AcquiredComplexity Investing & Semiconductors with NZS Capital (Extended Cut)
CHAPTERS
- 0:29 – 2:07
Show setup: why NZS Capital + semiconductors deserved an extended cut
Ben and David introduce a special episode sparked by their TSMC research and an outstanding prior interview featuring NZS Capital’s Brinton Johns and Jon Bathgate. They explain that the conversation ended up focusing first on NZS’s investing frameworks before getting into semiconductors.
- •Origin story: using The Knowledge Project as a source for the Acquired TSMC episode
- •Meeting Brinton at Capital Camp and deciding to invite NZS on the show
- •Promise of new investing frameworks + later semiconductor deep dive
- 2:07 – 6:15
Sponsor interview: Banco Inter’s transformation into a Brazilian digital bank and super app
A sponsored segment features João Menin, CEO of Banco Inter, detailing how a traditional Brazilian bank pivoted to a fully digital, free checking account model. He describes rapid growth to 14M clients and the evolution into a broader super app by adding non-banking services.
- •Banco Inter’s legacy origins (1994) and shift to retail banking
- •100% digital, free, full-service account as the wedge vs. Brazil’s big five banks
- •Scaling from 0 to 14M clients and layering marketplace/insurance/investments
- •Super app strategy: adding services that ultimately connect back to payments/credit
- 6:15 – 12:24
NZS’s starting point: complexity theory and admitting you can’t predict the future
The hosts bring on Brinton and Jon and open with NZS’s foundational idea: markets are complex adaptive systems, which makes precise prediction unreliable. Brinton traces the firm’s intellectual roots to complexity research and the Santa Fe Institute.
- •Investing philosophy born from repeated forecasting pain and error
- •Complex adaptive systems → emergent behavior → prediction limits
- •Santa Fe Institute: physicists challenging economists’ overly precise models
- •Shift from traditional economics metaphors to biology as a better investing lens
- 12:24 – 14:30
Resilience as the goal: the ant colony lesson and founder-like time horizons
A key metaphor comes from Deborah Gordon’s ant research: colonies keep “idle” capacity to survive shocks. NZS maps this to companies, arguing Wall Street’s quarterly focus over-optimizes for near-term efficiency at the expense of long-term durability.
- •Ant colonies: ~half the ants appear inactive to preserve resilience
- •Companies often optimize for productivity; resilience requires slack/optionality
- •Founder-led businesses tend to align with long time horizons
- •CEO incentives and short tenure undermine long-run compounding
- 14:30 – 16:51
Slow compounding beats hypergrowth: durable growers and capital allocators (e.g., TI)
The discussion reframes investing away from flashy hypergrowth toward long-duration compounding. Jon highlights businesses that can grow moderately for decades, often led by management teams that act like capital allocators.
- •Hypergrowth can create fragility; durability and longevity matter more
- •Examples of decades-long compounders (e.g., Danaher, Texas Instruments)
- •Decentralization and pushing decisions down the org as a resilience feature
- •“Set it and forget it” mindset for the resilient portion of the portfolio
- 16:51 – 20:44
Defining NZS portfolio halves: resilience vs. optionality (and why ‘moats’ can be dangerous)
NZS explains its barbell structure: a concentrated resilient ‘head’ and a diversified optionality ‘tail.’ They also critique some ‘moat’ thinking when it leads to value extraction rather than value creation across stakeholders.
- •Resilient bucket traits: mission criticality, switching costs, scale, network effects
- •Power-law profit pools where 1–2 firms capture most economics (e.g., TSMC)
- •Skepticism of ‘moat = raise price forever’ framing
- •NZS = ‘non-zero sum’: win-win value creation for customers, partners, society
- 20:44 – 25:44
Non-zero-sumness in practice: Apple as a cautionary case, TSMC as the counterexample
They apply the non-zero-sum lens to real companies: Apple’s value capture draws regulatory and ecosystem backlash, while TSMC’s restraint (relative margins vs. customers) reinforces trust and durability. The key idea is that leaving surplus on the table can buy long-term resilience.
- •Platform backlash risk as a resilience risk (developers, regulators, partners)
- •Argument: small margin sacrifices can create massive ecosystem goodwill
- •TSMC’s customer-first culture and comparatively lower margins as strategy
- •Compounding math: most value shows up in the out-years of duration
- 25:44 – 31:15
Optionality and ‘Root MOS’: designing asymmetric upside into a public equities portfolio
NZS defines optionality as exposure to wide outcome ranges where a few winners drive returns. They introduce ‘Root MOS’—resilience with out-of-the-money optionality—illustrated by Amazon (pre-AWS visibility) and eBay with PayPal as the embedded option.
- •Broad vs. narrow predictions (‘multiverse’ framing) determine risk/positioning
- •Portfolio construction: ~15 resilient names concentrated; ~40 optionality names diversified
- •Root MOS concept: resilient base business + undervalued embedded option
- •Examples: Amazon IPO (AWS surprise), eBay + PayPal during marketplace uncertainty
- 31:15 – 40:06
Power laws, non-ergodicity, and why traditional risk models fail in the real world
The conversation shifts to distributions: many economic outcomes are power-law, not normal. Using Ole Peters’ non-ergodic coin-flip game, they show how ensemble averages can mislead individuals, and why “three-sigma events” happen far more than Gaussian models suggest.
- •Complex adaptive systems naturally produce power-law winners
- •Coin-flip example: positive expected value can still bankrupt most participants
- •Non-ergodic reality: time-average ≠ ensemble-average for individuals
- •Risk models based on normal distributions underestimate tail risk
- 40:06 – 47:32
Anti-conviction culture: documenting theses, calling out bias, and playing for slugging %
NZS challenges the investing fetish for ‘conviction,’ calling it a proxy for overconfidence and bias. They describe cultural practices—team critique, written theses, leverage points, and comfort with frequent failure in the optionality bucket—to improve decision-making.
- •Conviction = sunk-cost bias and overconfidence in a personal future view
- •Goal is to ‘maximize probability of getting lucky’ via many small bets
- •Team sport: peers identify each other’s biases; discomfort is a feature
- •Write everything down + revisit leverage points to detect thesis drift
- 47:32 – 53:51
Managing the optionality tail: trimming winners, crossing into resilience, and valuation as a ‘forced prediction’
They explain how optional positions are managed as outcomes narrow: positions may be trimmed as they grow, or promoted into the resilient head when the business becomes structurally safer. Valuation is central—high multiples force narrower predictions and increase fragility.
- •Optionality winners can become oversized; NZS caps/trim to control drawdown risk
- •Promotion process: explicitly move companies into resilient bucket and ‘average up’
- •Avoid letting highly volatile names become large without resilience proof
- •Expensive valuations increase required certainty (‘forced prediction’)
- 53:51 – 1:15:27
Why semiconductors now: ‘ten out of ten multiverses’ need chips, and where NZS invests across the stack
Ben pivots to semiconductors as foundational ‘oxygen’ for the information age, which NZS sees as a resilience anchor in an expensive market. They outline positions across catalog semis, leading-edge digital, equipment, EDA, and emerging materials like silicon carbide.
- •Macro: high valuations everywhere; semis as core building blocks with durable demand
- •Two semiconductor worlds: leading-edge logic vs. catalog/long-life analog & MCUs
- •NZS portfolio exposure: ~⅓ in semis across foundry, equipment, EDA, designers
- •Optionality example: Cree and silicon carbide as a potential platform shift
- 1:15:27 – 1:28:12
TSMC vs. Samsung, vertical integration tradeoffs, and TSMC’s geopolitical ‘company-ending’ risk
They differentiate Samsung’s strength in memory (DRAM/NAND) from TSMC’s neutral foundry leadership in logic. The group discusses why vertical integration is hard in semis (Intel’s multi-front battle) and how TSMC’s centrality makes Taiwan geopolitics both critical and hard to hedge.
- •Samsung: dominant memory producer; foundry smaller and conflicted with customers
- •TSMC: trusted neutral partner; conflict-free model builds customer dependence
- •Vertical integration in semis is brutally difficult (process + design competition)
- •Geopolitical risk: TSMC’s shutdown would set back global tech and GDP for years
- 1:28:12 – 1:36:16
Moore’s Law today: EUV runway, chiplets, advanced packaging, and system-level scaling
Jon argues that even if ‘literal’ Moore’s Law is debated, the spirit continues through the mid-2030s. Progress now comes from multiple layers—EUV lithography, new transistor architectures, chiplets, packaging, and cluster-level interconnects—turning chip firms into full computing systems companies.
- •EUV as an enabler with ~10–15+ years of effective runway (per ASML framing)
- •Beyond transistor shrink: ‘more than Moore’ via packaging and chiplets
- •Gate-all-around transition after FinFET as another architectural lever
- •System-level performance: stitching processors into clusters with proprietary interconnects
- 1:36:16 – 1:59:57
Ecosystem orchestration: TSMC’s Open Innovation Platform, ASML/Trumpf/Zeiss ‘fractal’ dependencies, and why growth governors create durability
They explore why no single firm can vertically integrate the whole chain: each layer adds unique, hard-won capability, reinforced by alliances and learning cycles. The episode closes by linking semiconductor constraints (capacity, complexity) to NZS’s broader thesis: growth governors and shared value extend duration and maximize long-term compounding.
- •TSMC’s open alliance approach vs. Intel’s historically closed system
- •ASML’s EUV stack depends on deep partners (Trumpf lasers, Zeiss optics) and years of learning
- •Industry coordination example: Intel/TSMC/Samsung buying 25% of ASML to fund EUV
- •‘Negative feedback loops’/capacity constraints as healthy governors enabling decades of compounding
