According to a new projection from Goldman Sachs, spending focused on artificial intelligence is approaching levels last seen during the dot com boom. The bank predicts leading cloud and infrastructure providers could allocate nearly 98 percent of their operational cash flow to capital expenditures in 2026, marking a historic investment shift.
Infrastructure investment nears record heights
The analysis points to rapidly expanding budgets for data center capacity, computing infrastructure, networking equipment, and specialized AI hardware. Goldman Sachs emphasizes that the current trend is aligning closely with the technology, media, and telecommunications spending peak of the early internet era.
Mini glossary: Hyperscalers are technology companies that operate massive-scale cloud and data processing infrastructure. Capital expenditure refers to investments in long-term assets like data centers, servers, chips, and network equipment.
The financial blog Global Markets Investor also notes the remarkable shift, stating that major tech companies may soon route nearly all generated cash into new infrastructure buildouts. This increasing focus on investment has caught the attention of market watchers worldwide.
Goldman Sachs data shows that by 2026, hyperscaler companies are on track to allocate nearly 98 percent of their operating cash flow to capital expenditures, a rate that comes close to the highs reached in the dot com era.
An accompanying chart in the Goldman Sachs report tracks the ratio of capital expenditures to operating cash flow over time. Historical data reveals that telecom firms in the early 2000s’ infrastructure race exceeded 120 percent. In the broader technology, media, and telecom sector, the peak was around 95 percent.
| Period or Group | Capital Expenditure Ratio |
|---|---|
| Hyperscalers 2015 to 2018 | 30 to 40 percent |
| Hyperscalers 2023 | 55 percent |
| Goldman Sachs 2025 Projection | 68 percent |
| Goldman Sachs 2026 Projection | 98 percent |
Historically, hyperscaler firms spent at much lower levels. Between 2015 and 2018, they invested roughly 30 to 40 percent of operational cash. However, rising demand for cloud services and the accelerating race for AI development pushed this ratio to 55 percent in 2023. Goldman Sachs now predicts 68 percent in 2025 and an eye-popping 98 percent in 2026.
The debate over returns intensifies
The current investment cycle is largely fueled by soaring demand for AI computing power. Organizations are rapidly expanding data centers and continuing to purchase high volumes of GPUs and networking hardware to support advanced models.
Goldman Sachs estimates that total technology industry capital investments could approach 920 billion dollars by 2027, and in a more aggressive scenario, could climb as high as 1.4 trillion dollars. This would represent an 89 percent surge over the 2026 average projections, highlighting the unprecedented scale of spending.
As infrastructure spending continues to grow, companies are watching more closely to see whether AI-related investments are matched by corresponding revenue increases.
At the same time, some companies utilizing AI tools are questioning whether these heavy expenses are offset by adequate financial returns. As infrastructure costs rise, a key question emerges: will revenue growth keep pace? Price competition among model developers is also intensifying debate over long-term profitability in the sector.
Current data shows that hyperscaler investments are well above historical averages and closing in on that 100 percent threshold. This trend suggests that nearly all operational cash could end up being poured into growth initiatives, rather than shareholder payouts or other corporate uses, reflecting the aggressive tempo of technology reinvestment.
