memes. likes. views.
On Nov 8, Nvidia's CEO reportedly told staff he had been unfaithful. X answered the way it does, and kept answering as the story grew. This is a count of that answer: how many jokes, how fast they came, which ones won, and what the share price was doing at the time. It takes no position on the people in them.
The share price, everything X said, and the memes, hour by hour
Three charts share one timeline, each with its own scale. Nvidia closed at on Friday Nov 6, before the story, and stood at at the last reading, lower. The story broke over a weekend, so X had two days with it before the market could react.
How fast did it move?
The busiest day for memes was , with . The busiest single hour for X as a whole was , when posts mentioned the story.
Half of the memes were up by
The jokes came in waves, one for each turn of the story
From Saturday Nov 7, UTC. The shaded bands are 11pm to 6am in California, when the US went quiet and the rest of the world kept posting.
Which ones won?
The biggest meme has likes. Attention on a platform is never shared out evenly, and this was no exception.
The 15 most liked memes
Titles are our own descriptions. Pick one to read the post on X.
A hundred memes took of the likes
The bottom share between them.
About one view in turns into a like
Each step is ten times more. Dots above the dotted line were liked more often than usual for their reach.
What kind of joke was it?
Each meme is sorted two ways: by format, read from the post itself, and by joke type, judged by a reader.
Videos were of the memes and took of the views
A GIF counts as an image here.
"" was the most common joke
"" earned the most per meme: likes each across memes.
The joke changed as the story did
One row per joke type, earliest first. Each row is drawn to its own peak, so compare the shapes, not the heights. "" is the last type to arrive.
How much of it was actually jokes?
A search cannot ask for memes. It can only ask for words. We pulled every original post with 100 likes or more that matched, of them, and each one was read and sorted. Jokes were of the posts and took of the likes. Only of the jokes carried an image or a video, and those are the memes counted on this site.
News and opinion outnumbered the jokes. The jokes were liked more.
Which ones started an argument?
Likes are agreement. Replies are people with something to say. Across the memes there are replies for every 100 likes, so a post far above that got people talking back.
The most argued-with meme drew replies per 100 likes
The five that got people talking back
Memes with at least 5,000 likes, ranked by replies per 100 likes.
What did the AI ones cost to make?
of the videos are described as AI-made by a news source or by the person who posted them. That is a floor: nobody has checked the rest. Together the run seconds. Nobody publishes the bill for a meme, so the figures below are an estimate built from stated assumptions.
Longer clips cost more, and length is all we can measure
The estimate scales with seconds of video. One detector counted or more AI-made posts about the story on X by Monday Nov 9. At the same average length that is about tokens and GPU-hours, most of it very likely on Nvidia hardware.
What one meme wave teaches about data
You can only search for words
Our first search asked for his name and a scandal word together. It found 14 of 80 joke posts we already knew. Most jokes never spell it out.
Counting is cheap, reading is not
X will count posts for half a cent. Telling a joke from a headline meant reading of them.
Averages hide the shape
The average meme has likes, the median . When a hundred posts hold of the likes, the average describes almost nobody.
Together is not because
The share price and the meme count moved in the same week. The price was already falling before the first joke. A chart can show timing. It cannot show cause.