Citi's model predicts Meta will have negative free cash flow until 2028 due to aggressive AI infrastructure spending. The investment thesis relies on this capex driving user engagement (e.g., on Instagram), which will eventually translate to revenue, justifying the burn. Investors should watch engagement as a leading indicator of success.
The rapid scaling of Threads to 500 million MAUs demonstrates Meta has regained its ability to build and launch successful net-new products, not just copycat features. This success provides a blueprint and credibility for launching more ambitious products, including potential enterprise offerings like coding tools.
Based on current growth trajectories, an RBC analyst projects Microsoft's Azure is on a path to reach the same revenue scale as Amazon's AWS within several years. This would transform the cloud market from a clear leader-follower dynamic to a "1A/1B" scenario where both giants compete at a similar scale, though serving slightly different customer bases.
Microsoft mitigates the risk of over-investing in data centers by using third-party providers like Oracle and CoreWeave. If demand slows, Microsoft can reduce its reliance on these partners and bring workloads in-house first. This provides a crucial "escape valve" to avoid being stuck with costly, unused capacity.
An RBC analyst predicts an "80/20 world" for AI, where 80% of workloads can be handled by older, cheaper, or open-source models. However, the largest portion of the total addressable market (TAM) in terms of dollars will remain concentrated in the 20% of complex tasks that require cutting-edge frontier models.
Unlike the cloud transition which consolidated around a few hyperscalers, the agentic AI shift will be more fragmented. New agentic platforms will emerge from companies like ServiceNow and Salesforce, not just AWS and Microsoft. This creates a more diverse ecosystem for developers and enterprises, moving beyond a three-provider market.
To avoid chaotic spending, enterprises must replicate their "Cloud 2.0" governance models for AI. This means establishing a central platform engineering team to broker access to models, set budgets, and control the tools agents can use. This prevents runaway costs and security risks from decentralized AI development.