Global AI-Related Product Trade Rises to $4.18 Trillion
According to data from the World Trade Organization, the trade volume of AI-enabled products is expected to reach $4.18 trillion by 2025, a 21.9% increase from $3.43 trillion in 2024. These products account for about one-sixth of global goods trade but contribute approximately 42% of the annual increase in goods trade.
During the same period, global goods trade volume is projected to grow by 4.6%, driven mainly by semiconductors, servers, and data transmission equipment. The share of AI-related products in world trade is expected to rise from about 13% in 2023 to nearly 17% by the end of 2025. Asia is expected to handle about 62% of AI-enabled trade, while North America becomes the fastest-growing market.
WTO Director-General Ngozi Okonjo-Iweala announced these figures at the inaugural World Trade and Technology Day in Geneva, stating that AI demand could contribute an additional 0.5 percentage points to global goods trade growth in 2026. The Japan External Trade Organization estimates that AI-related exports will reach approximately $4.1547 trillion in 2025, a year-on-year increase of about 22.8%, contributing around 45.1% to global export growth.
Consumer hardware is moving away from a single mobile phone interface. Qualcomm has demonstrated a generative assistant on glasses that does not rely on a phone or cloud; products like Thunderbird have added eSIM capabilities for independent calling and real-time translation. Meta's Ray-Ban glasses are projected to sell about 7 million pairs in 2025, totaling around 9 million pairs, and work with wristband gestures to create a hands-free operation loop. Open-source projects are enabling glasses to relay visual and audio data to an Agent for task execution without needing to open a mobile app.
At the Hangzhou Yunqi Expo, AI glasses and self-assembling devices were showcased, synchronizing voice and images to be executed by an Agent without the need to open a mobile application. According to Counterpoint, global smart glasses shipments are expected to increase by about 110% year-on-year in the first half of 2025, with AI glasses accounting for about 78% of that.
The buyers are data centers, chip, and terminal brands, while the sellers are Asian assemblers and North American computing power purchasers. Funding is shifting from the premium of complete mobile phones to mid-range chips, servers, and glasses modules. Beneficiaries include semiconductors, optical modules, and independently connected wearables, while those under pressure are the old hardware chains that can only produce mobile peripherals and cannot delegate tasks to an Agent.
U.S. imports of mid-to-high AI-related products are expected to rise from $379 billion in 2023 to $654 billion in 2025, an increase of about 72.6%, with their share of imports rising to about 23%; excluding these products, the goods trade deficit could be reduced by nearly $200 billion.
Source: Public Information
ABAB AI Insight
The World Trade Organization has classified chips, servers, and transmission devices as "AI-enabled products," providing a quantifiable metric for trade growth for the first time. In 2023, these products accounted for about 13% of world trade, rising to nearly 17% two years later, contributing 42% to trade growth. This is not just a statistic of application explosion but a rebranding of the past decade's East Asian manufacturing and North American capital expenditure with an AI label. Ngozi Okonjo-Iweala shifted the narrative from tariff wars to computing power goods during Technology Day.
Money is shifting from mobile phone profits to three areas: wafers and advanced packaging, cabinets and optical modules, and glasses and modular terminals. U.S. imports of computing hardware have nearly doubled in two years, with Mexico and Taiwan handling about half of U.S. AI trade. Chinese manufacturers showcased modular devices and glasses at the Yunqi Expo, aiming to move the voice and image interface from app stores to facial sensors, allowing Agents to directly process orders and bypass Apple's and Google's distribution fees.
This is analogous to the smartphone boom from 2010 to 2015, which drove trade in touch screens and baseband chips: at that time, growth was also concentrated in a few HS codes. Today's position is during an infrastructure expansion phase, with consumer terminals just beginning to "de-mobile". Qualcomm is embedding small models into glasses, while Meta is validating that optical modules can operate independently of phones with 7 million units shipped; Thunderbird's addition of eSIM is learning from feature phones to connect independently. A failed comparison is the first generation of Google Glass: it had sensors but no Agent, only serving as a camera peripheral.
The structural changes belong to the reconstruction of the industrial chain. The largest portion of the $4.18 trillion trade volume remains computing chips, while glasses merely shift the interaction layer from pockets to faces. The mechanism is that once models can process voice and images in real-time, the pricing power of the phone as the sole gateway diminishes, with value added moving to sensors, local inference, and order-capable Agents. Whoever controls mid-range chips and independently connected modules will capture this round of trade growth; those only making phone cases and applications will become accessory suppliers for glasses.
ABAB News · Cognitive Laws
Trade growth appears before application profits.
Once the interface leaves the screen, distribution fees begin to lose effectiveness.
Chip coding rewrites the global goods ledger earlier than slogans.