure but face greater financial constraints than their U.S. counterparts, whose larger and more profitable core businesses provide stronger cash flow and greater capacity to sustain heavy investment, Moody’s Ratings said on Thursday.

Capital expenditure by leading Chinese technology companies is expected to rise to about $140 billion in 2026 and $165 billion in 2027, from $65 billion in 2025, the rating agency said in a report.

However, the six major U.S. hyperscalers are projected to spend more than $785 billion in 2026, roughly six times the combined total of their Chinese counterparts, with spending expected to approach $1 trillion in 2027.

The gap highlights the different financial positions of the two groups as competition intensifies to build the computing infrastructure needed to support increasingly sophisticated artificial intelligence (AI) models and applications.

Alibaba Group Holding plans to invest RMB 380 billion ($57 billion) over three years, a program announced in February 2025. ByteDance is planning capital expenditure of up to $70 billion in 2026, according to a Bloomberg report in May.

Quarterly results also point to an acceleration in AI-related spending by Chinese companies. For the quarter ended June 2026, Alibaba reported RMB 67.7 billion in capital expenditure, while Tencent Holdings’ spending rose 176 percent year-on-year to RMB 52.8 billion. Baidu Inc.’s capital spending nearly tripled to RMB 11.4 billion.

“While Alibaba noted that its quarterly spending is volatile because of hardware delivery cycles and procurement timing, the increase across the major Chinese hyperscalers indicates the continued expansion of AI spending in China,” Moody’s said.

Despite the faster growth in Chinese spending, the difference in investment levels is expected to result in a widening gap in installed computing capacity.

U.S. data center capacity reached 52 gigawatts (GW) at the end of 2025, compared with 28 GW in China, according to the International Energy Agency (IEA). The IEA expects U.S. capacity to reach 100 GW by 2030, representing a compound annual growth rate of 14 percent, while China’s capacity is forecast to expand to 67 GW, or a 19 percent of compound annual growth rate (CAGR).

“Even though China’s growth rate is faster, the absolute gap in total data center capacity will widen over the forecast period,” Moody’s said.

However, differences in infrastructure costs, labor costs and government support partly offset the headline disparity in capital spending.

Spending on data center facilities, cooling systems and related infrastructure is significantly lower in China than in the United States on a per-megawatt basis, according to International Data Corporation data cited by Moody’s. Lower land, construction and operating costs in China’s major computing clusters also improve deployment economics.

Government support further reduces the effective capital burden on Chinese technology companies, said Moody’s.

It noted that data centers remain a strategic priority under China’s digital and AI industrial policies, including the AI+ initiative and the Six Networks framework, which promote digital infrastructure development and broader AI adoption.

The National Development and Reform Commission estimates that investment linked to the Six Networks framework and related projects will exceed RMB 7 trillion in 2026.

Regional governments are also providing support through measures including land and green-energy access, infrastructure subsidies, faster project approvals and tax incentives.

State-owned enterprises are emerging as increasingly important builders of AI infrastructure. China is reportedly preparing to develop a nationwide network of interconnected AI data centers over the next five years, with most of the technology sourced domestically.

Funding is likely to come from sovereign debt, state-backed investment vehicles, bank financing and private co-investment, said Moody’s.

It noted power availability also gives China an advantage, Moody’s said, as electricity is generally more readily available and costs are lower than in the United States.

However, it sees access to advanced hardware remains China’s biggest constraint.

Servers and accelerators account for the largest share of AI infrastructure spending, and restrictions on access to leading-edge chips from Nvidia have pushed Chinese hyperscalers, state-owned enterprises and state-linked operators towards domestic alternatives.

Although China’s domestic chips and software ecosystem continue to improve, they still lag Nvidia’s technology and ecosystem, Moody’s said.

“As a result, China’s cost advantages are concentrated in non-chip infrastructure, while the U.S. is likely to retain a significant advantage in overall compute capacity,” it said.

Moody’s highlighted that Chinese hyperscalers are also building AI businesses from a materially smaller revenue base than their U.S. peers, although demand for cloud and AI services is growing rapidly.

Cloud businesses at Alibaba, Baidu and Tencent are benefiting from rising AI demand, and Moody’s expects rapid growth in AI-related products and services to drive higher infrastructure investment in 2026 and 2027.

Alibaba said in March that it was targeting more than US$100 billion in annual combined cloud and AI external revenue within five years, underscoring its confidence in the long-term monetization of AI.

Still, Moody’s said U.S. hyperscalers have more established monetization pathways across AI infrastructure, models and applications, supported by large, diversified and highly profitable businesses.

Amazon’s AWS, Microsoft’s Azure, Alphabet’s Google Cloud and Meta Platforms are all expanding rapidly, with AI-related revenue reaching meaningful annualized levels.

Large contracted but unbilled cloud commitments at Amazon, Microsoft and Alphabet also provide multi-year revenue visibility as enterprise AI workloads move from pilot projects into production.

This gives U.S. companies greater capacity to absorb elevated capital expenditure, although AI monetization still trails the scale of investment, said Moody’s.

It sees China’s cloud market remains more price competitive and less driven by enterprise software, constraining hyperscalers’ ability to finance large-scale AI infrastructure solely through cloud cash flow.

Chinese companies are therefore pursuing a lower-cost and higher-efficiency strategy to partly offset their scale disadvantage.

It is noted that open-weight models such as Qwen, ERNIE, Kimi and DeepSeek are reducing deployment costs and accelerating adoption, while several Chinese models now offer performance sufficient for many commercial applications.

A recent study by the Massachusetts Institute of Technology found that inference costs for open-source models are, on average, 87 percent lower than proprietary alternatives.

Growing enterprise adoption, sovereign AI initiatives and the development of domestic technology ecosystems are also supporting rapid growth in token consumption on Chinese cloud platforms, creating an increasingly important avenue for AI monetization.

The ability to sustain heavy AI investment ultimately depends on the resilience of companies’ core businesses and their access to external funding, Moody’s said.

Chinese hyperscalers are investing aggressively despite operating from a narrower earnings base and facing intense competition in their domestic markets.

Alibaba’s e-commerce business faces slower gross merchandise value growth and competition from Pinduoduo and content-driven e-commerce platforms operated by Douyin and Kuaishou.

Competition in instant retail and food delivery has also compressed margins at companies including Alibaba and Meituan.

Baidu’s online marketing business has weakened as advertising budgets shift towards short-video and other social media platforms, while Tencent’s resilient gaming, advertising and fintech businesses continue to support AI investment.

By contrast, U.S. hyperscalers benefit from highly profitable and cash-generative businesses outside cloud computing, said Moody’s.

Digital advertising at Alphabet and Meta, software sales at Microsoft, and retail and advertising businesses at Amazon continue to generate strong operating cash flow, helping fund AI investment while keeping leverage relatively modest.

It sees U.S. companies also benefit from deeper debt and equity capital markets and higher equity valuations, giving them greater financing flexibility.

Chinese hyperscalers generally trade at lower equity multiples, increasing their effective cost of equity, although large companies retain access to offshore and domestic funding markets, said Moody’s.

Alibaba, for example, announced on Aug. 23, 2026, a proposed HK$80 billion (US$10.2 billion) placement of new shares, with all net proceeds earmarked for its full-stack AI capabilities, including infrastructure expansion.

Meanwhile, AI investment is expected to weaken free cash flow and push leverage higher on both sides of the Pacific.

Moody’s expects several companies to turn free-cash-flow negative in 2026 and 2027 as capital expenditure outpaces growth in operating cash flow.

U.S. hyperscalers are increasingly using debt, leases and structured financing to support data center expansion.

Combined existing and future data centre lease commitments for the six major U.S. hyperscalers total about $1.2 trillion, with more than US$820 billion relating to leases that have yet to commence.

Chinese companies face similar pressure but from a smaller starting base, with capital expenditure-to-operating cash flow ratios rising as investment in AI infrastructure and chip procurement accelerates.

Still, Moody’s said substantial cash reserves remain an important credit strength for both groups.

Chinese hyperscalers have built sizeable net cash positions over years of strong cash generation, while their U.S. peers collectively hold several hundred billion dollars in cash and marketable securities.

“These buffers, combined with continued robust revenue and EBITDA growth, provide significant capacity to absorb 12 to 24 months of elevated capex without a material deterioration in credit quality,” Moody’s said.

Southeast Asia’s data centers gather pace with 85 percent of equity raised since 2024 – Tracxn