The primary national security threat from TikTok is algorithmic manipulation by the Chinese government, a risk that persists even after a forced sale if the algorithm remains under Chinese control.
AI will devalue SaaS companies with seat-based pricing by reducing employee counts, shifting their valuation from growth-based revenue multiples to more modest EPS multiples.
TSMC's conservative capital expenditure, driven by the immense financial risk of underutilized fabs, is creating the conditions for a massive semiconductor shortage around 2029.
The most effective advertising model for AI assistants will be based on deep user profiling (like Meta) rather than immediate conversational context (like Google Search).
Space-based data centers are only economically conceivable due to terrestrial constraints (energy, land) and would require a complete redesign from first principles, a feat only SpaceX is likely capable of.
Stratechery Launch Period
Launched his paid subscription model, which quickly achieved success by reaching 1,000 subscribers and a $100,000 annual run rate within six months.
Hong Kong Protests Era
Identified the national security risk of TikTok's algorithm after observing the platform censoring searches for the Houston Rockets following a pro-protest tweet by the team's GM.
Post-TikTok Forced Sale
Assessed the forced sale of TikTok as a 'big disaster' because it failed to address what he considers the core issue: the Chinese government's control over the content algorithm.
Post-ChatGPT Launch (2 years)
Observed that despite the AI boom, TSMC decreased its capital expenditures, leading him to analyze the firm's conservative strategy and predict significant future chip shortages.
Current Discourse
Analysis has expanded to the physical limits of AI growth, speculating on the viability of space-based data centers and the reliability issues of cutting-edge hardware like NVIDIA's Blackwell GPUs.
▶AI's Economic TransformationMay 2026
Thompson analyzes how AI will restructure industries, predicting the decline of seat-based SaaS models and the rise of AI agents that could create 'perfect competition.' He argues for profile-based advertising models as the most effective way to monetize AI assistants, viewing OpenAI's ChatGPT as a primary asset in the competition for user attention.
Investors should re-evaluate SaaS company valuations based on their vulnerability to AI-driven headcount reduction and consider which companies are best positioned to leverage large-scale user profile data for future AI advertising.
▶The Geopolitics of Technology
He identifies critical national security risks in the tech supply chain and information ecosystem, focusing on TSMC's singular importance in semiconductors and the Chinese government's control over TikTok's algorithm. Thompson views these as significant points of failure and strategic vulnerability for the West, arguing the forced sale of TikTok was a failure because China retained algorithmic control.
Geopolitical risk analysis for tech investments must extend beyond simple market access to include algorithmic control and chokepoints in the physical supply chain, like advanced semiconductor manufacturing.
▶Platform Dynamics and StewardshipMay 2026
Thompson consistently differentiates between a company's ability to create products and its capacity to steward a platform for third-party developers. He critiques Apple as a poor steward despite its excellent hardware and praises Microsoft's stewardship while noting its product weaknesses, suggesting these traits define their strategic limitations and opportunities.
The long-term success of a tech ecosystem depends as much on its governance and developer relations (stewardship) as it does on the quality of its core products, a factor that can determine its durability against competitors.
▶Infrastructure as the Ultimate Bottleneck
Thompson argues that the primary constraints on technological progress are increasingly physical, not digital. He points to TSMC's conservative fab expansion, long lead times for electrical transformers, and the fundamental physics challenges of cooling data centers in space as the real-world limits on AI's growth.
Analysts should shift focus from software and model innovation alone to the physical enablers of that innovation; the valuation of AI and cloud companies is directly tied to the capital expenditure and supply chain realities of semiconductors, energy, and data center construction.