Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Tokens Land on Double 11 Shelves: Prices as Low as 1 Cent, 2 Trillion Sold in 30 Minutes

AI services became a headline category in this year's Double 11 shopping festival, with token packages priced as low as one cent and one livestream selling 2 trillion tokens in half an hour, according to National Business Daily. The real test is retention: cheap tokens pull in first-time buyers, but converting them into paying users is a different problem.

Published

AI compute showed up this year as an actual consumer product on Double 11 shelves. According to a November 3 report from National Business Daily, token-based compute packages were listed on Chinese e-commerce platforms at prices as low as one cent, with one livestream channel selling 2 trillion tokens in half an hour.

Packaging tokens as a standard good is the trick that makes the model work. Enterprises used to buy compute through procurement contracts and monthly billing, and ordinary consumers had almost no way in. As a product listing with a near-zero first-purchase price, the barrier to trying an AI service drops to almost nothing.

Volume followed the discount. Multiple outlets reported the listings moving on the order of 100,000 units in half an hour, equivalent to those 2 trillion tokens—a marketing moment that also stress-tested payment, entitlement issuance and quota reconciliation on the back end.

The harder problem is retention, which the report singles out. A one-cent price cannot cover inference costs; the bet is that users keep paying at normal rates once promotional credits run out. If most buyers take the discount and leave, the campaign is simply customer acquisition spend dressed up as revenue.

Why it matters: AI services are shifting from per-call pricing for developers to per-item retail for consumers. That changes the pricing conversation from dollars per million tokens to bundles, livestream flash prices, platform commissions and subsidies—a much more consumer-shaped business model.

It also signals that model capability is becoming table stakes. Once capability converges, advantage moves to whoever sells tokens cheapest, lowers the usage barrier furthest and holds on to users longest. Double 11 provided a large-scale experiment; the results will only be visible after the subsidies fade.

What to watch: repeat-purchase rates on discounted tokens, whether platforms turn compute into a permanent category rather than a festival stunt, and how vendors balance margin against scale.

Why it matters

Selling compute as consumer retail pushes AI vendors to compete on channel, pricing and retention, not just model quality; the near-term fight is acquisition, the long-term one is subscription revenue.

TokenE-commerceDouble 11
Back to realtime news

Nearby Updates

All

11/04, 01:03

Alibaba Cloud Brings AI and Cloud Infrastructure to Dakar 2026 Youth Olympics

Alibaba is supplying cloud computing and AI technology for the Dakar 2026 Youth Olympic Games, the first edition of the event held in Africa. The deployment puts Alibaba Cloud's sports-technology stack into a high-pressure, low-tolerance environment far from its home market.

11/04, 01:07

Anthropic's IPO Filing Shows $17B Revenue—and Compute Eating 58% of It

Anthropic's updated IPO filing disclosed quarterly revenue of $17 billion, up from $11.5 billion in the prior quarter, while compute costs climbed to 58% of revenue from 48%. The operating loss widened, putting unit economics at the center of the company's listing story.

11/04, 00:43

Chargebacks911 sells merchants cover for chargebacks on AI agent purchases

Chargebacks911 has launched an AI agent purchase indemnification product that identifies agent-initiated transactions and covers qualifying chargebacks for merchants. The capability is bolt-on, so merchants keep their existing fraud prevention stack and payments infrastructure, and the company says it applies across verticals and markets.

11/04, 00:35

IBM's Palette splits decks into a designer and a coder, and a 20B model keeps up

IBM Research has introduced Palette, a 20B open-weight model fine-tuned to build slide decks in a specific design language by separating a designer that writes a JSON brief from a coder that generates each slide. On PresentBench it ranks first in the Economics and Education domains and second overall at 59.7 points, behind NotebookLM's 60.8.