The AI Capital Supercycle Is Real — And It's Just Getting Started
Global AI infrastructure spending is on track to exceed $3 trillion by 2028. With OpenAI valued at $730 billion and Big Tech deploying $650B this year alone, the capital supercycle is accelerating — and the winners won't just be the model makers.
The AI Capital Supercycle Is Real — And It's Just Getting Started
The numbers have stopped making intuitive sense — which is usually a sign that something genuinely structural is happening.
In February 2026 alone, global startup funding hit $189 billion. That is the highest monthly total in the history of venture capital. AI captured 90% of it: $171 billion. Three deals — OpenAI ($110B), Anthropic ($30B), and Waymo ($16B) — accounted for 83% of the entire month's global VC activity.
OpenAI is now valued at $730 billion, backstopped by Amazon ($50B), NVIDIA ($30B), and SoftBank. For context, that puts it in the same market-cap neighborhood as Berkshire Hathaway and JP Morgan — except OpenAI has no earnings, no publicly traded shares, and a product that didn't exist as a commercial proposition four years ago.
Meanwhile, Alphabet, Amazon, Meta, and Microsoft are forecasted to collectively spend between $635 and $665 billion on AI infrastructure in their 2026 fiscal years — up sharply from $410 billion in 2025. Bridgewater has noted this capex surge alone adds approximately 50 basis points to US GDP growth.
This is not hype. This is capital allocation.
Why This Cycle Is Different From 2021
The 2021 tech bubble was a liquidity-driven anomaly: cheap money inflated valuations of businesses with questionable unit economics. The AI supercycle has a different engine.
The capex being deployed is not being borrowed to fund losses — it is being invested in infrastructure that is already generating commercial returns, with pricing power that compounds. Microsoft's Azure AI services, Amazon's Bedrock, and Google's Gemini API products are all growing revenues faster than the infrastructure can scale to meet demand.
The fundamental dynamic is this: compute is currently the binding constraint on AI capability, and AI capability is directly translatable into commercial value. That means every dollar of capex is chasing a demonstrably expanding market, not a speculative one.
NVIDIA remains the clearest expression of this. But the strategic landscape is shifting. Jensen Huang's announcement that NVIDIA's $30B OpenAI stake may be its last major model-company investment — and the pivot toward a $26B commitment to open-source AI — signals that the next phase of the supercycle is about infrastructure commoditization, not model concentration.
That matters for how you invest in it.
This is where the analysis gets actionable. AlphaBriefing members get the full investment framework — scenarios, positioning, and the bottom line.
Subscribe to AlphaBriefing — Free, Member, and Paid tiers available.