Probability-Mass Flow (Compute scale)

Each horizontal band is a scenario branch; band height = scenario prior probability (taller = more likely). Predictions linked to a scenario sit inside that band at their expected resolution date. Visual area answers "what is likely to happen and when."

Compute scale (cumulative)Energy / grid (cumulative)Humanoid deployment (cumulative)Robotaxi (cumulative)AGIASI$1T+ IPOMars uncrewedAI pauseRecession
COMPUTE_1GW_2027prior 60%2027-06241_020 · Energy · prob 49% · resolves 2026-06 · 1 GW of power = ~$50 billion in hardware/software/data centers241_044 · Energy · prob 52% · resolves 2026-06 · Rate of data center construction will continue accelerating with government supportCMQ_057 · AI · prob 56% · resolves 2026-08 · Democratization of sophisticated AI tools means any 'kid in a basement' can now build world-changingINF_014 · Energy · prob 42% · resolves 2026-10 · Data-center industry could emit up to 2.5 billion tons of CO2 cumulatively through 2030 if powered p240_034 · Energy · prob 47% · resolves 2026-11 · Morgan Stanley forecasts 13-40GW data center power shortfall through 2028COD_AI_001 · Technology · prob 51% · resolves 2026-12 · Vera Rubin becomes commercially available across major cloud and OEM partners by H2 2026COD_AI_002 · AI/Compute · prob 40% · resolves 2027-03 · At least one 1 GW Vera Rubin frontier-training cluster reaches deployment by Q1 2027INF_007 · AI · prob 43% · resolves 2027-09 · OpenAI will dominate global AI data-center capacity by 2027 — centralization of compute power that pINF_017 · Energy · prob 48% · resolves 2027-11 · xAI Colossus Phase 2 will consume approximately 1.5 gigawatts of power — city-scale demand for a sinCMQ_014 · AI/Compute · prob 50% · resolves 2027-12 · Physical compute scaling will proceed at ~0.5 OOMs per year through 2027, driven by exponentially laAI_018 · Energy · prob 49% · resolves 2028-11 · Global data center construction spend will reach approximately $2.9 trillion through 2028 — early adINF_012 · Energy · prob 46% · resolves 2028-11 · AI data-center power consumption will grow by approximately 126 GW annually through 2028 — comparablINF_010 · Energy · prob 56% · resolves 2028-11 · US data-center demand will reach 74 GW by 2028 against a projected ~49 GW power-access shortfall — sSEM_005 · AI/Infrastructure · prob 72% · resolves 2028-12 · Stargate is a $500 billion multiyear capex program for distributed AI data-center construction (2025238_046 · Energy · prob 44% · resolves 2029-10 · xAI will build every new data center at 1.2 gigawatts scaleINF_048 · Energy · prob 48% · resolves 2030-11 · Global data-center energy demand will reach approximately 945 Terawatt-hours by 2030 — a roughly 175SEM_001 · AI/Compute · prob 55% · resolves 2030-12 · Compute clusters will scale from $10B → $100B → $1T by end of decade; 'another zero every six months231_030 · Energy · prob 49% · resolves 2031-10 · US AI industry needs 80 gigawatts of power in the next 3-5 years (Eric Schmidt cited).246_041 · Energy · prob 38% · resolves 2031-11 · Data centers will drive local electricity costs toward zero within ~5 years.INF_020 · Space · prob 26% · resolves 2035-10 · In-orbit data-center market will reach approximately $1.77 billion by 2029 and grow to approximatelyINF_006 · AI · prob 40% · resolves 2040-11 · Over time, very large fractions of the earth's land surface will be given over to data centers as AI21 claimsCOMPUTE_10GW_2028prior 40%2028-12241_020 · Energy · prob 49% · resolves 2026-06 · 1 GW of power = ~$50 billion in hardware/software/data centers241_016 · Energy · prob 51% · resolves 2026-10 · 92 GW power shortage in America between now and 2030238_047 · Energy · prob 45% · resolves 2026-10 · US will add 86 GW of new grid capacity in 2026, 51% from solarINF_014 · Energy · prob 42% · resolves 2026-10 · Data-center industry could emit up to 2.5 billion tons of CO2 cumulatively through 2030 if powered p241_017 · Energy · prob 50% · resolves 2026-11 · Equivalent of ~60 nuclear plants needed by 2030 but essentially zero being built240_034 · Energy · prob 47% · resolves 2026-11 · Morgan Stanley forecasts 13-40GW data center power shortfall through 2028235_022 · Energy · prob 53% · resolves 2026-11 · US will add a record 86 gigawatts of utility-scale capacity this coming year.CMQ_017 · Energy · prob 52% · resolves 2026-11 · US electricity production must grow by 'tens of percent' by end of decade to support hundreds of milCMQ_024 · AI/Compute · prob 54% · resolves 2026-12 · 'Tokens per Watt' is the defining Key Performance Indicator (KPI) for the modern digital economy.COD_AI_002 · AI/Compute · prob 40% · resolves 2027-03 · At least one 1 GW Vera Rubin frontier-training cluster reaches deployment by Q1 2027INF_017 · Energy · prob 48% · resolves 2027-11 · xAI Colossus Phase 2 will consume approximately 1.5 gigawatts of power — city-scale demand for a sinCMQ_040 · Semis/Memory · prob 62% · resolves 2027-12 · AI-driven demand has triggered an 'unprecedented supercycle' for dominant global memory suppliers (S241_023 · Energy · prob 49% · resolves 2028-06 · 10% of US electricity will be used by data centers246_036 · AI · prob 46% · resolves 2028-09 · Terafab will deliver 1 terawatt/year AI compute, 50x current global output of 20 gigawatt.INF_010 · Energy · prob 56% · resolves 2028-11 · US data-center demand will reach 74 GW by 2028 against a projected ~49 GW power-access shortfall — sINF_012 · Energy · prob 46% · resolves 2028-11 · AI data-center power consumption will grow by approximately 126 GW annually through 2028 — comparabl238_066 · Energy · prob 41% · resolves 2029-10 · Pete Donovan aspires to build 20,000 distributed 10MW data centers across US high-school-football to238_046 · Energy · prob 44% · resolves 2029-10 · xAI will build every new data center at 1.2 gigawatts scale232_045 · Energy · prob 48% · resolves 2030-06 · Tesla and SpaceX each aim to generate 100 gigawatts of solar per year.INF_056 · Energy · prob 46% · resolves 2030-10 · Helion will supply OpenAI with 5 gigawatts of fusion electricity by 2030 (roughly 100 Helion 50-MW rINF_048 · Energy · prob 48% · resolves 2030-11 · Global data-center energy demand will reach approximately 945 Terawatt-hours by 2030 — a roughly 175SEM_001 · AI/Compute · prob 55% · resolves 2030-12 · Compute clusters will scale from $10B → $100B → $1T by end of decade; 'another zero every six months231_030 · Energy · prob 49% · resolves 2031-10 · US AI industry needs 80 gigawatts of power in the next 3-5 years (Eric Schmidt cited).240_035 · Energy · prob 46% · resolves 2035-11 · Meta secured 6.6 GW of clean nuclear power for 2035INF_057 · Energy · prob 43% · resolves 2035-11 · Helion will scale to 50 gigawatts of fusion-generated power to OpenAI by 2035 — equivalent to more tINF_006 · AI · prob 40% · resolves 2040-11 · Over time, very large fractions of the earth's land surface will be given over to data centers as AI26 claimsCOMPUTE_100GW_2030prior 20%2030-12241_020 · Energy · prob 49% · resolves 2026-06 · 1 GW of power = ~$50 billion in hardware/software/data centers241_044 · Energy · prob 52% · resolves 2026-06 · Rate of data center construction will continue accelerating with government supportSEM_002 · AI · prob 56% · resolves 2026-09 · By 2025-2026, AI model outputs will outpace the cognitive capabilities of college graduates (driven INF_014 · Energy · prob 42% · resolves 2026-10 · Data-center industry could emit up to 2.5 billion tons of CO2 cumulatively through 2030 if powered pCMQ_017 · Energy · prob 52% · resolves 2026-11 · US electricity production must grow by 'tens of percent' by end of decade to support hundreds of mil241_017 · Energy · prob 50% · resolves 2026-11 · Equivalent of ~60 nuclear plants needed by 2030 but essentially zero being builtCMQ_045 · Semis · prob 43% · resolves 2026-12 · Agentic AI will create additional CPU market space of $10B-$100B by 2030, potentially driving total CMQ_030 · AI/Compute · prob 34% · resolves 2026-12 · In the modern AI pipeline, the CPU no longer merely supports the model — it drives the model (agentiCMQ_042 · AI/Compute · prob 35% · resolves 2026-12 · As AI evolves from Generative to Agentic, structural computing bottlenecks shift away from GPU and hCMQ_015 · AI · prob 50% · resolves 2027-08 · Algorithmic efficiencies will deliver ~0.5 OOMs per year of additional effective compute through 202INF_007 · AI · prob 43% · resolves 2027-09 · OpenAI will dominate global AI data-center capacity by 2027 — centralization of compute power that p240_017 · Energy · prob 35% · resolves 2027-11 · Optimization may eliminate need to tile the world with data centers or massive energy buildoutAUT_009 · AI · prob 22% · resolves 2027-11 · 'Powerful AI' (functional AGI) arrives 2026-2027 — data centers house a 'nation of geniuses' consistINF_001 · AI · prob 48% · resolves 2027-11 · Reaching AGI by 2027 will require deployment of hundreds of millions of AI GPUs — forcing total mobiCMQ_014 · AI/Compute · prob 50% · resolves 2027-12 · Physical compute scaling will proceed at ~0.5 OOMs per year through 2027, driven by exponentially la246_036 · AI · prob 46% · resolves 2028-09 · Terafab will deliver 1 terawatt/year AI compute, 50x current global output of 20 gigawatt.AI_018 · Energy · prob 49% · resolves 2028-11 · Global data center construction spend will reach approximately $2.9 trillion through 2028 — early adINF_012 · Energy · prob 46% · resolves 2028-11 · AI data-center power consumption will grow by approximately 126 GW annually through 2028 — comparablINF_010 · Energy · prob 56% · resolves 2028-11 · US data-center demand will reach 74 GW by 2028 against a projected ~49 GW power-access shortfall — sINF_002 · Geopolitics · prob 30% · resolves 2028-11 · By 2027-2028, the US national-security apparatus will effectively appropriate frontier AI data centeSEM_047 · AI/Hardware · prob 71% · resolves 2028-12 · At 200,000-GPU scale, orchestration becomes a literal 'battle against entropy' — single cosmic-ray-fAI_009 · AI · prob 49% · resolves 2029-08 · Silicon Valley boardroom compute-cluster conversations have escalated rapidly from $10 billion clust238_046 · Energy · prob 44% · resolves 2029-10 · xAI will build every new data center at 1.2 gigawatts scale235_023 · Energy · prob 48% · resolves 2029-11 · Eric Schmidt said AI will require 100 gigawatts by 2029, called it a crisis.INF_021 · Energy · prob 47% · resolves 2030-11 · AI data centers are the 'steel mills of the 21st century' — humanity will miss near-term climate-chaSEM_001 · AI/Compute · prob 55% · resolves 2030-12 · Compute clusters will scale from $10B → $100B → $1T by end of decade; 'another zero every six monthsINF_003 · Macro/Economy · prob 43% · resolves 2030-12 · The global AI infrastructure buildout is in its earliest phase — only a few hundred billion of a triCMQ_044 · AI/Compute · prob 65% · resolves 2030-12 · Future data-center architectures optimized for agentic workflows may require 1:2 or even 2:1 CPU-to-241_021 · Markets/Stocks · prob 47% · resolves 2031-09 · America can raise $5 trillion over 5 years for data center/AI buildout231_030 · Energy · prob 49% · resolves 2031-10 · US AI industry needs 80 gigawatts of power in the next 3-5 years (Eric Schmidt cited).FUT_018 · Energy · prob 39% · resolves 2031-11 · Rise of 'Energy Islanding' 2026-2031 — sovereign nations possessing BOTH geographic solar irradianceINF_050 · Energy · prob 39% · resolves 2032-11 · Within approximately 6-7 years (2031-2032), major tech companies will operate their own small modulaCMQ_025 · AI/Compute · prob 49% · resolves 2032-12 · The entire global installed base of data centers must be ripped out and replaced — legacy DCs unfit INF_044 · Energy · prob 48% · resolves 2035-11 · US total energy consumption will rise by approximately 10% over the next decade — the largest sustai240_035 · Energy · prob 46% · resolves 2035-11 · Meta secured 6.6 GW of clean nuclear power for 2035CMQ_063 · Quantum/AI · prob 15% · resolves 2035-12 · Quantum compute integration with AI optimization algorithms and material-science discovery could draSEM_009 · Economy · prob 30% · resolves 2035-12 · The AI boom rooted in semiconductor hardware will generate the world's first individual trillionaireINF_006 · AI · prob 40% · resolves 2040-11 · Over time, very large fractions of the earth's land surface will be given over to data centers as AI38 claimsCOMPUTE_STARGATE_FAILUREprior 15%2029-12241_016 · Energy · prob 51% · resolves 2026-10 · 92 GW power shortage in America between now and 2030240_034 · Energy · prob 47% · resolves 2026-11 · Morgan Stanley forecasts 13-40GW data center power shortfall through 2028INF_017 · Energy · prob 48% · resolves 2027-11 · xAI Colossus Phase 2 will consume approximately 1.5 gigawatts of power — city-scale demand for a sinSEM_005 · AI/Infrastructure · prob 72% · resolves 2028-12 · Stargate is a $500 billion multiyear capex program for distributed AI data-center construction (2025FUT_012 · Energy · prob 45% · resolves 2031-11 · Two-thirds of world's nations reaching tipping point 2026-2031 where building new renewable capacityFUT_025 · Energy · prob 52% · resolves 2031-11 · Runaway economic feedback loop 2026-2031: cheaper renewable energy incentivizes wider adoption, incrINF_006 · AI · prob 40% · resolves 2040-11 · Over time, very large fractions of the earth's land surface will be given over to data centers as AI7 claimstoday202620272028202920302031

How to read this

  • Tall bands are high-prior futures — the network thinks they're most likely.
  • Dots inside a band are the tracked claims linked to that scenario. Click any dot to drill in.
  • Dashed colored vertical line = the scenario's own resolution date. Scenarios with target_window_end past the right edge resolve "beyond horizon."
  • Claims appearing in multiple bands (Phase 2A's cosine-linked correlate edges often link a prediction to several scenarios) just render in each band they touch — a hint the claim is scenario-robust.