Top 9 Cloud Giants to Spend $830B in 2026 CapEx

Cloud Providers Raise 2026 Capital Spending to a Staggering $830 Billion

The world's largest cloud service providers (CSPs) are set to spend an estimated $830 billion in capital expenditures during 2026, according to the latest AI industry research from TrendForce. The upward revision — driven by surging AI infrastructure demand — pushes year-over-year growth from an already aggressive 61% to a remarkable 79%, underscoring the unprecedented pace at which hyperscalers are building out their AI capabilities.

The revised forecast covers 9 major CSPs spanning both Western and Chinese markets. On the U.S. side, Google, Amazon Web Services (AWS), Meta, Microsoft, and Oracle have all recently raised their 2026 capital expenditure guidance. Chinese counterparts ByteDance, Tencent, Alibaba, and Baidu round out the group, contributing significantly to the upward revision.

Key Takeaways at a Glance

• Total 2026 capex forecast: $830 billion across 9 major CSPs, up from previous estimates
• Year-over-year growth: Revised upward from 61% to 79%
• Primary driver: Strong and accelerating enterprise AI demand
• U.S. CSPs leading: Google, AWS, Meta, Microsoft, and Oracle all raised guidance
• Chinese CSPs contributing: ByteDance, Tencent, Alibaba, and Baidu included in forecast
• Source: TrendForce's latest AI industry research report

North American Hyperscalers Lead the Spending Surge

The bulk of the upward revision stems from North American cloud giants repeatedly raising their capital expenditure guidance in recent quarters. Microsoft alone has signaled plans to invest upwards of $80 billion in AI-ready data centers during its current fiscal year, while Meta has raised its 2025 capex outlook to between $60 billion and $65 billion — a figure many analysts expect will climb further into 2026.

Google parent Alphabet has been equally aggressive, committing $75 billion in 2025 capital spending and signaling no slowdown ahead. AWS, meanwhile, continues to expand its global data center footprint at breakneck speed, with CEO Andy Jassy describing current AI infrastructure capacity as 'nowhere near sufficient' to meet customer demand.

Oracle represents perhaps the most dramatic trajectory shift among the 5 U.S. providers. Once considered a legacy enterprise software company, Oracle has pivoted aggressively into cloud infrastructure, striking major AI partnerships and building out GPU-dense data centers. Its capex growth rate now rivals that of its much larger peers.

Why the 79% Growth Rate Matters

To put the 79% year-over-year growth in perspective, capital expenditure increases of this magnitude are virtually unprecedented in the technology sector. Even during the original cloud computing buildout of the 2010s, annual capex growth among hyperscalers rarely exceeded 30% to 40%.

The revised growth rate — up from an already substantial 61% — signals that AI demand is not only sustaining but accelerating. Several factors explain this:

• Enterprise AI adoption is moving from experimental pilots to production-scale deployments
• AI model training requires exponentially more compute with each generation
• Inference workloads are scaling rapidly as AI-powered applications reach hundreds of millions of users
• Sovereign AI initiatives are pushing CSPs to build localized infrastructure across multiple geographies
• Competition for AI leadership is intensifying, making underinvestment a strategic risk

Compared to 2023, when combined capex for these 9 providers hovered around $250 billion to $300 billion, the projected $830 billion for 2026 represents a near-tripling of spending in just 3 years. This is a capital deployment cycle unlike anything the tech industry has witnessed before.

Chinese CSPs Add Fuel to the Fire

While U.S. companies dominate the headlines, Chinese cloud providers are contributing meaningfully to the global capex surge. ByteDance — the parent company of TikTok — has emerged as one of the world's largest purchasers of AI chips, reportedly ordering tens of thousands of GPUs despite U.S. export restrictions on the most advanced Nvidia chips.

Alibaba Cloud, China's largest public cloud provider, has announced plans to invest more in AI infrastructure over the next 3 years than it did in the entire previous decade. Tencent is similarly scaling its AI compute capacity, driven by demand for large language models powering its WeChat ecosystem and enterprise cloud services.

Baidu, often called 'China's Google,' continues to invest heavily in its Ernie AI model ecosystem and autonomous driving infrastructure. Even amid a more cautious Chinese economic environment, these 4 companies show no signs of pulling back on AI-related capital spending.

The inclusion of Chinese CSPs in TrendForce's forecast highlights a critical dynamic: the AI infrastructure race is genuinely global, with parallel buildouts happening across geopolitical boundaries despite chip export controls and trade tensions.

Ripple Effects Across the Supply Chain

An $830 billion capex figure does not exist in isolation. It sends shockwaves through every layer of the technology supply chain, creating massive demand for:

• AI accelerators: Nvidia, AMD, and custom silicon from Google (TPUs), Amazon (Trainium/Inferentia), and Microsoft (Maia) all benefit
• High-bandwidth memory (HBM): SK Hynix, Samsung, and Micron are racing to expand HBM production capacity
• Advanced packaging: TSMC's CoWoS capacity remains the critical bottleneck for AI chip production
• Networking equipment: Broadcom, Arista Networks, and others see surging demand for high-speed data center interconnects
• Power infrastructure: Data center energy consumption is driving unprecedented demand for power generation and cooling solutions
• Construction and real estate: Data center construction firms and industrial real estate developers face record backlogs

Nvidia stands as perhaps the single largest beneficiary of this spending wave. The company's data center revenue has already grown from $15 billion in fiscal 2023 to over $115 billion in fiscal 2025, and the trajectory suggests further acceleration as next-generation Blackwell and Rubin architectures ramp into production.

The supply chain implications extend well beyond semiconductors. Electrical utilities, fiber optic cable manufacturers, and even steel and concrete suppliers are feeling the pull of the data center construction boom.

What This Means for Businesses and Developers

For enterprise technology leaders, the massive capex surge carries several practical implications. First, it signals that cloud AI services will become increasingly abundant and potentially more cost-competitive as providers achieve greater economies of scale. Companies that have been waiting for AI infrastructure costs to decline may find 2026 and 2027 to be inflection points.

Second, the spending indicates that CSPs are betting heavily on AI becoming a core enterprise workload — not a niche experiment. Organizations that delay their AI strategies risk falling behind competitors who are already building on these rapidly expanding cloud platforms.

For developers and startups, the infrastructure buildout means more accessible GPU compute, more diverse model hosting options, and better developer tooling. The competition among 9 major CSPs for AI workloads will likely drive innovation in pricing models, deployment tools, and managed AI services.

However, there is a counterpoint worth considering. Some analysts have raised concerns about overinvestment risk — the possibility that AI infrastructure spending could outpace near-term revenue generation, leading to a correction similar to the fiber optic overbuild of the early 2000s. So far, however, demand signals remain robust enough to justify the spending trajectory.

Looking Ahead: Can the Pace Sustain?

The central question facing the industry is whether 79% year-over-year capex growth can be sustained beyond 2026. Historical precedent suggests that spending cycles of this magnitude eventually moderate, but several structural factors could extend this one.

The transition from AI training to large-scale inference is still in its early stages. As AI agents, autonomous systems, and real-time AI applications proliferate, inference compute demand could dwarf current training requirements. This shift alone could justify continued infrastructure expansion well into the late 2020s.

Additionally, next-generation AI models — including multimodal systems, reasoning models, and domain-specific foundation models — require progressively more compute for both training and deployment. The scaling laws that have driven AI progress show no signs of hitting hard limits in the near term.

TrendForce's revised forecast of $830 billion in 2026 capex may prove to be conservative if AI adoption continues at its current pace. The firm has already revised its estimates upward once, and further revisions remain possible as additional CSPs update their spending plans throughout the year.

What is clear is that the AI infrastructure buildout represents the largest capital deployment cycle in technology history, and its effects will reshape industries far beyond cloud computing for years to come.