An Interview with Ben Thompson at the MoffettNathanson Media, Internet & Communications Conference
Ben Thompson discusses how compute shortages are reshaping Aggregation Theory and accelerating consumer AI adoption. The interview explores the tension between AI infrastructure constraints and the theoretical models that have governed internet platform dynamics.
Ben Thompson's discussion of compute shortages represents a critical inflection point for understanding how AI infrastructure challenges interact with established platform economics. The compute shortage directly challenges core assumptions of Aggregation Theory—the framework suggesting that winner-take-most dynamics emerge in digital markets through superior user experience and network effects. When computational resources become the bottleneck rather than distribution or data, traditional aggregators face unprecedented constraints on their ability to scale AI services.
Historically, Aggregation Theory explained how companies like Google, Facebook, and Amazon dominated by controlling user attention and data. AI commoditizes certain advantages these platforms previously held exclusive access to—models become increasingly available through APIs and open-source alternatives. However, the compute shortage inverts the traditional constraint: it's not data or users that limit AI deployment, but the physical infrastructure required to train and run large language models. This shifts competitive advantages toward entities with capital to secure GPU capacity and energy resources.
For investors and developers, this creates divergent outcomes. Companies with entrenched cloud infrastructure (AWS, Google Cloud, Azure) gain defensible moats through compute access, while traditional consumer platforms must compete for expensive resources. Startups face higher barriers to entry when training models requires millions in infrastructure costs. The shortage also accelerates consumer AI adoption paradoxically—urgency around capability differentiation pushes platforms to deploy imperfect but functional AI rather than waiting for optimal solutions.
Looking ahead, the resolution of compute constraints will determine whether AI remains concentrated among well-capitalized incumbents or becomes democratized. Supply-side solutions through new chip architectures and manufacturing capacity will reshape competitive dynamics across consumer technology and enterprise software.
- →Compute shortages are redefining competitive advantage away from traditional aggregation dynamics toward infrastructure control and capital access
- →AI infrastructure constraints create defensible moats for cloud providers while raising barriers for startups and smaller competitors
- →The scarcity of compute accelerates consumer AI deployment despite technical imperfections across platforms
- →Aggregation Theory's assumptions about network effects and winner-take-most dynamics require revision in resource-constrained AI markets
- →Resolution of compute supply will determine whether AI capabilities concentrate or decentralize across the technology industry