Renowned tech investor Gavin Baker has revealed that public financial reports from major GPU purchasers demonstrate AI is undeniably profitable.
In a podcast interview on December 9, Baker pointed out that the largest GPU investors are publicly traded companies. By analyzing their financial statements, he found that their Return on Invested Capital (ROIC) actually improved after massive GPU investments. Baker remarked:
*"I’ve always found it puzzling why there’s still any debate about this."*
The ROIC boost stems partly from operational cost savings, but more importantly, migrating large-scale recommendation systems from CPUs to GPUs has significantly enhanced efficiency, accelerating revenue growth. Baker emphasized that regardless of the source, the key takeaway is that the return on investment (ROI) is undeniably positive. He also observed that within every major internet company, revenue-generating departments are fiercely competing for GPU resources—because it’s a straightforward equation: more GPUs mean more revenue.
**Companies Are Cashing in on AI Dividends**
Baker highlighted Q3 2024 as a critical turning point, marking the first time non-tech Fortune 500 companies provided concrete, quantifiable examples of AI-driven performance improvements. For instance, freight brokerage firm C.H. Robinson saw its stock surge roughly 20% post-earnings due to AI-powered productivity gains.
One of the company’s core operations involves generating freight quotes. Previously, processing a single quote took 15 to 45 minutes, with only 60% of inbound requests fulfilled. After implementing AI, the company now responds to 100% of quote requests within seconds. This AI-driven efficiency uplift impacted both revenue and costs, leading to an earnings beat and a stock rally. Baker noted:
*"This is a clear case where AI-driven productivity improvements affect revenue, costs, and every metric in between."*
He stressed that this example is particularly significant because it alleviates market concerns about the "Blackwell ROI gap." Blackwell, Nvidia’s next-gen chip, requires massive capital expenditure. Since these chips are initially used more for model training than inference, there was a perceived risk period where high spending wouldn’t immediately translate to revenue, potentially depressing ROIC.
**Startups Show Dramatic Efficiency Gains**
Venture capitalists are more bullish on AI than public market investors, partly because they directly witness real productivity gains. Baker cited data showing that companies reaching certain revenue milestones today employ far fewer people than those at the same stage two years ago. The reason is simple: AI now handles substantial sales, customer support, and even product development tasks.
These efficiency improvements are evident in financial data. Firms like Accel and Andreessen Horowitz have published findings confirming this trend. Investors with access to startups see firsthand how AI creates value within businesses.
Young, AI-native founders stand out even more. Today’s 23- and 24-year-old entrepreneurs display the maturity level of founders in their early 30s from past generations. They leverage AI adeptly from day one—whether pitching to investors, resolving HR challenges, or refining sales strategies. Baker stated:
*"AI is already highly capable of handling these problems today."*
**SaaS Companies Repeating Retail’s Mistakes**
Baker expressed concern that traditional SaaS firms are failing to embrace AI, mirroring the missteps of brick-and-mortar retailers facing e-commerce. Retailers once dismissed e-commerce due to its lower gross margins, reasoning that in-store shopping was inherently more efficient than home delivery.
Yet, reality proved that with dense logistics networks, efficiency and margins improve. Today, Amazon’s North American retail margins exceed those of many mass-market retailers. Baker warned:
*"SaaS companies are making the same mistake. They enjoy 70% to 90% gross margins but resist AI’s ~40% margins."*
Unlike traditional software, where code is written once and distributed cheaply, AI inherently consumes compute power per transaction. However, Baker noted that AI firms, despite lower gross margins, generate cash flow earlier due to minimal staffing needs.
He called this a "life-or-death decision," where nearly every company except Microsoft has faltered. He advised SaaS firms to follow Adobe and Microsoft’s cloud transition playbook—accepting short-term margin pressure as long as gross profit grows in absolute terms.
Companies like Salesforce, ServiceNow, HubSpot, GitLab, and Atlassian possess cash-rich core businesses—an advantage AI-native startups lack—and should leverage this to compete in AI.
Comments