ARK Invest Forecasts AI Infrastructure Spending to Reach $1.5 Trillion by 2030

Falling costs of training neural networks are paradoxically driving greater demand for computing power — cheaper technology opens up new use cases. ARK Invest estimates that annual global investment in AI infrastructure could reach $1.5 trillion by 2030, tripling over five years.

"AI adoption is outpacing the internet, and infrastructure is scaling to match. We believe this is the beginning of a massive buildout, as consumers and enterprises signal strong demand." — ARK Invest (@ARKInvest), original post

Why This Matters

Artificial intelligence is transitioning from an experimental technology to a core component of enterprise operations worldwide. The scale of investment in AI infrastructure is now comparable to telecom spending during the early internet era. For crypto markets, this trend holds direct relevance: growing demand for computational resources fuels decentralized compute projects, and massive AI capital expenditures shape the narrative around AI-related tokens.

Prices Drop, Demand Surges

According to ARK Invest's analysis, training costs for neural networks are declining by 75% annually. Inference costs for models scoring above 50% on benchmarks are falling even faster — at roughly 95% per year.

Training and inference cost reduction dynamics. Source: ARK Invest

Yet lower costs are not reducing total expenditure. Instead, affordability widens the range of economically viable applications. AI adoption is occurring at twice the speed of the internet: penetration reached 20% in just three years, compared to over six years for the World Wide Web.

Enterprise demand is the primary driver. Token requests routed through OpenRouter have grown 28x since December 2024. Anthropic scaled its annualized revenue from $100 million in 2023 to $14 billion by February 2026. OpenAI reached 1 million business customers by November 2025.

Corporate AI adoption trajectory. Source: ARK Invest

The Infrastructure Investment Boom

Since ChatGPT's launch, demand for accelerated computing has skyrocketed. Nvidia's annual revenue climbed from $27 billion in 2022 to $216 billion in 2025, with analysts projecting $350 billion for 2026. Global server system investment growth accelerated from 5% annually (in the decade prior to 2022) to 30% over the past three years.

ARK's data shows that GPU and ASIC-based solutions now dominate 86% of the server computing market. Private investment in AI infrastructure exceeded $200 billion in 2025, with approximately $80 billion flowing to foundation model developers.

Hyperscalers are pursuing alternative financing structures: Meta's $30 billion deal with Blue Owl became the largest private capital transaction in history.

The Chip Battle Intensifies

AMD has matched Nvidia on total cost of ownership (TCO) for inference of smaller models. However, Nvidia retains its performance lead in the heavy model segment thanks to its Grace Blackwell architecture.

Competition among AI chip manufacturers. Source: ARK Invest

Major cloud providers are developing proprietary semiconductor solutions. Google has been designing TPUs for 10 years, and SemiAnalysis estimates that custom chips reduce compute costs by up to 62% compared to Nvidia architectures. Amazon is pushing its Trainium accelerators, which have become the preferred training platform for Anthropic. Microsoft is deploying its second-generation Maia accelerators, optimized for inference workloads.

Broadcom dominates backend chip design, partnering with Google TPU, Meta MTIA, and OpenAI's upcoming chip. Citi forecasts Broadcom's AI revenue will grow from $20 billion in 2025 to $100 billion by 2027.

Startups are also making moves. Cerebras, known for its Wafer Scale Engine chip, plans to go public this year. Groq has signed a $20 billion licensing agreement with Nvidia.

ARK's Forecast: $1.5 Trillion by 2030

ARK Invest projects that annual AI infrastructure investment will hit $1.5 trillion by 2030. The share of specialized ASIC chips in total compute capacity is expected to grow to one-third of the market.

Projected annual AI infrastructure investment. Source: ARK Invest

ARK's analysts emphasize that the infrastructure being built today is not a speculative bubble but the foundation for a once-in-a-generation platform shift. AI agents — still in their early deployment phase — are compute-hungry yet far more capable than current tools. Scaling these agents across millions of businesses will require massive computing resources, justifying the current level of capital expenditure.