Coexistence Over Conquest: What Greg Graffin's Wars & Torments Teaches Us About AI Leadership
Summary
Drawing on Greg Graffin's book Population Wars, this article explores how his evolutionary-biology argument for coexistence over conquest reframes AI industry competition. I look at NVIDIA's "five-layer AI cake" as a real-world example of resource partitioning: no single company can own the whole stack, and the winners are the ones building ecosystems within their layer rather than trying to dominate every layer.
My Perspective
Graffin's central claim is that species and civilizations that endure do so through coexistence and resource partitioning, not by wiping out rivals. Watching the AI infrastructure industry organize around interdependent layers — energy, chips, infrastructure, models, applications — convinced me the same lesson applies. NVIDIA's partner-led approach to building an ecosystem, rather than trying to own every layer, is coexistence as strategy, not conquest.
I've been a Greg Graffin fan for a long time. His work as an artist made me curious and eager to learn, and the same is true of his life as an evolutionary biologist and professor. I've been learning from both sides of him ever since — the artist asking why, the scientist asking how. So when I found his book Population Wars, it made me think differently.
The book explores one of the world's longest-running narratives: that every species, population, and empire is locked in a zero-sum war for survival, with winners declared by who wipes out the rest. Graffin flips the script. Drawing from microbes to civilizations, he shows that long-term success isn't driven by extinction — it's driven by coexistence. Populations rarely vanish; they assimilate and adapt. The ones that endure carve out a niche, accommodate neighbors, and reach resource partitioning — a stable equilibrium where everyone has a seat at the table. As Graffin puts it, "the real hell of war is that you can never really win one."
That idea has stuck with me while watching the AI infrastructure industry organize itself in real time.
The AI Industry Looks More Like Graffin Than Conquest
The headlines sell conquest — one model, one winner, one company owning the stack. What I see with customers and partners is the opposite. As NVIDIA describes it, AI is a five-layer cake, and every layer depends on the others:
- Energy — the power that makes real-time intelligence possible
- Chips — the GPUs and accelerators that do the actual computation
- Infrastructure — networking, storage, and the systems that keep everything fed and running at scale
- Models — the foundation models and agents
- Applications — the end-user products and industry solutions that deliver real value
Remove any layer and the cake doesn't work. Try to own every layer yourself and you slow the entire industry, including your own progress. The companies succeeding are the ones that go deep in their layer and build a thriving ecosystem around it. That's not zero-sum competition. That's resource partitioning. That's coexistence as strategy.
NVIDIA and the Partner-Led Ecosystem
What stands out in my observations is how NVIDIA operates as a deeply partner-led organization. Rather than trying to control every part of the stack alone, it focuses on enabling the broader ecosystem with reference architectures and making CUDA the connective tissue that partners and customers build on. This approach turns the company into an essential platform that strengthens everyone around it.
The Coexistence Stack
Graffin's conclusion in the book is that populations that lasted weren't the loudest or strongest. They were the ones that figured out how to share the planet.
The AI industry is learning the same lesson. The five-layer cake exists because no single company can bake it alone. The players pretending otherwise are already losing ground to those who never tried.
The dots connect clearly. Graffin's evolutionary lens shows us that the most resilient systems thrive through cooperation, adaptation, and shared niches, not conquest. In AI, that means embracing the full stack, building strong partnerships, and measuring success by real output, like lowest cost per token and highest tokens per watt. The winning organizations will be those that learn early: coexistence is the strategy that delivers lasting value and sustainable progress.
Greg Graffin's music and lessons continue to teach us. I'm still taking notes, and the more I watch the tech industry unfold, the more I believe his insights belong at the center of how we build the future.
Related Articles
What a Family Seal Taught Me About Governing AI Agents
Drawing on a centuries-old family wax seal, this article explores what history's original approval gate teaches us about governing AI agents. I examine four failure modes of ungoverned multi agent work — source drift, role collapse, hidden context dependence, and candidate creep — and the harness that contains them: one canonical source, separated roles across separate models, a quarantine for new ideas, and a single human seal.
Shaping the Future with AI, Punk Rock, and Skateboarding Power
This article draws on punk rock and skateboarding culture — specifically the mentorship and camaraderie I've seen at my son's skatepark — to make the case that AI progress depends on collaboration, not competition. I use the NVIDIA/DDN partnership as a concrete example of an ecosystem built on shared knowledge rather than any one company trying to own the whole stack.
Reflecting on My Professional Journey: Decision Analysis & Lessons Learned
A look back at my career across Cisco, AWS, Stripe, MongoDB, and now DDN, framed through decision analysis — the idea that every decision is an irrevocable commitment of resources. I trace how that framework shaped partner ecosystem strategy, cloud migrations, and AI infrastructure launches, and connect it to Sundar Pichai's point that the human element only gets more important as AI takes on more of the work.