Investors reflexively look for what's new. The bigger, more persistently underestimated opportunity is usually what's being replaced — an existing installed base swapped for a better version, on a timeline nobody prices correctly because replacement demand rarely announces itself the way a new market does.
Every growth narrative this publication has covered — AI, energy transition, reindustrialisation — eventually runs through the same underlying mechanism: an existing installed base of assets gets replaced by a better one. This letter argues that mechanism deserves its own framework, separate from whichever sector happens to be doing the replacing this decade.
A genuinely new market announces itself — nobody had a smartphone before 2007, so every unit sold was incremental demand, easy to size and easy to get excited about. Replacement demand is different: it transforms an existing market rather than creating a new one, so the same dollar of revenue looks like cannibalisation of an old category rather than growth of a new one, and investors systematically discount it as a result. This is the single most persistent mispricing this letter has identified in eighteen months of coverage across this publication's archive.
The 2026 EV transition is the cleanest live example. BloombergNEF's Electric Vehicle Outlook 2026 projects that 27% of cars sold globally this year will be electric, up from just 9% five years ago, rising to 52% of all cars sold by 2035. That is unambiguous, well-covered growth. But BNEF's own Head of Intelligent Mobility, Andrew Grant, points to the more important and less-covered half of the story directly: even with record sales, existing combustion vehicles simply aren't leaving the road fast enough to matter for net-zero targets — in the US specifically, only 24% of the fleet is projected to be electric even by 2040. The replacement cycle, not the sales cycle, is the actual constraint, and it is measured in vehicle lifespans (12-15 years), not sales quarters.
Installed base — how many units of the old asset exist right now, and how is that number distributed across geographies with different replacement incentives. Asset life — the physical or economic lifespan of the thing being replaced; a car (12-15 years) replaces on a fundamentally different clock than a data center server (3-5 years) or a coal plant (40+ years). Replacement cost — not just the sticker price of the new asset, but the full system cost; BNEF's own numbers show EVs still carry a 17% price premium over combustion equivalents in major European markets (down from 34% in 2024, but still a real gate). Adoption curve shape — replacement cycles are rarely linear; they cluster around subsidy deadlines, regulatory mandates, and the point where total cost of ownership crosses over, which is precisely why BNEF has now cut its long-term EV adoption forecast for two consecutive years even as near-term sales hit records. Winners in transition — the businesses that profit are frequently not the asset manufacturer but the infrastructure layer the replacement requires.
That fifth variable is where the money hides. BNEF projects the global grid will require over $800 billion in investment by 2040 purely to incorporate EV electricity demand — a cost that has nothing to do with who wins the car-manufacturing race and everything to do with the physical reality that every combustion vehicle replaced by an electric one shifts energy demand from a liquid-fuel supply chain to an electrical one that was not sized for it.
Replacement cycles create their real money not in the asset being swapped in, but in the infrastructure layer that has to absorb the swap — the grid upgrade, the charging network, the software migration — which is systematically the less-covered, less-obvious side of every transition this letter has examined.
ICE vehicles to EVs is simply the most visible current example of a mechanism running in parallel across the rest of the economy this publication already covers in isolation: ageing copper grids being upgraded for electrification and AI power demand (this publication's own Letter 28), coal plants retiring into gas and renewables, legacy enterprise software migrating to cloud architectures, cash giving way to digital payment rails, and — the replacement cycle with the largest single number attached to it — human labour being replaced by AI agents across knowledge work. Each of these is usually analysed as its own sector story. The framework this letter proposes is to analyse them as instances of the same underlying mechanism, because the investment lesson (installed base, asset life, replacement cost, adoption curve, infrastructure winners) transfers directly from one to the next.
The single most underpriced number in this letter is not an EV statistic — it is the observation that BNEF has downgraded its long-term EV adoption forecast for two consecutive years while short-term sales keep setting records, evidence that the replacement cycle and the sales cycle are genuinely different variables that the market keeps conflating. The framework this letter proposes — installed base, asset life, replacement cost, adoption curve, infrastructure winners — applies with almost no modification to every other replacement cycle this publication has covered sector-by-sector. The risk to this framework is not that replacement stops happening; it is that investors keep pricing the visible new-unit sales number and continue missing the slower, larger, infrastructure-layer number sitting underneath it.
Founder, NextGen Economics · Bangalore, India · July 2026
Sources: BloombergNEF, Electric Vehicle Outlook 2026 (June 2026) · Global Fleet (BNEF EVO 2026 analysis) · IEA, Global EV Outlook 2026. Builds on and generalises this publication's own Letter 28 (The Grid).
Not investment advice. This letter proposes a cross-sector investment framework; it does not constitute a recommendation regarding any security.