Quantifying Wealthspire's investment thesis: How upfront semiconductor cash realization decouples from lagged hyperscaler cloud monetization across 2024–2030 AI infrastructure spend.
| Chipmaker Margin \ Adoption | Low (Lagged) | Moderate (Base) | High (Accelerated) |
|---|
The model starts with six trillion dollars of assumed capital expenditure. Forty five percent silicon share produces two point seven trillion chipmaker revenue, and sixty five percent gross margin yields one point seven five five trillion gross profit. Bars encode these at one hundred pixels per trillion dollars. Revenue and gross profit are different accounting quantities; neither is investor return on capital. The seven annual allocation weights sum to one and distribute capex from twenty twenty four through twenty thirty. In the offline original, missing D3 throws during initial chart setup before the first update, leaving some starting captions and an empty chart. Native range changes reach updated metrics but throw before matrix and evidence-preview rendering. The explanation preserves that failure rather than claiming the original full visualization works. Setting the native capex range to eight trillion with all other defaults unchanged produces chip revenue three point six trillion and gross profit two point three four trillion. Bars compare original and changed revenue at one hundred pixels per trillion. Cloud revenue instead uses twelve times year index to power one point eight times an adoption multiplier, independent of capex, margin and depreciation. The default seven rounded annual cloud revenues sum to twelve hundred four billion. Subtracting that from thirty six hundred billion chip revenue gives divergence twenty three hundred ninety six billion. This gap compares two revenues, not profit or a causal allocation of value. Native JSON download works independently of chart rendering, while the matrix remains empty offline. Peak depreciation is cumulative assumed capex divided by hardware lifetime: six trillion over four years gives one point five trillion per year at the final sample. Bars encode the stock and annual expense at one hundred pixels per trillion, with different time units explicitly stated. Earlier vintages never expire from that cumulative depreciation formula. Payback tests cumulative cloud revenue minus cumulative capital expenditure rather than the displayed cloud revenue minus depreciation expression. Chip margin changes gross profit, but does not enter hyperscaler payback, so the sensitivity table margin rows repeat when rendering is available. The twenty thirty plus caption means no crossing inside the seven sample horizon, not a known eventual break-even date. These are assumed equations rather than verified forecasts, financial recommendations or exhaustive browser completion.