Lots of healthy disagreement in this week’s THAT WAS THE WEEK tech show with Keith Teare. We debate the impact of AI on coding jobs, with Keith suggesting that while traditional coding skills may become less important, system architecture and AI guidance skills will be crucial to maintaining the value of human labor. We also discuss the rise of early-stage unicorns, military-tech AI start-ups, and disagree strongly on the status of billionaires, with Keith arguing that it’s “not hard” to be a billionaire in Silicon Valley today.
Here are the five KEEN ON takeaways from today’s conversation:
* Divergent Market and Valley Sentiment: While the stock market is having its worst performance since Trump's inauguration, Silicon Valley remains optimistic, particularly about AI. Keith argues there's no short-term correlation between Silicon Valley sentiment and market performance.
* Evolution of Tech Skills: The rise of AI is changing the nature of technical skills needed in startups. Keith suggests that traditional coding skills are becoming less crucial, while the ability to architect systems and guide AI is becoming more important. He notes that universities are already adapting their computer science programs to include AI.
* Rise of Efficient Startups: AI is enabling lean startups to do more with fewer people. Keith uses his own company Signal Rank as an example, noting they've built a complex system with just five people, two of whom are coders, highlighting a shift in how startups can be built efficiently.
* Military-Tech Convergence: There's a growing trend of Silicon Valley companies entering the defense sector, exemplified by Saronic raising $600 million for autonomous warships. This represents a broader shift in how military technology is being developed and funded through private companies.
* Debate about Wealth Creation: The conversation concludes with a debate about wealth accumulation, sparked by Robert Reich's controversial X post about billionaires. Keith argues that technology's global reach and distribution capabilities have made it easier than ever to build valuable companies, with Andrew strongly disputing the idea that becoming a billionaire is "not that hard."
That Was The Week - February 22, 2025
With Andrew Keen and Keith Teare
Andrew Keen: Hello everybody. It is Saturday, February the 22nd, 2025. The last Saturday in February, the last Saturday we're going to do That Was The Week tech roundup. It's been an odd week. On the one hand, the stocks notched the worst week since Trump's inauguration six weeks ago. It's been a long six weeks. According to the Financial Times, the geopolitical rupture, which of course has been caused by Trump, has sparked a quiet market rebellion. Niall Ferguson had an interesting piece in today's Wall Street Journal about the demise of the United States because of its massive debt, and Elon Musk has been continuing to make a public fool of himself this week, waving a chainsaw and pretending to be an Argentine politician, which I'm not sure reflects that well on him. However, in spite of all that bad news, Keith Teare's That Was The Week newsletter is actually very optimistic. Unicorns are back, according to Keith, and we have an image, of course, created by AI of these imaginary beasts horses with horns. Keith is joining us, as always, from Palo Alto, the home of optimism. Keith, do you think it's coincidental that suddenly everyone is optimistic again in Silicon Valley whilst the market is sliding to those two things in an odd way, kind of go together?
Keith Teare: There's no correlation between Silicon Valley and the markets at all in any day to day sense. There's long term correlation, but not short term. Silicon Valley is having a moment because of AI, and Grok Three was launched this week. Crunchbase launched its new AI driven data platform, and the CEO declared that historical data is dead, meaning only future predictive data is any good anymore.
Andrew Keen: And historical data being dead. The future is predictive intelligence. What does that mean?
Keith Teare: He means that it's now possible, because of AI, to see patterns and trends and predict them. Just knowing the past is not the point anymore. Obviously it's stretching a point. You still need the history from the past to see the trends. But he's saying the needle has turned from looking backwards to predicting the future because of data. That's true in biology as well. There's a massive arc this week announced a new model that understands DNA and can predict the likelihood of solving diseases.
Andrew Keen: Your editorial this week, Keith, is quite personal. You know that as the person in charge of Signal Rank, your startup, AI has been remarkably helpful in it. You refer in the editorial to an interesting piece in the New York Times about how AI is changing Silicon Valley build startups like your own Signal. What does your experience at Signal Rank tell us about the future of startups?
Keith Teare: Signal Rank is five people. Two of us have coding skills. We've raised $5 million ever to spend on building Signal. All the other money we raised is to invest in companies. That article is focusing on the fact that it's almost like the Lean Startup story from the early 2000s, except it's true this time, because the most expensive thing in a startup is people. And the one thing you need less of is people. That's a massive shift. Of course, if you're building large language models, the opposite is true, because the most expensive thing is GPUs, which you pay Nvidia for. And that's super expensive. But everything else that's sitting on top of that is getting faster, cheaper and better.
Andrew Keen: You also refer to a New York Times piece about how AI is prompting an evolution, not an extinction for coders. Your son's a coder, in a sense, you're a coder. Ultimately, one and I was at this thing with Tim Draper a couple of weeks ago where he was talking about companies, billion dollar companies built and managed by single people won't ultimately make most coders extinct. Maybe not all. But when founders like yourselves simply become coders and you won't have the need for other help.
Keith Teare: I make the point in an editorial that I didn't write a single line of code, but I've built a very complex system with lots of AI agents working together and delivering results for users. Learning to code is going to be a low requirement. A very high requirement is learning to architect and guide the AI because the AI can code, but it can't imagine systems to build or know when it got it right or when it got it wrong. The skill base is going to shift to what normally would be the domain of a product manager who has coding skills and can understand what's happening and can understand what it can ask for and what it can't ask for. But coding itself, learning Python, learning JavaScript or Java? Probably less essential.
Andrew Keen: So what happens to kids like your son who just graduated and now works in Silicon Valley as a coder?
Keith Teare: He'll still be needed for some time. In his company, they're not allowed to use AI yet. It's a little bit like dying skills always protect themselves until they can't. Engineers that are defensive or companies that are defensive about using AI are going to fall behind a little bit. But eventua...
