Atomic energy yes? I raise your fission and small modular reactors (SMRs) with fusion? If the timelines for SMRs to power datacentres were mismatched, you just wait…

I’ve been on somewhat of a “journey” with fusion. I wrote a few years ago:

“Best case scenarios from private companies are “within a decade”. But these timelines are at odds with publicly-funded projects from the EU, China and the UK coming in between 2035 and 2050. The reality is there will be a long lag between “net gain” and commercial electricity supply despite the huge demand. Even if net gain was achieved tomorrow, it will still take the best part of a decade to make a material impact on grid supply. For our purposes we do not need to distinguish between 2030 or 2040. We can put nuclear fusion on the 2030+ timeframe making it hard to see a pathway for VC funding.”

I’ll still out here calling shots.

To be clear, I was bullish on fusion generally, but specifically I thought the 15 year timeline for a electricity producing facility was sad for VCs. Especially for early-stage VCs.

It holds up well, despite recent funding by Pacific Fusion and Proxima Fusion and net gain at LLNL a few years ago. But I read somewhere that the top decile performers change their mind three times a year? So let’s see, am I a top decile performer?

My earlier assessment relied on two main assumptions:

1. The "first wall problem": Protecting the reactor from neutron damage is unsolved and, even if solved, would likely require a separate testing facility, which itself would take at least five years to build.
2. Cost: ITER’s estimated cost is $20–25 billion per GW, with private companies targeting $2–5 billion for 100–500 MW plants and Levelized Costs of Energy (LCOE) of $50–100/MWh. These costs would compete with advanced SMRs at $60–100/MWh (likely trending lower), gas with carbon capture at $70–120/MWh, and wind-plus-storage at $40–80/MWh.

So in theory, if we could find a way to solve the first wall problem and reduce costs, maybe we could speed up deployment time?

The big hope with my so-called research is to build epistemic infrastructure enabling unusual connections between disparate technologies to uncover novel investment opportunities. What’s striking, is how similar the analysis of market dynamics is for SMRs, quantum and fusion. There is no “ground truth”. Just the same trade-offs with proponents arguing that their particular trade-off will win. I will never interview someone who will give me the “truth”. I’ll just find different scientists making different trade-offs.

Scientist 1: “Getting to market first is most important because of path dependency."

Scientist 2: “No, we need a scalable system to win long-term”

Scientist 3: “No, the only thing that matters is cost for adoption, we need to focus on lower capex and opex to win long-term”

Choose your player…

• Multiple Technical Approaches:
  • Fusion: Tokamaks, stellarators, magnetic mirrors, z-pinch, inertial, magnetized target
  • SMRs: Light water, molten salt, high temperature gas, liquid metal, microreactors
  • Quantum: Superconducting, ion trap, photonic, neutral atom, topological, silicon spin
• Trade-offs:
  • Performance vs Complexity
  • Time to Market vs Ultimate Potential
  • Operating Costs vs Capital Costs
  • Scalability vs Initial Capability
  • Material Requirements vs Technical Maturity
• Market Dynamics:
  • Early leaders might not be long-term winners (short and long term bets)
  • Different approaches might serve different market niches (no ring to rule them all)
  • Supply chain and manufacturing capabilities matter as much as core technology (manufacturing not science risk)

The key insight is that, like in quantum computing, we might see different fusion/SMR approaches winning in different timeframes:

• Short/medium term (2025-2035):
  • SMRs: Light water designs like NuScale and BWRX-300, leveraging decades of PWR/BWR experience
  • Fusion: Simpler magnetic configurations like Helion's field-reversed configuration or Commonwealth's compact tokamak, which prioritize speed to market over ultimate performance
  • Quantum: Superconducting qubits and trapped ions, which are more mature but face scaling challenges