200M Viewers Put Prime Video AI on Bezos’s Hot Seat

On Thursday (July 23), Prime Video AI stopped looking like a feature roadmap and started looking like Amazon’s next public audition in artificial intelligence.

Jeff Bezos has urged Prime Video chief Mike Hopkins to overhaul the streaming service so AI takes a starring role, according to PYMNTS, citing Reuters and four sources with direct knowledge of the matter. The internal project is called Lighthouse, and its audience could be massive: more than 200 million people use Prime Video.

That timing matters because Amazon has spent hundreds of billions of dollars developing AI, yet its most visible consumer AI story remains uneven. Reuters sources described Alexa’s long-running upgrade as producing mixed results, with that division still losing money. Prime Video gives Amazon a cleaner stage. People already open it, browse it, search it, abandon it, and come back.

XOOMAR analysis: Lighthouse signals that Amazon wants Prime Video to do more than retain viewers. It wants the service to make Amazon’s AI feel tangible.

Last fall’s Bezos meeting reportedly turned Prime Video AI into Project Lighthouse

The Prime Video AI push traces back to a presentation last fall that reportedly became “contentious.” According to Reuters sources cited by PYMNTS, Bezos was unhappy that the proposed Prime Video update did not do enough to spotlight AI/personalization capabilities.

That reaction reportedly killed the prior plan and produced Lighthouse.

The project is being led by Kam Keshmiri, global head of Prime Video design, according to the related Reuters-based report carried by The Economic Times. Bezos has also received occasional updates, which is notable because he stepped back from Amazon’s CEO role in 2021.

Amazon has not offered a public explanation of the project. PYMNTS said it contacted Amazon for comment and had not received a reply. The Economic Times report said Amazon declined to comment.

The causation is the story. A streaming redesign became an AI mandate after Bezos decided the earlier plan did not show enough ambition. That makes Lighthouse less like a normal interface refresh and more like a reputational project.

More than 200 million users make Prime Video Amazon’s broadest AI showroom

The scale explains the choice. Prime Video reaches more than 200 million users, and the service is one of Amazon’s best-known brands. The Economic Times report also noted that Prime Video is available in markets where Amazon has limited or no e-commerce presence.

That gives Amazon something Alexa has struggled to deliver recently: a high-frequency consumer surface where AI can be judged by usefulness, not demos, a key front in the AI habit war.

Reported Lighthouse ideas include:

• AI recommendations: Better film and TV suggestions based partly on viewer preferences.
• Voice requests: Responses to spoken requests, potentially tied to Alexa.
• Homepage redesign: A new main page being tested with some users.
• AI-driven tiles: Suggested categories such as “action movies from the 1980s” or “Christmas rom-coms,” though one source said a current version does not include text-heavy tiles.
• Traditional search: The existing search function would remain.
• Sports and release promotion: Space at the top would still promote new releases or events such as Thursday Night Football.

The strongest version of Prime Video AI is boring in the right way. It helps viewers find something faster. It reduces clutter. It makes spoken search less brittle. It turns personalization into a practical tool instead of a label slapped onto the homepage.

The weak version is corporate AI theater. If viewers feel they are being pushed through AI-branded prompts that serve Amazon’s narrative more than their viewing intent, Lighthouse will make Prime Video noisier, not smarter.

For readers tracking how streaming services compete for attention week by week, XOOMAR’s coverage of 5 New Netflix Movies and Shows Rescue a Thin July Week shows the same consumer problem from the content side: discovery matters most when the catalog feels thin or overwhelming.

The numbers show why Amazon wants the homepage, not just the model

Prime Video has reach, but it does not dominate U.S. TV viewing. April Nielsen data cited by The Economic Times puts Prime Video behind YouTube, Netflix, and Disney+.

That gap sharpens the Lighthouse logic. Better personalization can lead to more hours spent on the service, according to the Reuters-based report. More time spent also makes the Prime Video homepage more valuable real estate.

But that same real estate is contested.

“The real estate on the home screen is very valuable to studios, so it would be a big change to take away any of that coveted space,” said Michael Goodman, director of entertainment research for Parks Associates.

That quote gets to the hardest product tradeoff. Prime Video, like other streaming services, uses paid studio placement and software algorithms to decide where content appears. If Lighthouse gives more space to AI-personalized rows or tiles, something else may lose visibility.

XOOMAR analysis: Amazon is not just redesigning navigation. It is potentially changing the economics of attention inside Prime Video.

Subscribers, studios, and Amazon won’t want the same Lighthouse

The subscriber pitch is easy to understand. Better recommendations, cleaner search, and useful voice commands can make Prime Video less frustrating. Viewers do not need to care whether the feature is powered by AI. They need it to work.

Studios may see the same redesign differently. If AI personalization changes where shows and films appear, paid placement becomes less predictable. Goodman’s warning suggests a real tension between viewer-specific discovery and the homepage positions studios value.

Amazon’s incentives are broader. Reuters sources described Lighthouse as part of Amazon’s effort to improve its standing in AI while competitors including OpenAI and Anthropic push ahead. The company has spent heavily on AI, and Prime Video is a consumer-facing way to show progress.

That does not mean every AI idea belongs in the streaming interface.

A useful test is simple:

• Viewer benefit first: Does the feature help someone pick, search, or resume content faster?
• Promotion second: Is Amazon using the screen to prove AI ambition rather than solve a viewing problem?
• Partner impact: Does personalization weaken paid placement enough to create new friction with studios?
• Feedback risk: Do early testers accept the redesign, or does Amazon retreat from the more visible AI elements?

The related Alexa discussion matters here. Amazon has discussed integrating the Alexa voice assistant into Prime Video search as part of Lighthouse, according to the Reuters-based report. Amazon released an overhauled generative AI version of Alexa in early 2025 and integrated it into its main shopping site in May 2026, according to that same report.

XOOMAR has also covered the broader challenge of making AI useful outside slide decks, including in Physical AI Delivery Robots Hit the Costly Last 50 Feet. The lesson travels: visible AI only earns trust when it fixes a real bottleneck.

The next decision is whether Lighthouse follows tester feedback or Bezos’s AI mandate

Amazon is already testing versions of the Prime Video redesign with a few users, according to one person cited in the Reuters-based report. The final design has not been settled and could change because of tester feedback, financial concerns, or other issues.

That is the decision point that matters now.

If early feedback favors practical features, Prime Video AI may emerge as a quieter but stronger product: better discovery, better spoken requests, and less friction on the homepage. If the redesign over-indexes on making AI visible, the service risks turning a streaming session into an Amazon strategy presentation.

The forward read is clear. Lighthouse will matter if it makes Prime Video easier to use. Evidence that would confirm that thesis: Amazon keeps traditional search, improves recommendations, limits text-heavy AI tiles, and avoids crowding out the core reason people opened the app.

Evidence that would weaken it: more branding than utility, more homepage clutter, and studio placement fights that force Amazon to dilute personalization before most viewers ever see the product.

The Bottom Line

• Prime Video gives Amazon a high-traffic venue to showcase AI to more than 200 million users.
• Project Lighthouse suggests Amazon wants AI to improve everyday entertainment discovery, not just backend technology.
• Bezos’ reported involvement signals strategic urgency around making Amazon’s consumer AI efforts more visible.