The Next Telco Boom in GenAI
(and maybe that's ok)
Benedict Evans has an annual report that's as great a summary of the tech industry as any. Always worth your time: https://www.ben-evans.com/presentations
I’ve highlighted some slides and my key takeaways here:
There's still a lot to do to make a 'sticky' (durably useful) product for most people. This means that while there's a lot of potential for AI, there are still many challenges to overcome before it can be truly mainstream.
Enterprise adoption of AI is still largely exploratory. This suggests that we're still in the early stages of the hype cycle, and there's a lot of uncertainty about how AI will be used in the long term. Claims of current revenue are not good indicators of long-term revenue.
The capital expenditure (capex) needed for generative AI base models is turning base model companies into something closer to telcos like Verizon, AT&T, etc. This has implications for strategy and culture, as these companies will need to focus on building and maintaining infrastructure. Former employee-first cultures were downstream of how employees were the main cost. Now they are only one of the main costs, and so only one of the main priorities.
Google excels at 'speeds-and-feeds'. This refers to Google's ability to provide fast and reliable infrastructure, which could be a key advantage in the AI market for genAI models. There is an open question for the industry on whether the key value-capture part of the value chain is in base models versus at a layer above. A historical example might be that many telecom companies in the 1990s dotcom bubble built a lot of great infrastructure but it was the advertising companies that sat on top that got all the rewards (Google, Meta, etc.)
Note: I am a google employee so only analyzing based on this chart.
All signs still point to how the AI market is still in its early hype-cycle stages. And maybe that’s ok!
has a great book out this past week about exactly how bubbles are maybe Good, Actually: they pull forward new invention that wouldn’t have happened otherwise by creating both a place and a time for everyone to build this stuff out.