As a PR pro, I love to build buzz for clients; it’s in my veins. Also, I work in the AI space at Fusion PR. We were all quite amazed by the sudden rise of DeepSeek. The Chinese AI vendor seemed to come from nowhere and pull the rug out from the industry this week.
It was important for us to quickly get our arms around this news and its implications. I was curious about how the company seemingly overnight commanded massive buzz and media coverage, rising while U.S. AI stocks and related sectors like energy fell. It’s kind of a geeky story, and DeepSeek had virtually no U.S. brand recognition up until one week ago.
Why all the press attention?
It may be easier to answer this question than exactly how the buzz unfolded. In reading up on the story; about who DeepSeek is, and what they offer – it seemed pretty clear.
All this press attention was not due to some cleverly conceived PR campaign. It is what happens when an overheated AI market meets fears about threats to U.S. tech innovation and dominance meets the chink in the AI armor.
Let’s face it, many if not most have been captivated by the siren call of AI. It’s pretty amazing, and progressing rapidly. We know there are costs and risks. But many are willing to look past these as VCs pour billions into the space and stocks climb for the major players.
We’ve accepted the premise that it will take tons of money, data and energy to keep the ship moving forward. But we try not to thing too hard about bothersome things like environmental impact, whether the tech will bite back, and if all the investment will be worth it – for the companies, investors and society.
It seems to be a Faustian bargain we’ve accepted. And then, a company comes along from out of nowhere (well, from China, actually) that says you can have all that great AI stuff with much less of the downside. China’s DeepSeek apparently found a way with their just-released R1 model to provide similar solutions that require less energy, fewer chips, etc. Moreover, they did it with low cost tech, e.g. relying on open source software; a “joke of a budgret” as reported in Fortune.
This Reuters story, by Eduardo Baptista, said that DeepSeek R1 is “20 to 50 times cheaper to use than OpenAI o1 model, depending on the task, according to a post on DeepSeek’s official WeChat account.”
‘Tastes great, less filling,” as the old Miller Lite beer TV ad said. DeepSeek’s accomplishment made the current crop of U.S. AI leaders seem flat footed and was of great interest to anyone watching and reporting on the AI space.
How much buzz did they get and how did it unfold?
I tried to answer these questions by asking people and technology. It is apparently not that easy to track the trajectory of such (hint hint PR tech vendors, yell if you have an answer).
I wanted to research exactly how the DeepSeek R1 buzz unfolded. When did US media pick up on the story and when did it start to spike? Did one major publication lead the wave? (Yes, I know it isn’t always media that set the wheels in motion – e.g. the buzz could have started on social media perhaps giving a reporter a scoop; but I did want to study media coverage, as one big scoop or story often triggers others).
Answers weren’t jumping out so I asked around.
Is that your final answer, ChatGPT?
I did what people do in growing numbers these days. I asked AI, figuring what the heck – costs me nothing, is easy enoough, and sometimes does truly amaze.
But the answers were inadequate. I asked Bing CoPilot, ChatGPT and Perplexity to help me understand when and how the massive flood of coverage started to take off. They mistakenly pegged the first articles to The Atlantic, Wired even the NY Post circal 1/23. A simple Google News search showed that they were wrong – articles came out earlier, from major and minor news outlets (I’ll get further into that in just a bit).
I informed the chatbots that they were wrong, and asked each to try again. They were all apologetic; but let’s face it, we know AI can tell lies, even about other AI. I’ve taken to chiding them and asking in frustration: “Is that your final answer?”
Fusion staff chime in
I asked our team for their thoughts on the DeepSeek media frenzy. Seth Menaker shared some interesting numbers, from MuckRack
I also asked my friend Tressa Robbins, who is VP of Client Success at global media monitoring and PR measurement company Truescope. She did some digging around in their system this Wednesday and said:
On January 20 VentureBeat and TechCrunch both posted stories within minutes of one another around mid-day. That evening, Yahoo!Finance published a story. IMHO those are the stories that kicked off the current surge…
Tressa Robbins, Truescope
She sadded: “As you’ve found, it’s increasingly difficult to trace back to a single source.”
Tracking the DeepSeek Big Bang Event
It seems AI should do be able to do a better job of tracking these things. I am sure vendors are working on improving their wares. Indeed, Tressa showed me the nice chat interface they now have with Truescope.
But none seem to be there yet, that can give us inquisitive PR folks a way to just ask for what we want and get answers that we can consistently rely on. In the meanwhile, I did some down and dirty detective work – aided by Google News searches, reading articles, and yes dabbling with a range of tools.
On Jan. 20, the Hangzhou, China-based DeepSeek released R1, a reasoning model that outperformed Open AI’s latest o1 model in many third-party tests… The buzz around DeepSeek’s R1 seemingly picked up steam after Alexandr Wang, CEO of Scale AI, touted its competitiveness against the best products from the U.S.
“What we found is that DeepSeek, which is the leading Chinese AI lab, their model is actually the top performing, or roughly on par with the best American models,” Wang said on CNBC from the World Economic Forum in Davos, Switzerland, last week.
Wang spoke with CNBC at WEF on Thursday, January 23. Then, over last weekend, the “hype… reached a fever pitch on social media,” wrote CNBC reported Yun Li, Noted investors like Marc Andreesen sang DeepSeek’s praises on X.
I could not find a formal announcement from DeepSeek, but did catch this Tweet, from 1/20. They have about 800K X followers.
As Tressa Robbins noted, TechCrunch and VentureBeat were amongst the first U.S. media to write about it, shortly after DeepSeek tweeted about the new model. ZDnet was also early to the game: Radhika Rajkumar wrote this story for the publication on 1/21.
So, it seems the story simmered for a few days; social media buzz built over last weekend (when most reporters are not filinng stories). The media coverage really began to take of on January 23, as shown in the above chart from MeltWater.
