VC: The $575B AI Bet (Biggest Since the World Wars) and the New Two-Buyer Reality in Software | Tomasz Tunguz, Theory Ventures

The AI boom has become the fifth‑largest infrastructure effort in U.S. history, consuming roughly 5% of GDP and representing a $575 billion wager by hyperscalers. For every dollar earned from AI services, companies are investing twelve dollars in data‑center capacity, power, and networking. This unprecedented scale, projected to reach 6% of GDP by 2030, mirrors the massive mobilizations of the world wars and underscores the infinite demand for inference workloads across text, image, and video models.

In the near term, the battle is over market share: Google, OpenAI, Anthropic, and emerging rivals are racing to dominate the AI platform landscape. The longer game shifts to margins, where profitability hinges on delivering more intelligence per watt of electricity. Hardware bottlenecks—energy, GPUs, memory, and high‑speed interconnects—are driving a cascade of investment into traditional “picks‑and‑shovels” sectors, from semiconductor fabs to heavy‑equipment manufacturers building data‑center infrastructure. This ripple effect reshapes the broader economy, creating new opportunities for companies that sit at the intersection of AI software and specialized hardware.

For founders and investors, the fusion of data and AI stacks demands a rethink of organization and product strategy. Startups must design AI‑native structures, reducing headcount through intelligent automation while building trust‑based, multi‑agent platforms that can conduct B2B discovery and close deals without human intervention. Continuous product‑market fit replaces static milestones, as foundation‑model companies have mere weeks to commercialize breakthroughs before competitors catch up. Venture capital is flowing toward hybrid hardware‑software ventures that can serve as trusted partners across a three‑to‑five‑year horizon, positioning them to capture both share and margin in the evolving AI value chain.