Goldman Sachs highlights that US technology companies are poised to deliver significant returns on their substantial investments in AI infrastructure, overcoming the challenge posed by low-cost, open-source Chinese AI models.
- US AI infrastructure investments top $700 billion in 2026
- Strong demand and constrained compute supply boost AI adoption
- Chinese open-source models have limited impact on US market dominance
What happened
US technology giants are ramping up investments in AI infrastructure, including data centers and specialized chips, with total spending expected to surpass $700 billion this year. This surge in capital expenditure is fueling increasing demand for agentic AI tools, which operate autonomously to perform complex tasks and drive new efficiencies in enterprise operations.
Goldman Sachs analyst Eric Sheridan points out that while there is an influx of cheaper open-source AI models emerging from China, US firms maintain a significant advantage in compute capacity and economically valuable AI applications. This advantage has resulted in accelerating revenue growth for leading cloud providers such as Alphabet and Amazon, whose infrastructure businesses recently reported revenue increases of 63% and 28%, respectively.
Why it matters
These results suggest the AI sector is not experiencing a speculative bubble but has instead reached a critical 'inflection point' where advanced AI adoption is proving commercially viable. The imbalance between AI compute demand and supply is expected to persist through late 2027, validating the massive infrastructure investments made by US companies so far.
Furthermore, this sustained demand for AI tokens—the core measure of AI usage—positions US hyperscalers to benefit from falling token costs. Goldman Sachs predicts this will lead to improved gross margins over the next year, amplifying revenue potential as token adoption expands in enterprise environments. This momentum is backed by an $800 billion revenue backlog held by the largest US cloud providers, offering strong visibility into future returns.
What to watch next
Market observers should monitor how US companies manage ongoing compute constraints and whether they can maintain their innovation lead amid emerging Chinese foundational models. While China holds an edge in energy supply critical for training AI models, US firms currently dominate in sophisticated agentic AI technologies and enterprise integration within Western markets.
Additionally, tracking the expansion of AI token consumption will be key, as Goldman Sachs forecasts a 24-fold increase by 2030 and 55-fold by 2040. This growth trajectory is expected to create robust new revenue streams independent of achieving artificial general intelligence milestones, reinforcing the sustainability of AI investments made today.