Google's Bard demo gets a single fact wrong, and Alphabet's market cap drops by 100 billion dollars
A scripted promotional video for Google's chat-AI product showed the model attributing a James Webb space-telescope first to the wrong instrument. By the time the trading day in New York closed, the company had lost roughly the GDP of Hungary.
On 8 February 2023, the day after Microsoft had unveiled an AI-augmented Bing in Redmond, Google held a hastily organised event in Paris to introduce Bard, its own consumer chat product. The session was brief, the demo was pre-recorded, and the model in the demo attributed the first picture of an exoplanet outside our solar system to the James Webb Space Telescope. The first such image was, in fact, taken in 2004, by the European Southern Observatory's Very Large Telescope, eighteen years before Webb launched.
Astronomers caught the error within minutes. As The Verge reported the same afternoon, the correction circulated on social media before the Paris event had finished. By the close of trading on Wall Street that evening, Alphabet's share price had fallen by 7.7 per cent, wiping approximately 100 billion US dollars from the company's market capitalisation.
What the market actually priced
Analysts who had been writing variations of "Google's AI lead is unassailable" for a decade adjusted in real time. As the Financial Times noted in coverage the following morning, the move was less a vote of confidence in Microsoft's product than a vote of no confidence in Google's ability to translate research leadership into a comparable consumer experience. Google had quietly run state-of-the-art language models internally since 2019; what it had not done was show it could ship them.
It was not a 100bn dollar error. It was a 100bn dollar permission slip for the market to ask a question it had been avoiding.
The factual error in the demo became, within weeks, more myth than fact. Google's Bard product corrected on launch and the Webb-versus-VLT mix-up entered the bin of widely repeated anecdotes that no longer match the underlying record. The market reaction, which the demo did not directly cause but did precipitate, has aged better. By the end of 2023, Alphabet had recovered the losses; by mid-2024, the company was again the AI-narrative incumbent. The episode is now studied as the moment a research-leadership story stopped being enough on its own.
Discussion