AI's Strategic Vacuum: Beyond Better Models, Seek Defensible Leverage

In a landscape rapidly reshaped by generative AI, a critical question emerges: where does strategic advantage lie when the traditional engines of tech dominance--network effects and proprietary technology--seem absent? This conversation with Benedict Evans and Tony Cam Brown delves into the unsettling reality for AI newcomers like OpenAI. It reveals that without the user-driven flywheel of network effects or a distinct technological moat, companies risk becoming mere infrastructure providers, vulnerable to incumbents with established distribution and user bases. The core implication is stark: the path to sustained dominance in AI is not through simply building a better model, but through a yet-to-be-defined strategy that can create genuine, defensible leverage in a rapidly commoditizing field. This analysis is crucial for founders, product leaders, and investors seeking to navigate the uncharted territory of AI's competitive dynamics and identify where true, lasting value will accrue.

The Ghost of Network Effects: Why "Better" Isn't Enough

The bedrock of consumer tech for decades has been the network effect: the phenomenon where a product or service becomes more valuable as more people use it. Think of Google Search, whose vast user base feeds it data, making its search results superior, thereby attracting even more users. Or iOS and Android, where developer ecosystems flourish because users are locked into those platforms. This creates an almost insurmountable moat, a virtuous cycle where the leader gets stronger, and challengers struggle to gain traction, regardless of their technical merit. Microsoft’s long, expensive battle to make Bing competitive with Google Search is a prime example of this dynamic at play.

Generative AI, however, appears to defy this historical pattern. As Benedict Evans points out, "from ground zero, from day one, you can't see network effects." The models themselves, while incredibly complex and expensive to build, do not inherently improve with more users in the way a social network or search engine does. This absence of a self-reinforcing user loop fundamentally alters the competitive landscape.

"The point here being that all consumer tech since the 80s, certainly, consumer computing has been based on network effects... Their dominance is based on network effects, and Google Search is based on network effects."

This lack of inherent network effects forces a re-evaluation of where sustainable advantage can be found. Evans suggests that while scale effects in model training are undeniable--requiring billions in investment--this doesn't automatically translate to leverage further up the stack. Companies like Intel or TSMC, controlling crucial low-level components, don't dictate the applications built on top. Similarly, AI model providers might become akin to the hyperscalers (AWS, Azure, Google Cloud), offering commodity infrastructure at competitive prices, rather than dictating the terms of engagement. The question then becomes: how does a company like OpenAI, lacking existing distribution channels or deep user engagement, break into the exclusive club of vertically integrated giants like Apple or Microsoft?

The Strategy Vacuum: Beyond "Being Better"

The historical playbook for tech dominance involved either leveraging network effects or building a defensible platform. Apple’s iOS, for instance, benefits from both: a massive user base and a vast app ecosystem that developers must support. Google Search, as mentioned, thrives on its data-driven network effect. But what happens when neither is present?

Evans argues that the current state of generative AI presents a strategic vacuum. Without a clear path to a unique, defensible advantage, companies are left with the challenging task of simply executing better than everyone else. This is a precarious position, especially when competitors include tech behemoths like Google, Meta, and Microsoft, who possess immense resources, existing distribution channels, and a history of strategic pivots.

"And so the issue here is, you know, you can hire, you have a whole bunch of really clever, really aggressive, really driven people, and you have to execute. But then you're doing this thing, you've got this strategy, and you're following the strategy, and that is what's delivering fundamentally the defensibility and the sustainable competitive advantage of your company. It's the strategy, it's not hire lots of clever people."

The conversation highlights how established companies have clear "flywheels"--self-reinforcing cycles of growth. Amazon's flywheel, for example, links more customers to lower prices and better service, leading to more volume. OpenAI's published "flywheel" of "more capex, more infrastructure, more revenue" is critiqued as merely a statement of investment, not a virtuous circle that compounds advantage. This lack of a clear, compounding strategy leaves AI newcomers vulnerable.

The Product Head as a Strategy Taker, Not Setter

A particularly striking insight emerges from the discussion of product leadership within AI labs. The traditional product development model, where a product lead sets a strategic vision and roadmap, seems inverted. As Evans notes, quoting Figgy Simo, the head of product at OpenAI or Anthropic might receive an email from researchers saying, "Hey, we've got this cool thing. What can you do with it?"

This dynamic, contrasted with Steve Jobs' famous dictum to "start with the user experience and work back to the technology," reveals a fundamental challenge. When the technology itself is rapidly evolving and unpredictable, and the research agenda dictates the product possibilities, the product team becomes a "strategy taker, not a strategy setter." They are reacting to emergent capabilities rather than proactively shaping a user-centric vision.

"So you don't really know or control your roadmap. Yeah. Now, I've paired this quote from Figgy with the classic quote from Steve Jobs from 1997 or so when he went back to Apple, and he said, 'You can't start with the technology and work to the user experience. You've got to start with the user experience and work back to the technology.'"

This lack of roadmap control, coupled with the commoditization of core AI models, means that differentiation is incredibly difficult. The Netscape analogy is invoked: when the underlying technology and product become largely indistinguishable, competition devolves into brand, marketing, and distribution--areas where incumbents have a significant advantage. This is where companies like Anthropic, with its "lifestyle product" branding and Super Bowl ads, are attempting to carve out space, even if their actual user engagement metrics remain low compared to established players. The question remains whether this brand awareness will translate into sticky user adoption or merely serve as a temporary differentiator in a market where the core technology is rapidly becoming a commodity.

The Unsettling Future: Commoditization and Distribution Wars

The overarching theme is the potential commoditization of foundational AI models. If the technology becomes undifferentiated, and the product built on it is also largely the same, then market share will likely be dictated by factors outside the core AI capability itself. This is where the established players--Google, Meta, Microsoft, Apple, Amazon--hold a significant advantage. They can integrate AI capabilities into their vast existing product suites and leverage their enormous distribution networks to expose these new features to billions of users.

For OpenAI, this presents a stark challenge. They are, in essence, starting from a "completely blank slate" without the cash flows or established user bases of their competitors. While they possess a large user base for ChatGPT, its engagement is described as "shallow" and lacking "stickiness" or network effects. This leaves them competing with the entire tech industry, including thousands of entrepreneurs, to invent what comes next.

The conversation concludes by circling back to the core questions: What does "platform" mean in this new AI era? What constitutes "power," "leverage," or "lock-in" when the underlying technology is rapidly evolving and becoming accessible to many? The current uncertainty suggests that the companies that will ultimately thrive may not be those with the most advanced models today, but those who can successfully navigate this evolving landscape by building genuine, defensible strategic advantages--advantages that are not yet clearly apparent. The immediate future, it seems, will be defined by a battle for distribution and brand relevance, rather than solely by technological superiority.

Key Action Items:

• For AI Model Providers (e.g., OpenAI, Anthropic):

  • Immediate: Develop a clear, actionable strategy for differentiation beyond model performance. Focus on unique product integrations, developer ecosystems, or novel user experiences.
  • Immediate: Aggressively pursue distribution channels, leveraging partnerships or integrating into existing platforms where possible.
  • Next 6-12 Months: Invest in building genuine user engagement and "stickiness" that moves beyond shallow usage, exploring how user actions can create value for other users or the platform itself.
  • 12-18 Months: Explore potential for platform lock-in through developer tools, APIs, or specialized applications that are difficult to replicate.
  • Longer Term: Identify and cultivate unique, defensible strategic advantages that competitors cannot easily replicate, moving beyond simply being "better."

• For Developers Building on AI Infrastructure:

  • Immediate: Focus on building applications with strong user experience and clear value propositions that go beyond basic AI functionality.
  • Immediate: Prioritize brand and distribution strategies, as these will likely be key differentiators when underlying AI tech is commoditized.
  • Next 6-12 Months: Consider how your application can create its own form of network effect or user-generated value, even if the underlying model doesn't offer it.
  • 12-18 Months: Evaluate the long-term viability of your chosen AI model provider, considering their strategic positioning and potential for future differentiation.

• For Investors:

  • Immediate: Scrutinize business models for clear, defensible competitive advantages beyond technological parity.
  • Next 6-12 Months: Favor companies with strong distribution strategies or unique product-market fits that can leverage AI capabilities effectively.
  • 12-18 Months: Assess the long-term strategic positioning of AI companies, looking for evidence of compounding advantages rather than just current performance.