Video taking off 9.7x its usual first day

Asianometry traces Zhipu AI's path from campus to corporate customers

The GLM company profile looks beyond model performance to a harder question: how to pay for the research.

Channel:

Asianometry Story of November 3, 2026Trari Newsroom

Why this story? The video opened at a pace of 99,394 views a day, 9.7 times a normal first day on this channel (10,221).

The Rise of GLM and China’s Zhipu AI

Things we can learn from the video

Press to jump

0:34Zhang Peng's work with Tang Jie at Tsinghua provides the academic roots of Zhipu.
4:28OpenAI's GPT models push Zhipu toward building its own bilingual language models.
10:19Zhipu trains GLM-130B on Nvidia servers and releases it openly.
16:48Crowded consumer chatbot markets prompt a shift toward enterprise customers.
20:00The essay follows GLM-4.5's open-weight release and subsequent model updates.
22:04The narrator questions the business model as research spending exceeds twice the company's revenue.

A Chinese AI company with university roots has given Asianometry an unusually fast start. "The Rise of GLM and China's Zhipu AI" drew roughly 100,000 views in its first day, nearly ten times the channel's usual opening audience.

The essay follows Zhipu from Tsinghua University's Knowledge Engineering Group into the business of building language models. Zhang Peng's work with professor Tang Jie on the academic research platform AMiner provides the starting point. The company takes shape as China makes it easier to turn university research into commercial ventures.

Then GPT changes the assignment. Asianometry explains how Zhipu moves away from traditional language-processing pipelines toward its own bilingual models, eventually releasing GLM-130B openly. ChatGPT's arrival brings a rush of interest, investment and new Chinese model developers.

The most revealing turn is less about a chatbot's answers than who will pay for them. Facing a crowded consumer market, Zhipu shifts toward enterprise customers. The essay follows that strategy through later GLM releases and a Hong Kong listing, before returning to the expense underneath it all: the narrator says research and development spending exceeds twice the company's revenue.

Viewers are bringing their own experience to that business history. Several discuss using GLM for coding; one describes its accidentally comic responses while programming. Another asks Asianometry for more profiles of AI labs focused on their origins and operations, rather than just their models. For a channel whose most-watched regular video is "Why the Soviet Computer Failed," the interest in how technology companies are built is familiar territory.

Published on Trari

Where next?

Also in Explainers, history and educationSubscriber surge +99,972 subscribers in a dayTED reaches 28 million subscribers with connection on the agendaTalks about keeping friends through parenthood and reconsidering romantic dealbreakers accompany an unusually large subscriber gain.Channel: TED · Story of November 3, 2026

A different discovery todayVideo taking off 12x its usual first dayFrom "please don't leave" to "ask Grok": John Oliver's midterm tourThe election-week main story drew about 3.8 million views in under a day, and its comment section filled more than twice as fast as usual.Channel: LastWeekTonight · Story of November 3, 2026

What viewers say

Is this Asianometrys AI era? Would be great to have primers on the big labs, foucsing on their history + operations rather than just the models like everyone else
@tanjeeschuan4999See it on YouTube
Nice video.
After programming with GLM for two months, I can say: I like the humor in GLM 5.x, but it doesn’t know it’s being funny.
“Oops, ended up on the wrong branch again. Cleaning up my mess.” :D
@MeffenOneSee it on YouTube
Watching this while Claude burns my last tokens before weekly reset. 😊
@Tential1See it on YouTube

3 more comments

I've been using GLM 5.3 Flash for the past few months as my primary coding agent. It's absolutely fantastic and it compares very strongly to the leading frontier models. In everyday tasks, it gets the job done and more.
@uncaringbearSee it on YouTube
Great perspective on Chinese AI. It is opaque for most Western Chinese model users, except for benchmarks and cost.
@flyingkiwiguySee it on YouTube
Zhipu’s challenge is turning strong research into a sustainable business. Enterprise customers may offer steadier revenue than consumer AI, but heavy R&D costs can still outpace adoption. The key question is whether GLM can become infrastructure businesses depend on, rather than just another competitive model.
@SiliconNodeReportsSee it on YouTube

Public comments from the video's page on YouTube