The first-session gap has defined the ceiling for cloud gaming since the category emerged: to reach a new player, a service must clear the hardware prerequisite and the subscription decision before a single session occurs. Amazon and Microsoft are now attacking that constraint from different directions, with Luna moving into Prime Video's established subscriber surface and Microsoft testing an ad-supported streaming option for titles its users already own. The two moves point at the same distribution problem while targeting different user populations.
Luna inside Prime Video: inheriting the subscriber layer
The mechanism behind Amazon's move is stack compression. Prime Video already operates inside an established subscriber identity and billing infrastructure. Folding Luna into that surface means cloud gaming can inherit the authentication and discovery layer that Prime members already use, rather than requiring a separate app install or a separate subscription commitment. The constraint this addresses is top-of-funnel acquisition: players who pay for Prime but have never opened a dedicated cloud gaming application.
Where this sits in the stack is the content discovery surface, the same screen users open to watch video, which becomes the screen where a gaming session can now begin. That surface conversion is the structural logic of the integration. The user already paid; the session just needs to start.
Microsoft's ad-supported test: the owned-title angle
Microsoft's model targets a different point in the same funnel. Players who have already purchased titles represent a population that has cleared the intent barrier but not the cloud-access barrier. An ad-supported streaming option for those owned titles sets the incremental subscription cost to zero, replacing it with advertising revenue on a per-session basis.
The specific unit that drives the economics here is cost-per-session: whether ad yield can cover the compute and bandwidth expenses generated by each player. Microsoft has framed this as a test, not a committed rollout, meaning the model's commercial viability is still being evaluated against actual session data.
Two funnels, one distribution problem
The divergence reflects a category-wide reorientation. Cloud gaming's infrastructure costs and latency profile are no longer the primary barriers being addressed at the platform level. Both moves acknowledge, by their structure, that the binding constraint is now distribution: getting a session to start at all. The two companies have chosen different user populations to prove that out. Microsoft's answer is in test. Amazon's routes through a video streaming service its subscribers already open.