If the adoption of the AI feels different from any technological revolution that you may have known before – mobile, social and cloud computing – this is in fact the case.
The venture capital Mary Meeker has just fallen A 340 -page slideshow report – who used the word “unprecedented” on 51 of these pages – to describe the speed at which AI is developed, adopted, spent and used, saved with a graph after the graph.
“The rhythm and scope of change linked to the evolution of artificial intelligence technology are indeed unprecedented, as the data support it,” she writes in the report, entitled “Trends – Artificial Intelligence”.
There is a certain poetic story to this person who writes this kind of report. Meeker is the founder and general partner of VC Firm Bond and was once Known as Queen of Internet For its previous annual internet reports. Before founding Bond, she led the growth practice of Kleiner Perkins, from 2010 to 2019, where she supported companies like Facebook, Spotify, Ring and Block (then Carré).
She has not published a report on trends since 2019. But she dusted her skills to document, in laser detail, how AI adoption has exceeded any other technology in human history.
Cat achievement 800 million users in 17 months: unprecedented. The number of companies and the rate to which so many others reach high annual recurring income rates: also unprecedented.
The speed at which the costs of use drop: unprecedented. Although the training costs of a model (also unprecedented) reach up to 1 billion dollars, inference costs – for example, those who pay to use technology – have already dropped 99% over two years, when calculating the cost by 1 million tokens, she writes, citing research Stanford.
The pace to which the competitors correspond to the characteristics of each other, to a fraction of the cost, including the open source options, in particular the Chinese models: unprecedented. For example, she points out that Nvidia’s Blackwell 2024 GPU uses 105,000x less energy per token than the company GPU predecessor 2014 of the company.
Meanwhile, Google chips, such as its TPU (tensor processing unit) and Amazon trainium, are developed on a large scale for their clouds- which moves quickly too. “These are not secondary projects – these are fundamental bets,” she writes.
The only area where AI has not exceeded all other technological revolution is in financial returns. While VCs pay money on the AI fire as quickly as possible, AI companies and cloud service providers are also burning in cash. AI requires massive investments in infrastructure.
It’s good for consumers and businesses, beneficiaries of rapid improvements, while competition reduces costs, underlines Meeker. But the jury is still absent on which of the current business harvest will become new, long -term, profitable long -term technological giants. “The only time will tell us what side of the money equation that the current aspirants of the AI will land,” she writes.
As for the rest of us: keep your hats.
