Gemini Launches Shopping Feature: 1000 Days of AI Transforming the Consumption Entry Point

Author | Xia Sijia
Editor | Qiao Qian

At the beginning of 2026, retail giants once again appeared on the global AI competition stage.

On January 11th, Walmart and Google announced plans to integrate the products of Walmart and Sam's Club into Google's Gemini. Meanwhile, at the National Retail Federation (NRF) conference, Google officially launched the Universal Commerce Protocol (UCP) to provide intelligent shopping capabilities for Google Search and Gemini's AI models. US users can browse products and complete purchases in Gemini's dialog box without leaving the AI chat interface.

Previously, the last company to turn AI conversations into a shopping scenario was OpenAI. At the end of September 2025, ChatGPT launched "Instant Checkout", which completed the full shopping loop from conversation to order placement.

Since the release of ChatGPT at the end of 2022, in just three years, the technology industry has experienced multiple rounds of high - frequency "catch - up" competition. After a round of competition for search entry points, the focus of the AI competition is further extending to the e - commerce field.

According to the New York Post, on Black Friday at the end of November 2025, the busiest shopping day in the US, consumers used artificial intelligence to search, compare prices, filter, and make decisions, driving online consumption to a record $11.8 billion. Data from Adobe Analytics showed that the online consumption in the US on this day increased by 9.1% compared with the same period last year. AI is gradually becoming an important variable in the shopping chain.

In the information age, those who control the entry points control the world. Whoever can explore real and sustainable transaction capabilities first will gain the upper hand. Now, this competition is no longer limited to a single track, but a three - way game around transaction entry points. Shopping platforms, search engine companies, and large - model manufacturers are being forced to sit at the same table.

In contrast, the commercial competition around large models in China still mostly occurs within existing platform ecosystems. Even in the case of the Doubao phone, which appears in the form of a system, the transaction behavior in specific operations still has to return to the e - commerce platform to be completed. Therefore, although companies are frequently taking actions in product iteration, organizational adjustment, and talent competition, the real direct confrontation around "cross - platform entry points" has not fully begun.

In this context, observing the struggle history of overseas technology companies in this round of transformation may help us better understand how the business order will be reconstructed when AI moves to the front end of transactions.

01 From Search to Payment

The launch of UCP by Gemini 3 is another attempt to counter OpenAI.

At the end of September 2025, ChatGPT under OpenAI launched the "Instant Checkout" function and quickly reached a cooperation with US retail giant Walmart in October. "Instant Checkout" is an upgrade based on the "shopping function" that ChatGPT launched in April of the same year.

This means that generative AI is no longer just providing product search links but is starting to directly intervene in the transaction process. Users can complete inquiries, place orders, and make payments (supporting Apple Pay, Google Pay, Stripe, and credit cards) within ChatGPT's dialog box without jumping to third - party e - commerce platforms, forming a complete shopping loop. Currently, this function is only available to US users.

At that time, Fortune magazine pointed out that this move was an important step for OpenAI to commercialize its 700 million weekly active users (as of July 2025) and might take a large market share from traditional Google search ads.

In the past, people usually searched for target products on Google Shopping first, compared prices repeatedly, checked reviews, and then placed orders on e - commerce platforms or merchants' official websites after making a decision. Now, this process has been greatly compressed - as long as users ask shopping - related questions to AI, such as "What are the best sports shoes for mountain climbing?", AI can directly display relevant products, pictures, reviews, prices, and purchase links from merchants, and even complete payments directly.

As is well - known, most of Google's revenue is based on the model of "leading users to websites and then monetizing through ads". Once users start their shopping journey from external AIs like ChatGPT, and the starting point of shopping is not to search for products first but to ask questions to AI, Google will lose a part of its traffic source. For Google, this means not only the loss of a single click but also potential high - profit advertising fees.

In fact, since its birth, the conflict between OpenAI and Google has never stopped.

For the past two decades, Google has been dominant, occupying 90% of the global search market. Until November 2022, OpenAI emerged out of nowhere. In less than three months, OpenAI's user number exceeded 100 million. This speed far exceeded TikTok's nine months and Instagram's two and a half years.

At that time, smelling the crisis, Google immediately issued a "red alert" and recalled its two founders, Sergey Brin and Larry Page, to the company to jointly face the challenge. According to the New York Times, the rapid rise of ChatGPT was regarded as a potential threat to Google because it "seemed to be able to provide a brand - new way to search for information on the Internet".

As one of OpenAI's largest investors, in February 2023, Microsoft took action. It launched NewBing before all its competitors, which embedded OpenAI's generative function into its web search engine Bing, providing more intelligent conversational answers in search results. As the first mainstream product to integrate generative AI into the search entry, the emergence of NewBing itself was a rewrite of the traditional search logic. For more than a decade, Bing's market share had always been suppressed by Google. The emergence of OpenAI gave Microsoft the first real possibility to challenge Google's dominant position in the search market.

Facing the siege, eager to prove its AI capabilities, Google quickly launched the large model Bard (later renamed Gemini) in March 2023, trying to counter ChatGPT. It is worth noting that before its official launch, Bard made a factual error in a demo video, causing Google's stock price to plummet by 7.68% after the US stock market opened and wiping out about $105.6 billion in market value overnight, exposing its dilemma when facing new technologies at that time.

It wasn't until May 2024 at Google I/O that Google announced the official launch of the AI Overviews function. AI could directly provide answer summaries generated by Gemini at the top of search results. The launch of this function was also called "one of the biggest changes in Google Search in 25 years".

But soon, Google faced a more direct challenge: on July 26, 2024, OpenAI officially launched SearchGPT, allowing ChatGPT to connect to the Internet in real - time, search for the latest Internet information to generate answers, and provide source links. This marked that ChatGPT was transforming from a simple AI chatbot to a search tool. After the launch of SearchGPT, Google's stock price dropped by more than 3% on the same day.

This transformation also laid the foundation for its subsequent application in consumption. As people talk to AI more and more frequently, some people gradually start to leave their consumption decisions to AI during the chat, asking AI to recommend products and provide purchase links. In 2025, OpenAI's "Instant Checkout" function was the commercial implementation of this search logic - in addition to providing shopping links, it further took over the payment process to form a complete loop.

In fact, as early as November 2024, the artificial intelligence search engine Perplexity entered the e - commerce field. The function it launched could provide shopping recommendations in Perplexity's search results and allow users to place orders directly without visiting retailers' websites. Meanwhile, Perplexity also launched a merchant program, stating that if merchants joined the program, their products were more likely to be recommended because the company would have more complete information in its index. Users could also use the one - click checkout function to purchase products from the company's partner merchants.

The above trends show that large - model companies are no longer satisfied with just providing information. They are trying to control the entry point of consumption decisions as shopping assistants. This not only subverts the original shopping paradigm in e - commerce but also forces traditional search giants to re - examine their competition strategies.

However, despite being highly challenged, Google's strong strength still worries its opponents. Especially the latest artificial intelligence model, Gemini 3, released in November 2025, received amazing reviews. Not only was it first applied to Google's most core search business on the day of its release, but it also topped the LMArena ranking (the global AI model arena) with 1501 points as soon as it went online, becoming the first model to break through 1500 points.

More importantly, Gemini is no longer just limited to answering questions but can complete complex tasks. For example, if a user gives a requirement for game development, it can automatically generate runnable code and a complete interface. Regarding this, Google CEO Sundar Pichai commented on Google's official blog: "Gemini 3 is our smartest model so far, aiming to turn any user's ideas into reality."

This also strongly indicates that after three years of development, Google is fully prepared.

Google's rapid progress even made OpenAI, which was once in the lead, issue a red alert. In response, OpenAI founder Sam Altman said in the podcast "Big Technology" that it was quite common for the company to issue red alerts. He said, "I think it's good to stay vigilant and take quick action when potential competitive threats emerge." He also said, "I estimate that we'll issue such alerts one or two times a year for a long time, which is actually to ensure that we stay ahead in our field."

Facing this complete reshuffle regarding traffic distribution, business models, ecological control, and discourse power, major technology companies are obviously stepping up their layouts to ensure a place in this transformation.

02 Amazon Is in a Hurry

The reshaping of the shopping entry point by AI soon put e - commerce giant Amazon at risk of being overturned.

To cope with the impact, in August 2025, Amazon added six AI - related crawler programs to its robots.txt file, intensifying efforts to prevent artificial intelligence companies from accessing its e - commerce platform. These companies included Meta, Google, and Huawei. Before that, Amazon had already imposed restrictions on crawler programs for companies such as Anthropic's Claude, Perplexity, and Google's Project Mariner.

Independent analyst Juozas Kaziukėnas, who first discovered these changes, posted on LinkedIn: "Amazon is doing its best to prevent artificial intelligence companies from using its data to train models." But he thought that the timing of these restriction measures might be too late to stop the training of artificial intelligence models because Amazon's data already existed in the data sets used by ChatGPT and other models. He pointed out: "But Amazon is obviously not interested in helping anyone build the future of AI shopping. If this is really the future, Amazon wants to build it itself."

In fact, Amazon is also developing its own artificial intelligence shopping tools. In mid - July 2024, Amazon launched the AI shopping assistant Rufus across the United States and later promoted it to other countries. Rufus can help users obtain product information more quickly and accurately, compare products, and access order details through conversational interactions. In April 2025, Amazon also conducted a small - scale test in the US of a function called "Buy For Me". This function can help consumers go to other platforms to search for and purchase products that cannot be bought on Amazon.

On the other hand, in early November 2025, Amazon also filed a lawsuit against Perplexity in the US Federal Court in San Francisco. The reason was that in July of the same year, Perplexity's AI - native browser Comet launched an intelligent assistant function that could search for and purchase products on multiple websites, including Amazon, on behalf of users. Amazon accused it of violating the platform's service terms and suspected of bypassing technical detection by disguising as a normal browser. Amazon also based on legal provisions such as the Computer Fraud and Abuse Act, asking the court to rule that Perplexity stop this behavior and seeking injunctive relief from the court.

Amazon hopes to continue to control the shopping process and commercialization rights to counter the "forced" cross - platform shopping operations of external AI companies. However, Shopify, Amazon's main competitor in the e - commerce field, adopted a different strategy.

Shopify did not try to keep AI companies out like Amazon but chose a more moderate approach. It requires that any AI shopping agent cannot place orders completely automatically and must retain the manual confirmation link. Meanwhile, Shopify also extended an olive branch to developers - if an AI tool wants to help users complete a purchase, the final payment and checkout steps are best directly connected to Shopify's checkout system. That is to say, although Shopify does not oppose AI becoming a new shopping entry point, it hopes to keep the place where transactions occur within its own system.

In addition, Shopify has always been actively seeking opportunities to cooperate with artificial intelligence companies. According to the Financial Times, OpenAI and Shopify started cooperation negotiations around allowing users to shop directly within ChatGPT in April 2025. At the end of September, when OpenAI launched the Instant Checkout function, Shopify announced the cooperation, stating that merchants could directly sell products in ChatGPT conversations.

Shopify's official announcement on December 10th showed that through the Agentic Storefronts program, it is pushing merchants' product catalogs to multiple conversational AI interfaces, including ChatGPT, Perplexity, and Microsoft Copilot, allowing consumers to discover products, compare, and complete transactions in conversations.

Ultimately, the difference between Shopify and Amazon in AI shopping stems from two different business positions. Shopify is essentially a provider of transaction infrastructure, like a "shovel seller". As long as merchants can sell more products and make more profits, Shopify can benefit. Therefore, it naturally welcomes any new entry point that can bring orders.

Amazon, on the contrary, relies on search, recommendation, and advertising as important sources of income. Once users make consumption decisions outside the platform, Amazon will lose not only a single order but also the stability of its traffic monetization system. This is what it cannot tolerate.

In general, in this competition around the next - generation AI - dominated search entry point and platform discourse power, all participants are constantly infiltrating each other's core areas. Amazon is accelerating the development of AI shopping and intelligent agent tools to stabilize its transaction system; Google is advancing its generative large - model capabilities to compete for the entry point; and large - model companies are starting to turn their attention to the browser itself.

However, at present, the functions of AI shopping are not yet mature. The long - criticized "hallucination" problem also exists in the shopping recommendation scenario. For example, when users ask ChatGPT about a specific product, they may get recommendations for non - existent or expired information. Many users have to return to Google or Amazon to make shopping decisions.

In response, Justine Moore, a partner at a16z, believes that although OpenAI is trying to embed e - commerce functions more deeply into the user experience and provide more relevant and timely product information, which will indeed cause Google to lose some search volume, he points out that so far, we have not seen this phenomenon occur on a large scale.

In each major technological transformation, the result of the game is often complementarity and symbiosis rather than complete replacement. Just as smartphones did not eliminate personal computers but reshaped the way we obtain and use information. In the future, the boundaries of search, transaction, and decision - making will be re - defined rather than simply disappearing.

Although AI shopping agents are still in the early stage of application, their impact on the business structure has already begun to show. The only thing that is certain is that in this transformation, no one can stay out of it.