Deflationary Cascade
The simultaneous collapse in cost across energy, intelligence, computation, biology, and production. Unlike any previous revolution (which deflated ONE resource), this one deflates everything at once. Phase shift, not incremental improvement.
The Data
[EVIDENCE]
Every sector where technology is the primary input follows exponential learning curves -- Wright's Law or analogous experience curves. The consistency is the point: these are not isolated anomalies but a unified phenomenon.
| Technology | Start | End | Decline | Approximate rate |
|---|---|---|---|---|
| Solar PV ($/W) | $76.67 (1977) | $0.24 (2024) | 99.7% | ~10%/year |
| Li-ion batteries ($/kWh) | $1,100 (2010) | $108 (2025) | 90% | ~15%/year |
| Genome sequencing | $95M (2001) | $200 (2022) | 99.9998% | ~40%/year |
| Compute ($/GFLOPS) | $18.75M (1984) | $0.03 (2017) | >99.9999% | OOM/5 years |
| Digital storage ($/GB) | $193,000 (1980) | $0.014 (2022) | >99.9999% | OOM/4 years |
| LED lighting ($/klm) | ~$90 (2008) | ~$1-3 (2020) | ~97% | ~25%/year |
| Internet transit ($/Mbps) | $1,200 (1998) | $0.50 (2020) | 99.96% | ~25%/year |
| Satellite launch ($/kg LEO) | $54,500 (Shuttle) | $2,720 (Falcon 9) | 95% | ~5%/year |
| AI inference | $20/M tokens | $0.07/M tokens | 99.65% | ~50x/year median |
China-manufactured solar modules now approach $0.10/watt. BloombergNEF reports the lowest observed battery pack price at $50/kWh (LFP chemistry in China). Genome sequencing outpaced Moore's Law by an order of magnitude after next-generation sequencing was adopted in 2008. The BLS Producer Price Index for prepackaged software declined ~74% from 1997 to 2022.
The Cascade Order
[CONVICTION]
The deflation follows a predictable sequence: software first, then services, then physical goods, then energy. Software and compute are already effectively free at scale. Energy is on the same trajectory -- utility-scale solar LCOE fell ~90% from ~$360/MWh in 2010 to ~$30-50/MWh in 2022.
The critical frontier: whether AI pushes these deflation curves into services and knowledge work. The early evidence says yes. The empirical-case-for-abundance documents that AI is automating knowledge work first -- reversing every previous automation pattern. Computer programmer employment fell 27.5% in two years (2023-2025). The St. Louis Federal Reserve found a 0.57 correlation between AI adoption intensity and unemployment increases. Physical labor remains resilient: construction added 190,000 jobs in 2024.
This cascade order creates asymmetric national vulnerability. The US (77.8% services) and UK (73%) face maximum exposure to AI-driven service deflation. China (36.5% industry) and South Korea (32%) maintain structural buffers. The most AI-capable nations are the most structurally exposed -- a paradox the empirical-case-for-abundance quantifies in detail.
Why This Time Is Different
[CONVICTION]
Every previous revolution deflated one domain while scarcity persisted in others. Steam deflated mechanical labor. Electricity deflated distance. The internet deflated information cost. This one deflates energy, intelligence, computation, and eventually production ALL AT ONCE.
When everything deflates simultaneously, the entire economic architecture (designed to manage scarcity) hits a structural mismatch with the reality it's supposed to coordinate. GDP registers this as stagnation or deflation rather than progress -- a measurement failure the lossy-compression framework predicts. The "invisible economy" of ecosystem services, unpaid care, open source software, and digital consumer surplus likely exceeds $150 trillion annually, none of which GDP captures.
Capital Is Responding
[EVIDENCE]
The market already prices in the cascade. Hyperscaler capex exploded from $24 billion (2015) to $211 billion (2024), with projections of $315-443 billion for 2025. 75% flows to physical AI infrastructure -- data centers, GPUs, power systems. Global clean energy investment crossed $2.2 trillion in 2025, doubling from 2015, at a 2:1 ratio over fossil fuel investment. Infrastructure fundraising surpassed real estate fundraising in 2024 for the first time.
Meanwhile, traditional software valuations collapse: public SaaS multiples fell from 18-19x EV/Revenue (2021) to 5.1x (2025) -- a ~70% compression. The market message is clear: the future is physical, and asset-light software generating outsized returns is narrowing to AI or ending altogether.
The Novel Synthesis
No unified "deflationary cascade" model exists in the literature. Individual curves are well-documented (Wright's Law, experience curves). The synthesis -- that simultaneous deflation across domains is qualitatively different from sequential -- is original to the Mesocosm thesis.
The empire-collapse-pattern shows what happens when abundance pressure meets scarcity-based institutions. The money-as-scarcity-tool explains why the coordination mechanism breaks. The cascade is the forcing function that makes both inevitable.
Related
- lossy-compression -- why the current economic architecture cannot handle abundance
- verification-infrastructure -- what enables value to flow in the new regime
- money-as-scarcity-tool -- why the coordination tool breaks when abundance arrives
- empirical-case-for-abundance -- the hard-data backbone for the cascade thesis
- empire-collapse-pattern -- historical pattern of scarcity tools failing under abundance
- economics -- domain overview
- 15-the-deflationary-cascade -- chapter treatment