TLTD #35 - Best reads of 2025
A year of AI reads
Happy new year, everyone! I hope you had a good break and took some time to recharge, hang out with loved ones and maybe vibe code a few apps! Most people seem to be back at work this week so I guess the year has officially started! And boy, what a start to 2026 have we had! America, are you Ok?
Let’s cut to it.
In 2025 I read 21 books and while some were absolute duds (it happens…), five really made me think. Here are my top reads of 2025.
5. Transformed: Moving to the Product Operating Model
If AI-assisted software engineering accomplished anything in 2025 is to shift the focus away from purely technical discussions to solving real customer problems. That’s what Transformed is about: it challenges the feature factory model that still plagues many companies, replacing it with a framework where empowered product teams take ownership of outcomes rather than outputs.
This is very relevant in the context of enterprise AI adoption. The most advanced AI capabilities don’t matter if you don’t have a product operating model that focuses on customer value and business outcomes. Empowered products teams are key as we move from experimentation to ROI, and this book offers a path towards that goal.
4. The NVIDIA Way
Jensen Huang’s journey with NVIDIA is a lesson in patience and execution. Yes, NVIDIA managed to ride three massive waves — gaming, crypto, and generative AI — but what makes this book worth the read is the story of how NVIDIA positioned itself to catch each wave. As Seneca said: luck is when preparation meets opportunity.
luck is when preparation meets opportunity
There are valuable leadership lessons here. Huang’s ability to foster a culture of urgency while maintaining humility creates a tension that drives innovation. He’s famous for treating the company as if it’s “30 days from going out of business”, which might sound extreme, but it does captures the importance of staying sharp and nimble.
3. Empire of AI
Wow. Karen Hao’s inside look at OpenAI was a very interesting read, to say the least. Empire of AI adds a lot of context to how we got here technically, culturally and politically. The way the book frames Sam Altman’s leadership, and OpenAI’s pivot away from its nonprofit roots, raises some key questions about who actually controls this power technology, and what incentives drive that control.
massive [AI] data centres often require clean water for cooling, including in regions where clean water is already scarce
A couple of things stayed with me. One is the environmental cost of AI that we rarely talk about: for instance, those massive data centres often require clean water for cooling, including in regions where clean water is already scarce. Hardly an acceptable trade-off for memes-on-demand. The other is the human cost: the terrible conditions in data labeling companies that make modern AI possible. We’ve outsourced the bad parts of AI development to the most vulnerable people.
Building responsible AI will be hard if we don’t shed light into these ethical questions.
2. More Everything Forever
Adam Becker’s examination of technological maximalism hits hard and raises questions about the AI work we do every day. The central tension here — sacrificing immediate wellbeing for hypothetical future benefits at massive scale — is the operating principle behind much of Big Tech’s AI strategy, and it has real consequences for real people right now.
The future is not a permission slip for the present
This “AI will save billions of lives” narrative dominates tech news and, while AI has great potential for good, Becker makes you ask: whose suffering are we willing to accept today in service of an uncertain future? Accountability has never been more important: we need to be aware of the impact AI has on people’s lives today, not just its supposed future benefits.
The book doesn’t offer easy answers, but it asks the right questions.
1. The Skill Code
This is probably the most important book I read in 2025. I’ve written before about the impact of AI on software engineering (here and here) and Beane’s work across manufacturing, surgery, and tech gives us the empirical evidence of one of my biggest concerns: we’re breaking the apprenticeship model that has always been how expertise transfers from one generation to the next.
where will the next generation of senior engineers come from?
This book came out at a good time. In AI Writes Better Code Than You, I wrote that we may be creating a new cohort of engineers who can generate functional code but may never develop the deep systems thinking that comes from working alongside experienced practitioners and struggling through hard problems on their own — those are the lessons you never forget!
The book seems to support this idea: the surgeons who train with robots, the factory workers who monitor automated systems, the junior developers who rely on Copilots, they’re all experiencing the same disconnection from the expert-novice relationship we have normally relied on.
The question that comes to mind, then, is: if agentic coding is so effective that junior engineers never struggle through the hard problems, where will the next generation of senior engineers come from? This book should be required reading for anyone who cares about the future of our profession.
You may have seen some claim that this is overblown and that “we’ve been here beofre”. i.e.: unless you have a very good reason to, nobody allocates and manage their own memory pointers anymore.
I see it differently: while it’s true we no longer handle pointers ourselves, the underlying principles that make robust systems remain the same: decomposition, tradeoffs, debugging, reasoning about failure modes, etc…
2025 was a big year in technology. There was a lot of good, and a lot of not so good. I’m still optimist about where we’re going, but I feel some things need to change, and others need additional focus. These books helped me think more clearly about the trade-offs we’re making, often without realising it.
What were your most impactful reads from 2025? Drop them in the comments!