《WIRED》的 Steven Levy 在 2026 年 2 月 13 日报导 Zillow，指出美国房市仍然疲弱。CEO Jeremy Wacksman 说市场正「在底部弹跳」，去年仅售出 4.1 million 间二手房，低于「正常」的 5.5 to 6 million，并预期 2026 年只会小幅改善；他也提到 Zillow 的表现优于整体产业，但估值仍仅约为 2021 年高峰的 1/4。Zillow 上季公布获利较高后，股价隔天仍下跌近 5 percent；Wacksman 将 AI 描述为「一种配料而非威胁」，既能保护 Zillow 的地位，也能改变人们的搜寻方式（例如：每月低于 $3,000、并带有特定生活型态限制的租屋查询）。

Wacksman 主张 Zillow 长期依赖 AI，并以 Zestimates 作为以机器学习为核心的功能例证；他表示近期的 LLM 浪潮正加速 AI 在公司各处的整合。Zillow 正推动 AI 成分更重的房源展示工具，包括 SkyTour（使用 Gaussian Splatting 将无人机影像转为 3D 渲染）与 Showcase 的 Virtual Staging（加入合成家具），同时承认信任风险，并呼吁清楚的浮水印与揭露。尽管在技术上投入 billions，Zillow 表示这些进阶展示功能的采用率仍仅在 single-digit percentage；一项重要押注 Zillow Immerse for Apple Vision Pro 于 February 2024 上线，但未成为广泛受消费者关注的产品。在内部，Wacksman 说 AI 提升了工程、客服与设计的生产力，使人力规模维持「relatively flat」，并将近期裁员描述为仅涉及少数表现不佳者。

策略风险在于分发（distribution）：房地产房源多属公开资讯，AI 聊天机器人与搜寻平台可能成为前端入口，绕过 Zillow 的 app。Zillow 早期就与 OpenAI 整合，但在测试中，若未明确启用 Zillow app，连结在 Zillow 与 Homes.com、Realtor.com 等竞争对手之间的呈现并不一致，凸显微小的路由选择就可能改变流量。更大的威胁是 Google 在数个城市与 HouseCanary 合作的实验：在 Google agent ads 旁展示房源，并提供类似 Zillow 的图库与预约连结；Google 表示该测试不涉及 Gemini，且目前未在 AI model 上投放任何广告，但此举仍与 Zillow 股价再次下挫同时发生，并引发收购揣测。Wacksman 淡化来自约 $4 trillion 的 Google 的威胁，押注 Zillow 的直接流量与「深度整合的体验」再加上与真人房仲的合作，将在未来 20 years 持续存在；不过文章也警告，以 AI agent 驱动的典范最终可能取代今日以 app 为中心的模式。

WIRED’s Steven Levy reports on Zillow on February 13, 2026, as the US housing market remains weak. CEO Jeremy Wacksman says the market is “bouncing along the bottom,” with only 4.1 million existing homes sold last year versus a “normal” 5.5 to 6 million, and he expects only marginal improvement in 2026; he notes Zillow has outperformed the broader industry but is still valued at about 1/4 of its 2021 peak. After Zillow reported higher earnings in the last quarter, its stock nonetheless fell nearly 5 percent the next day, and Wacksman frames AI as “an ingredient rather than a threat” that can both protect Zillow’s position and change how people search (for example, queries like rentals under $3,000 per month with specific lifestyle constraints).

Wacksman argues Zillow has long relied on AI, citing Zestimates as a machine-learning core feature, and says the recent LLM wave is accelerating AI integration across the company. Zillow is pushing AI-heavy listing presentation tools, including SkyTour (using Gaussian Splatting to turn drone footage into 3D renderings) and Showcase’s Virtual Staging (adding synthetic furniture), while acknowledging trust risks and calling for clear watermarking and disclosure. Despite spending billions on technology, Zillow says adoption of these advanced display features is only in the single-digit percentage range; a prominent bet, Zillow Immerse for Apple Vision Pro, launched in February 2024 and has not achieved broad consumer prominence. Internally, Wacksman says AI has boosted productivity in engineering, support, and design, helping keep headcount “relatively flat,” while characterizing recent layoffs as involving only a handful of underperformers.

The strategic risk is distribution: real-estate listings are largely public information, and AI chatbots and search platforms could become the front end that bypasses Zillow’s app. Zillow integrated early with OpenAI, but in testing, links surfaced inconsistently across Zillow and rivals like Homes.com and Realtor.com unless the Zillow app was explicitly activated, underscoring how small routing choices can shift traffic. A larger threat is Google’s experiments in several cities, built with HouseCanary, that show listings alongside Google agent ads with Zillow-like galleries and scheduling links; Google says the test does not involve Gemini and currently runs no ads on the AI model, but the move still coincided with another Zillow stock dip and sparked acquisition speculation. Wacksman downplays the danger from the roughly $4 trillion Google, betting Zillow’s direct traffic and “deep, integrated experience” plus human realtor partnerships will endure for the next 20 years, even as the article cautions that an AI-agent-driven paradigm could eventually replace today’s app-centric model.