Startup Funding Up 64%, Hiring Down 10%

The venture capital landscape is increasingly favoring "ultra-lean" startups, which prioritize automation and efficiency over large teams. This has led to a significant increase in funding per employee; for instance, US-based tech startups in 2025 raised over $320,000 per employee in Series A rounds, a jump from about $160,000 in 2020. The median number of employees in these startups has also seen a decline, dropping from 57 in 2020 to 44 in 2024. This shift is largely driven by the adoption of AI, which enables companies to achieve higher productivity with smaller workforces. Startups heavily invested in AI show a 93% positive outlook on their financial prospects, compared to 71% for those not adopting AI. Consequently, more than half of startups are reallocating their budgets from traditional tools to AI technologies. The focus for both founders and venture capitalists has moved from rapid, aggressive growth to a more measured approach centered on capital efficiency and a clearer path to profitability. This disciplined approach to hiring is seen as a strategic advantage, with an emphasis on resilience and return on investment for every new team member. The growing reliance on AI has put a premium on computational power, with investors beginning to view GPU infrastructure as a long-term asset class. The global data center GPU market is projected to grow from $26.3 billion in 2026 to $178.1 billion by 2033. This demand is causing a structural shift in data center investment, prioritizing compute density and utilization over physical space. This intense demand and price volatility for GPUs have led to the concept of "compute as a commodity," similar to oil or electricity. The argument for creating a futures market is based on the high capital costs of building GPU clusters, supply and demand shocks, and the need for standardized units of compute power. Several firms are already working to create financial instruments for the GPU market. Silicon Data has launched a daily GPU rental price index on Bloomberg terminals, and companies like OneChronos are developing auction markets for GPU compute. These platforms aim to bring transparency and risk management tools to what is being called the "world's largest unhedged corporate asset class."