Etzioni on AI: Startup speed is not a strategy

Artificial intelligence researcher and Farecast founder Oren Etzioni published an analysis on November 15, 2026, challenging the technology industry obsession with development velocity, as detailed in GeekWire Startups's report. While software tasks that required an entire operating quarter in 2023 now take a single weekend in 2026, Etzioni argues that rapid execution alone fails to build defensible commercial operations because competitors and foundation models quickly erase simple feature leads.

This breakdown examines the vulnerabilities within common velocity arguments, the operational separation between rapid iteration and premature scaling, and the cultivation of proprietary data moats. It provides tactical guidance for software founders, venture investors, and engineering executives navigating resource allocation in an environment where frontier laboratories routinely absorb standalone application layers.

01Flaws across four popular velocity arguments

Advocates of development velocity often claim that market opportunities vanish quickly and rivals launch identical clones within weeks. Etzioni observes that while lower barrier costs allow competitors to duplicate software rapidly, frontier laboratories reliably close application gaps during scheduled release cycles. When software construction becomes inexpensive, velocity cannot serve as an enduring moat, reinforcing Peter Thiel's thesis in Zero to One that lasting defensibility generates superior long-term profits.

The standard premise of data flywheels similarly breaks down under rapid foundation model evolution. Customer interactions collected to fine-tune specialized models in 2024 lost their commercial relevance once frontier systems resolved those problems natively out of the box. Meaningful data advantages instead stem from restricted assets such as clinical health records, industrial sensor telemetry, and private transaction logs that frontier laboratories cannot readily crawl.

Similarly, relying on growth hype to secure excessive capital can trap young enterprises in restrictive corporate expectations. Securing capital at elevated valuations forces teams to meet inflated performance trajectories, eliminating the operational leeway necessary to change strategic direction. Startups that pace fundraising conservatively retain critical optionality to adapt whenever foundation models introduce fundamental architectural shifts.

02Cheap iteration against premature scaling risks

The sole valid argument for development velocity is accelerating organizational learning through low-cost iteration. Teams can test prototypes, launch pilot programs, or deploy experimental features to twenty users before quickly reversing course if adoption lags. Because these lightweight experiments carry minimal financial consequences, founders should execute them as rapidly as possible to uncover authentic product-market fit without locking down substantial enterprise resources.

In contrast, aggressive operational scaling introduces structural hazards that remain exceedingly difficult to unwind. Expanding headcount ahead of confirmed revenue, committing to multi-year infrastructure leases, and purchasing compute capacity represent irreversible one-way doors, matching the framework Jeff Bezos outlined in his 2015 Amazon shareholder letter. Empirical research supports caution, as a 2011 study examining more than 3,200 high-growth technology startups determined that premature scaling caused approximately 74 percent of venture failures.

03Founder judgment and defensible business models

Executive acumen and problem selection determine startup viability far more than raw coding throughput. At startup studio AI House, managing director Sri Chandrasekar explains that partners expect founding teams to identify meaningful commercial pain points through deliberate analysis rather than velocity. Because automated tooling has reduced basic software construction costs toward zero, teams should run experimental workflows manually before investing capital into permanent engineering infrastructure.

While velocity accelerates feedback loops, it cannot substitute for analytical discernment or sound strategic judgment. Running rapidly without directional clarity mirrors a hamster operating an exercise wheel at high speed without achieving forward momentum. Even Sam Altman's guidance notes that Y Combinator specifically seeks commercial enterprises that expand their market advantages at scale while remaining exceptionally difficult for emerging rivals to clone.

04Proprietary datasets as durable secret weapons

Sustainable competitive moats rely on structural assets that competitors cannot acquire simply by increasing engineering pace. Carnegie Mellon University professor and Nobel laureate Herbert Simon historically urged researchers to select problems where they possessed a secret weapon in unique knowledge or methodology. For modern technology businesses, that irreplaceable foundation translates directly into exclusive commercial partnerships and proprietary operational data accumulated over extended operating periods.

Etzioni demonstrated this principle at Farecast, the airfare forecasting venture he established in 2003. By 2007, the company mined more than 175 billion airfare records, scoring algorithmic predictions against realized price movements over time to assemble an insurmountable labeled dataset. Modern defensibility mirrors this paradigm when enterprise startups secure protected hospital network agreements, clinical records, industrial sensor feeds, or proprietary transaction logs from payment processing streams.

05Strategic patience and measured capital commitments

Navigating foundation model shifts requires founding teams to select specialized domains that frontier laboratories are unlikely to pursue directly. Authors like Laurence Peter long warned that lacking clear direction guarantees arriving at unintended destinations, while Good to Great author Jim Collins recommended firing small bullets to calibrate trajectory before launching definitive cannonballs. Operators must pursue difficult relationships and secure non-public data through patient trust rather than attempting to outpace well-capitalized platform providers.

Securing venture survival also demands raising funding at a disciplined cadence that preserves the latitude to pivot across generative model releases. Founders should formulate their core competitive differentiation into a single sentence and resist the instinct to rush headlong into broad market skirmishes. While contemporary market participants sprint blindly on momentum, the ultimate economic returns will belong to operators who balance lightweight tactical experimentation with methodical long-term strategic execution.