Keith Rabois: OpenAI is winning the AI battle, Opendoor needs a new CEO and AI-native transformation

Summary

Keith Rabois is unambiguous on the AI competitive landscape: OpenAI is winning, and Google is losing. ChatGPT is the fastest-growing consumer app of all time, and normal users are substituting searches with prompts at a pace that directly threatens Google's core business. Rabois argues that even with world-class AI research talent, Google has failed to ship a product that reverses this trend. His proposed counter — fusing Gmail, YouTube, and other personal data into a deeply personalized product — remains unshipped and unproven.

The deeper structural threat to Google, in Rabois's view, is the business model itself. Performance-based advertising, the engine behind Google's rise, may not be how the next generation of consumers expects to be monetized. AI applications are demonstrating that users will pay directly for value, making subscription models a credible replacement. A rough per-user revenue comparison, he argues, suggests consumers are already willing to pay more through ChatGPT than Google currently extracts through ads.

Opendoor: A Leadership Problem, Not Just a Rate Problem

Rabois frames Opendoor's struggles as only partially explained by the macro environment. In a hot market, roughly 5 to 6 million U.S. homes transact annually. The Fed's rapid rate hikes compressed that to approximately 4 million, a meaningful reduction but not an existential one. The real problem is a fixed cost base — specifically G&A — that is too high to sustain profitability at lower transaction volumes. That structural flaw was flagged to management as far back as 2015 by an executive coach, and by Rabois's account, the company has made no meaningful progress cutting those costs in the three years since rates rose.

He sees AI as the direct solution. Inspection, repair assessment, and most back-office functions are candidates for agentic replacement. He floats the possibility of running Opendoor with as few as 100 people or fewer, converting the fixed cost base to variable, and returning the company to profitability on home-by-home underwriting decisions alone. Rabois notes that with one exception in company history, Opendoor has priced and purchased homes profitably — the pricing model works, the cost structure does not.

On valuation, he draws an explicit comparison to Carvana, a $40 billion company operating in a more competitive market. The largest residential real estate platform in the Western world is currently worth approximately $18 billion, a figure Rabois calls absurd given that residential real estate is the largest asset class globally. He puts the credible target range at $50 to $150 billion, contingent on the right leadership.

Retail investor pressure, led by Eric Jackson, has refocused public attention on Opendoor's potential after three years of narrative neglect. Rabois is supportive of that dynamic, framing retail investor activism as legitimate capital allocation in action. He explicitly calls for a new CEO — someone AI-native, founder-minded, and willing to make hard unilateral calls about what the company stops doing. He is not a candidate himself but says he will help identify, assess, and recruit the right person. His public benchmark for the role: operators in the mold of Will Gaybrick at Stripe or Greg Brockman at OpenAI.

His criticism of the current strategic direction is direct. The company's recent moves toward partnering with legacy real estate brokers are, in his view, the wrong direction entirely. The value proposition Opendoor offers buyers and sellers should be so superior that the comparison to traditional agents becomes irrelevant.

Macro: AI as a Deflationary Growth Engine

Rabois pushes back on the assumption that growth necessarily triggers inflation, arguing it reflects outdated post-2010 thinking shaped by quantitative easing rather than historical norms. From 1950 to 2010, he notes, the U.S. averaged 3.6 decades of years with over 4% GDP growth without triggering inflation. AI-driven productivity gains, he argues, can deliver consistent 3 to 5% growth without wage-driven inflation because the growth engine is not labor cost increases.

He is aligned with Treasury Secretary Scott Bessent's three-three-three framework and believes the U.S. can exceed 3% consistent GDP growth within the next few years. The Federal Reserve, in his view, remains stuck in a model that reflexively brakes on growth signals — a mistake he attributes to intellectual frameworks that predate the current technology cycle. He calls for replacing Jerome Powell as Fed chair, expressing confidence that change happens by September, and explicitly ties lower interest rates to Opendoor's transaction volume recovery alongside a CEO transition.