On September 10th, Amap unveiled its 2026 edition of the "Street-Sweeping List" alongside its annual awards ceremony, an event that signaled far more than just a ranking release.

The same day marked the debut of ABot-Earth 0.7, a 3D native urban world model, with the company leveraging "spatial intelligence" to redefine the product experience of its curated list. A year prior, the list relied on navigation and check-in data to gauge whether a venue was worth visiting; now, with features like Flight Street View 2.0, a "Pitfall Avoidance Guide," and Live Navigation, Amap is applying Agent capabilities to forecast the nuances of an entire trip.

This pivot represents Amap's ambition to reposition itself within the user decision-making framework in the AI era. By drawing on its extensive reservoir of location, routing, and real-time traffic data, the navigation giant aims to anticipate potential obstacles and better align with user needs against the backdrop of a dynamic physical world.

Yet, a critical question emerges: can spatial intelligence genuinely improve the predictability of consumer choices, helping Amap break free from its "search-and-leave" tool status and forge a sustainable business model?

Expanding Horizons in Space and Time

Amap's foray into AI through spatial intelligence is rooted in twenty years of accumulated data tied to cartography. The platform now boasts nearly one billion monthly active users, handles close to a trillion daily BeiDou satellite positioning calls, and has amassed tens of trillions of geospatial samples. These datasets span road networks, terrain, street views, and countless user behaviors like navigation sessions and route adjustments, all stemming from genuine movement rather than user-generated content.

Whether it's a road's congestion, the number of intentional visits to a shop, or the ebb and flow of foot traffic in a commercial district, each action continually refines Amap's understanding of the world. The "Street-Sweeping List" initially represented Amap's first concentrated foray into channeling this behavioral data into consumer decision-making. For the 2025 edition, the company calculated ratings based on actions such as purposeful trips, repeat visits, and local user preferences. The 2026 version elevates the significance of these intentional behaviors further, even integrating expert scores for niche categories like coffee shops.

In essence, a single online review pales in comparison to the time and travel costs borne by users who journey specifically to a location, making their choices a more reliable indicator of genuine preference. As Amap CEO Guo Ning aptly puts it: "Comments can be generated, but footsteps cannot be forged."

However, determining a shop's worthiness only addresses the initial phase of consumer choice. The journey from decision to arrival is fraught with variables—traffic, weather, queues, parking, and business hours. According to Amap's presentation, the real world comprises four facets: how roads connect, the spatial structure, temporal variations, and the flow of people and vehicles. Amap's vision is to consolidate this into a single system, allowing it to recreate the world's three-dimensional form, sense real-time changes, and predict future developments. "Expanding toward space" thus means maps go beyond providing coordinates and routes to illustrating the interior and essence of destinations.

In January, Amap launched the "Million Bustling Stores Support Plan," committing hundreds of millions in computing resources to offer free "Flight Street View" integrations for one million merchants. Version 2.0, announced at the event, extends this free access, adding up to 4,800 RMB in annual operation fee reductions, a free AI store manager, and consumption subsidies, broadening its support from three-dimensional displays to daily operations. According to sources from Wall Street News' All-Weather Tech, Flight Street View currently focuses mainly on scenic areas, with indoor merchant scenes reliant on self-captured uploads; coverage is expected to expand gradually through merchant authorization and external partnerships.

Simultaneously, "expanding toward time" enables the map to move beyond depicting current states to predicting complex variables throughout a user's journey. During the launch, Amap presented a five-day Guizhou travel itinerary generated by a large language model. While seemingly complete, when factors like holiday traffic, weather, scenic area capacity, parking, and companions' ages are introduced, the plan falls apart. Delays earlier in the trip can cascade, affecting subsequent bookings and routes.

The new "Pitfall Avoidance Guide" embeds this predictive logic into a tangible product, scrutinizing plans across 15 dimensions—including business hours, pacing, parking, and congestion—to preemptively flag closures and timing conflicts. Adding Agent-based planning capabilities also expands the informational scope Amap can process. With user input and authorization, the acceptance, adjustment, and completion of plans become feedback for future optimization, ultimately enhancing trip predictability. Amap's AI evolution, therefore, transitions from simply recording relationships between people and places to comprehending user intent and simulating real-world journeys. This is both a product extension and a renewed effort to shed its identity as a mere navigation tool.

An Unsettled New Direction

Rather than resolving Amap's long-standing questions, AI has amplified them: beyond massive map traffic, what business model and capability boundaries should Amap establish? Historically, it has experimented with advertising, ride-hailing aggregation, and local services to monetize traffic. The "Street-Sweeping List" pushes navigation upstream into consumer decisions, addressing not just "how to go" but also "where to go."

For consumers, AI advances this agenda by increasing Amap's touchpoints through a deeper understanding of needs and context. Yet, this only represents a portion of its AI strategy. Amap possesses a substantial user base and aims to differentiate itself with real behavioral data and a focus on distinctive, authentic local shops. However, unlike Meituan's formidable offline sales force, Amap's local services ecosystem still lags, creating constraints on merchant services and the overall experience loop. AI can help Amap understand demands earlier but cannot independently bridge these gaps.

To penetrate the local services market more deeply, Amap would need to leverage Alibaba's existing merchant, transaction, and fulfillment systems, intertwining its commercialization prospects with its position within the group. When Amap fully embraced AI in 2025, it found itself at the intersection of Alibaba's two core strategic trajectories. In August, Alibaba Group CEO Eddie Wu described Amap as "the world's first map-based AI-native application," suggesting it could become a significant new entry point for future life services through AI upgrades. Around that period, Alibaba's Tongyi Lab partnered deeply with Amap, using the Qwen model as a base to construct a model cluster for spatial perception and spatiotemporal intent understanding.

A shift occurred in November, however, when Alibaba launched the "Qwen" project and its namesake app, which quickly surpassed 100 million monthly active users within two months, culminating in the establishment of a dedicated Qwen consumer business unit in December. Qwen morphed from a model capability into an independent consumer portal. By 2026, Alibaba began reorganizing these two tracks: Qwen's consumer business, model lab, and QwenWork were integrated into an "AI Lab and Applications" unit, while the consumer side coalesced into a larger e-commerce system encompassing Taobao, Flash Purchase, and Fliggy. Amap found itself not wholly integrated into either camp.

These internal shifts force Amap to redefine its mission. Alibaba envisions numerous AI agents performing digital tasks in the future, with Qwen coordinating cross-platform resources through a unified interface. Within this structure, a unified entry point and vertical applications are not necessarily conflicting, leaving room for Amap to continue specializing in location-based scenarios. The newly launched "Pitfall Avoidance Guide" exemplifies this approach, offering more than just place information; it conducts spatiotemporal simulations and risk assessments for entire trips using real-time data, showcasing a more task-oriented Agent quality.

Concurrently, Amap is exploring direct productization of its spatiotemporal capabilities to unlock B2B revenue streams. In April 2026, it released "Tutu," a quadruped robot applying world models and embodied navigation across open environments. This was followed by the ABot-Earth series and ABot-World Studio, a general-purpose world model workshop. In July, Amap partnered with Hello, Shanghai HiSilicon, and OpenHarmony to deliver an AI-native navigation solution for shared two-wheelers. On September 10th, Mercedes-Benz became the first global automotive partner for the "Street-Sweeping List," which will now be integrated into its intelligent cockpit.

From robotics and world models to two-wheelers and automotive cockpits, Amap is attempting to package its maps, navigation, and spatiotemporal data as spatial intelligence services for automakers, hardware manufacturers, and industry clients. In the AI era, Amap's challenge has evolved beyond monetizing map traffic to fundamentally redefining the commercial value of its entire spatiotemporal capability stack.