Energy mix in 2086

What powers the world when there are 2 billion people, an AGI data-centre fleet, and a small footprint on Mars.

A lightly edited transcript of a conversation with Claude, the large language model developed by Anthropic. Four turns. The first sets up the baseload question against a specific 2086 worldbuilding scenario; the second forces an account of fossil fuels that the first answer skipped; the third returns to storage and tidal — categories the previous answers underweighted, with Australia as the lead case; the fourth pushes into the esoteric energy sources, including micro black hole power in the Lachlan-Trilogy mode. My questions are quoted; Claude's responses follow as prose, with internal sub-headings preserved.

Lars

Let's talk about what sort of power sources we can reasonably extrapolate to provide the bulk of baseload power by 2086. Let's factor in three things: a population of around 2 billion, a large network of conventional data centres that provide AGI capabilities, and a reasonable presence in space — including the Moon, plus a small outpost on Mars.

Claude

The 2-billion population assumption changes the answer significantly, so worth starting there. A 2086 with 2 billion people implies the demographic collapse the romance and AGI pieces projected has actually happened — current 8B trajectory has reversed sharply rather than plateaued at 10-11B. This is plausible on the timescale and fits the rest of the worldbuilding, but it has large implications for the energy question that aren't obvious at first.

Lower population doesn't mean proportionally lower total energy demand. Several factors push the per-capita demand up dramatically: complete electrification of transport, heating, industry, and agriculture; AI infrastructure that didn't exist in 2024; high automation handling the work the missing 6 billion would have done; space operations; climate adaptation infrastructure that doesn't scale with population. The remaining 2 billion plausibly consume per-capita electricity at 2-3x current developed-world levels.

Rough estimate: global electricity demand in 2086 is somewhere between 60,000 and 90,000 TWh/year, against current ~30,000 TWh/year. The major new demand center is the AGI data infrastructure, which could be 15-25% of total. Industrial automation for 2 billion at high consumption is another 25-35%. Residential and transport, fully electrified, perhaps 30-40%. Space-related terrestrial loads (launch facilities, ground stations, manufacturing for orbital industries) maybe 5-10%. The total is significantly higher than today despite the smaller population, because the world is doing more per person and doing all of it electrically.

This is the demand the generation mix has to meet, with the additional constraint that data centers and industrial automation need genuinely firm power, not intermittent renewables with statistical assurances about availability.

What actually provides baseload in 2086

The honest projection requires walking through each candidate against its own technical and political constraints.

Nuclear fission is the high-probability major contributor. By 2086, Generation IV reactors and SMRs (Small Modular Reactors) are mature technologies in their third or fourth deployment generation. The current SMR designs being demonstrated now (NuScale, Rolls-Royce, KLP-2 in Russia, ACP100 in China, BWRX-300) represent the first commercial generation. By 2050 these are factory-produced at scale; by 2070 they're a commodity; by 2086 they're the default for medium-scale industrial and data-center power and a major contributor to grid baseload.

Several things favor nuclear in the 2086 picture specifically. The 2-billion population reduces the political friction around siting: there's less not in my backyard when there are fewer backyards. The collapse of the original anti-nuclear movement is largely demographic: its core constituency was the Baby Boom, and by 2086 that constituency is gone. The replacement generations have grown up with climate change as the dominant frame and with the failure of pure-renewables strategies in several major economies as a cautionary tale. The political economy of nuclear in 2086 looks more like the political economy of nuclear in 1970 than in 2010.

Fast-spectrum reactors and closed fuel cycles solve the fuel-supply and waste problems that constrained 20th-century nuclear. Thorium and uranium-238 utilization extends fuel reserves to thousands of years. Spent-fuel recycling produces a much smaller waste stream with much shorter hazard duration. The Russian BN-800 fast reactor has been operating commercially since 2016; by 2086 this technology is standard. The waste problem in 2086 is technical and manageable rather than the existential question it was treated as in 2024.

I'd project nuclear fission at 25-35% of total generation in 2086, with significantly higher penetration in jurisdictions that committed early (France, possibly Korea, China, India, the UAE, and depending on political trajectories the US and parts of the Nordics).

Nuclear fusion is the wildcard and the biggest source of uncertainty in the 2086 projection. The current state has more credibility than any prior fusion-development cycle: the National Ignition Facility achieved ignition in December 2022; ITER is producing first plasma in the 2030s; multiple private programs (Commonwealth Fusion, Helion, TAE, Tokamak Energy, First Light, Type One Energy, and others) are pursuing different approaches with serious capital backing.

The optimistic case is that commercial fusion plants come online in the 2040s, scale through the 2050s, and provide a significant fraction of new generation by the 2060s. By 2086 in this scenario, fusion is 20-30% of global electricity and growing. The pessimistic case is that fusion remains an engineering challenge for another two cycles: commercial plants in the 2060s, scaling through the 2070s, modest deployment by 2086. In this scenario fusion is 5-10% of generation by 2086 and is the dominant new technology of the early 22nd century rather than the late 21st.

I'd weight these scenarios as roughly equal, which means the realistic projection is fusion contributing somewhere between 5% and 30% of 2086 electricity, with the actual figure depending heavily on engineering progress in the 2030s. Worth noting: even in the optimistic case, fusion in 2086 is mature for the first time, which means the technology improvement curve is steep through the rest of the century. The fusion-dominant grid is the 22nd-century story, not the late-21st.

Solar continues to scale. The cost curve is unrelenting: solar PV has been the cheapest new generation source globally since the early 2020s and continues to get cheaper. By 2086, solar dominates installed generation capacity, possibly providing 30-40% of total energy delivered, but constrained as baseload by its diurnal cycle.

The interesting solar development for 2086 is space-based solar power. Microwave power transmission from geostationary or sun-synchronous solar arrays to ground receivers has been studied since the 1970s. The economics have been blocked by launch costs. With Starship-class reusable systems and beyond, launch costs fall to levels where SBSP becomes economically competitive. By 2086, with established space infrastructure as you've assumed, space-based solar is plausibly contributing 5-15% of global power, providing genuine 24/7 generation that complements the diurnal terrestrial solar. The first major SBSP installations probably come online in the 2050s; significant deployment in the 2060s and 2070s. China has been the most aggressive investor in this technology in the 2020s and is likely to be the first major operator.

Wind continues to scale, particularly offshore and floating designs. Mature technology by 2086, contributing maybe 10-15% of generation, with the deepest deployment in coastal regions and at higher latitudes. Same intermittency caveat as solar.

Geothermal is the under-discussed contributor that probably matters more than current models suggest. Enhanced Geothermal Systems (EGS) (drilling deep wells, fracturing hot dry rock, circulating fluid to extract heat) are scaling now. The drilling technology transfers directly from the oil and gas industry, which has the expertise and equipment idle as fossil extraction declines. By 2086, EGS plus conventional geothermal could be 5-10% of global generation, with very high penetration in geologically favored regions (Iceland-like, but also large parts of the western US, East Africa, parts of Southeast Asia). True baseload, 90%+ capacity factor.

Hydro is largely tapped. Marginal growth from new sites and pumped storage. Probably 3-5% of generation in 2086, down from current ~7%.

Storage doesn't generate but enables the intermittent renewables to function as effective baseload. By 2086, grid-scale storage is solved technology: some combination of next-generation batteries (sodium-ion, iron-air, flow batteries, possibly solid-state derivatives), pumped hydro where geography permits, thermal storage in molten salt systems, and hydrogen storage for long-duration applications. The 2086 grid has multiple days of storage at high reliability levels, which transforms solar and wind into effective baseload at the system level even though they aren't at the source level.

The data center question specifically

AGI-grade data centers in 2086 probably aren't drawing baseload from the grid the way current data centers do. The most likely architecture is co-located dedicated generation: typically SMR nuclear, sometimes geothermal where geology permits, occasionally combined with on-site solar plus storage for sites with good solar resource.

The reasoning: AGI inference and training facilities need extremely high reliability, predictable load, and continuous high-power supply. Grid power at 99.99% reliability isn't quite enough; data centers want 99.999%. Dedicated co-located nuclear achieves this with the lowest infrastructure cost. The pattern is already emerging in 2024 with Microsoft's deal to restart Three Mile Island for AI workloads, Amazon's purchase of nuclear-powered data center capacity, and similar moves by other hyperscalers.

By 2086, the typical AGI data center is paired with a 300-1000 MW SMR or AMR (Advanced Modular Reactor), sited together, possibly with shared cooling infrastructure. The largest facilities have multiple co-located reactors providing redundancy. The data center is essentially a private power-and-compute integrated facility, with grid connection as backup rather than primary.

This decouples data center demand from grid baseload concerns in a useful way. The grid no longer has to plan for the largest single loads being unpredictable demand spikes from AI training runs. The data centers operate as their own ecosystems with their own power, and the grid provides power to everything else. The total nuclear footprint includes both grid-supplying reactors and the co-located industrial reactors; the latter category by 2086 is probably 30-40% of total nuclear capacity.

Space, Moon, and Mars

In Earth orbit and cislunar space, power is overwhelmingly solar. Free, abundant, no atmosphere to attenuate. Large stations have massive solar arrays. The ISS in 2024 has about 120 kW of solar power; the major stations of 2086 are operating at megawatt to gigawatt scale, all solar. Some nuclear for specific applications: radioisotope thermoelectric generators for small craft on long missions, fission reactors for high-power applications and for shadow operations where solar isn't available.

The Moon has the 14-day night problem, which makes pure-solar infeasible for any continuously-operating base. The standard architecture for a permanent lunar presence in 2086 is solar plus battery storage for short outages plus a small fission reactor for the lunar night. NASA's Kilopower / Fission Surface Power program is targeting 10 kW reactors for early deployment in the 2030s; by 2086 the technology is mature and 1-10 MW lunar reactors power the permanent bases. Solar concentrators near the lunar poles, where continuous sunlight is available at certain crater rims, complement this.

The Helium-3 fusion story for the Moon is mostly fiction at current understanding. The energy cost of extracting He-3 from regolith at meaningful concentrations is large; the energy yield from He-3 fusion requires fusion technology several generations beyond what's been demonstrated; the math doesn't favor it. The plausible lunar power story is solar plus small fission, not He-3.

Mars is harder. Solar flux is about 40% of Earth's, and dust storms can suppress solar output for weeks at a time. A small Mars outpost (which is what you've specified) needs nuclear as the primary power source: likely a 100 kW to 1 MW reactor, factory-built on Earth and shipped, providing reliable power independent of weather. Solar supplements but cannot be primary. NASA's Kilopower work is exactly this technology. By 2086 a small Mars outpost runs on perhaps 1-5 MW of installed reactor capacity plus solar for daytime augmentation.

The integrated 2086 picture

Putting it together, the realistic 2086 generation mix for the world you've specified:

Nuclear fission (Gen IV + SMRs, both grid and co-located): 25-35% of total energy

Solar (terrestrial + space-based): 25-35%

Fusion: 5-25% (the largest source of uncertainty)

Wind: 10-15%

Geothermal (including EGS): 5-10%

Hydro: 3-5%

Other (biomass, tidal, etc.): 2-5%

Storage and grid management make this work as effective baseload even with high intermittent shares.

The actual baseload providers (the 24/7 firm generation) are nuclear (fission and fusion together at 30-60% of generation depending on fusion progress), geothermal (5-10%), and the firm fraction of solar+storage and wind+storage systems. By 2086 the distinction between baseload and dispatchable is partly dissolved: the grid is managed as a portfolio with multiple firm and variable sources, with storage and AI-coordinated demand-response making the whole thing work as if it were baseload-dominated.

For Lars's fiction, the things worth noting:

The energy system of 2086 looks dramatically different from current discourse implies. The renewables-dominant grid that dominates current policy projections is one piece of the picture, not the whole picture. Nuclear is back as a major contributor in most jurisdictions, with the early-21st-century anti-nuclear consensus largely a generational artifact. Fusion is finally real but its degree of impact depends on whether the engineering scales on the optimistic or pessimistic timeline. Space-based solar is plausibly contributing meaningfully. The data centers run on their own power, mostly nuclear. The Moon and Mars run on small reactors with solar augmentation.

The most interesting story-fact for fiction: the 2086 world has solved its energy problem in the sense that it has abundant clean power, but the energy system is much more centralized and much more dependent on specific high-capital infrastructure than the early-21st-century discourse imagined. The decentralized renewables-and-storage future that some advocates pictured isn't what happened. What happened is that the existing utility-scale model continued, with renewables added in bulk, with storage at scale, and with nuclear (both old and new types) handling the firm-power role. The end result is clean, abundant, and operated by a small number of large institutional players. Energy democracy was promised and didn't arrive; energy sufficiency did.

For your worldbuilding specifically, this means the 2086 power system is operated by 2-3 dozen major institutional operators globally: large utilities, sovereign energy authorities, data-center-and-reactor combined operators, space power consortia. The political economy of energy in 2086 is closer to the political economy of the 1970s telephone system than to anything that was envisioned in the 2010s. Whether this is a feature or a bug depends on what your story is asking. It's certainly fertile territory.

The single most-important uncertainty for any 2086 energy projection is fusion. If it works on the optimistic timeline, the 2086 world is on the steep ramp of a fusion-dominant transition that will define the early 22nd century. If it doesn't, fission carries the firm-power load and fusion remains the not-quite-arrived next-generation technology it has been for so long. I'd bet, with moderate confidence, on a middle scenario: fusion is real and meaningful by 2086, but isn't yet dominant. The dominant transition is still fission + renewables + storage. Fusion is what powers 2120, not 2086.

That's the honest projection, taking your inputs as given.

Lars

I noticed you didn't make any predictions around coal, oil or gas. Why is that, given that for most countries one or more of these are providing significant baseload capabilities?

Claude

Honest answer: I treated the clean-energy transition as more complete than it will be, and jumped straight to the post-transition picture without seriously engaging the persistence of fossil fuels. That's a real gap and worth correcting, because the realistic 2086 picture has fossil fuels in it — significantly less than today, but more than my previous answer implied.

The unstated assumption I made was that 60 years is long enough for fossil fuels to phase out of electricity generation almost entirely. This is the optimistic-discourse default position and it's wrong in specific ways worth naming.

Current state: fossil fuels provide about 60% of global electricity (coal ~35%, gas ~22%, oil ~3%) and about 80% of total primary energy when transport, heating, industry, and chemical feedstock are included. That's the baseline. Any serious 2086 projection has to account for what happens to that 80%, not assume it away.

The reasons fossil fuels persist longer than clean-energy advocacy admits:

Existing infrastructure has economic momentum: a coal plant built in 2020 in India or Indonesia is still operational and economic in 2050. The plants being built now will be running for 30-40 years on financial logic alone, and shutting them early requires either regulatory force or carbon pricing high enough to make operation uneconomic. Many jurisdictions won't apply either.

Gas in particular is dispatchable in a way intermittent renewables aren't: gas turbines start in minutes, ramp fast, fill the gaps when solar and wind aren't delivering. Even with massive grid storage, gas peakers remain economically attractive for the rare high-stress events that storage alone can't cover cost-effectively. A grid that's 80% renewable + storage may still need 5-10% gas peaking capacity for the worst-case weeks per year. The capacity sits idle most of the time and earns its returns during the few periods when it's essential. This is the part of the gas industry that doesn't die.

Petrochemicals need oil and gas as feedstock even if energy use declines. Plastics, fertilizers, lubricants, specialty chemicals. The chemical industry's hydrocarbon demand is structural: you can replace some of it with biomass-derived alternatives and some with synthetic chemistry powered by clean electricity, but a significant fraction persists because the alternatives are more expensive and the chemistry itself is built around hydrocarbon feedstocks.

The political economy of fossil-fuel-producing states creates active resistance to phase-out. Saudi Arabia, Russia, Iran, the US (in specific states), Australia, Canada, Norway, the UAE, Iraq, Nigeria, Venezuela, and others have economies built around fossil extraction. Some have diversified (UAE notably); most haven't. The political pressure these states apply to slow the transition has been significant in the 2020s and will continue. By 2086 most of them are diminished, but the transition has taken longer than it would have without their resistance.

Developing economies have legitimate development arguments for using cheap fossil fuels. The argument is roughly: rich countries got rich by burning carbon for two centuries, and now they want to deny that path to countries trying to industrialize. By 2086 most of this is moot because renewables are cheaper than fossil fuels in most applications, but the political and economic delays from this dynamic have shifted the transition curve to the right by years to decades.

So what's the realistic 2086 fossil fuel picture?

Coal is the fuel most likely to be largely phased out by 2086. The economics are bad even without carbon pricing: new coal can't compete with new solar or wind on cost per MWh in almost any market. Existing coal plants are aging out. The political will to retire coal is strongest because the air quality and health arguments are unambiguous.

The realistic 2086 coal picture: 5-15% of global electricity, concentrated in jurisdictions that didn't transition aggressively. India almost certainly still has significant coal capacity, possibly 15-25% of its generation, much of it with CCS attached or pending. China likely has phased out new coal but retains some installed capacity for grid stability. Southeast Asia (Indonesia, Vietnam, Philippines) probably has more coal than the optimistic scenarios suggest because they built so much in the 2020s and 2030s. Most of the OECD is essentially coal-free by 2086. Total coal consumption is probably 20-35% of current levels, increasingly with CCS where it persists.

There's a non-trivial scenario where coal is effectively gone by 2086: under 5% of generation, persisting only in specific niches. This requires aggressive policy and faster-than-expected economic obsolescence. I'd put this at 25-35% likely. The more probable outcome is meaningful residual coal in specific jurisdictions.

Gas is the more interesting case because it has multiple roles and dies more slowly. Gas in 2086:

For electricity generation, gas is probably 10-20% of total, with most of that being peaking and balancing capacity for renewables-heavy grids. The large continuous-baseload gas plants of the 2020s have been retired or converted; what remains is the rapid-response capacity that fills gaps. Some of this is fueled by bio-methane or synthetic methane (produced from renewable electricity via electrolysis and methanation), which is technically carbon-neutral but uses the same combustion infrastructure.

For industrial process heat, gas persists where electrification is difficult: high-temperature industrial processes, certain chemical reactions, some legacy installations. By 2086 most of this has been electrified or shifted to hydrogen, but a meaningful residual remains.

For chemical feedstock, gas is critical and persistent. Ammonia production, methanol, ethylene precursors, the whole petrochemical industry. This use doesn't decline much with energy decarbonization because it's not energy use; it's material use.

For heating in cold climates, gas is largely displaced by heat pumps in most of the OECD by 2086, but persists in some applications and in regions where the transition was slower.

Total methane consumption in 2086 is probably 50-70% of current levels, with the share going to electricity generation falling much faster than the share going to chemical feedstock.

Oil has the most dramatic decline trajectory because its largest current use (transport) is being electrified rapidly. Light-duty vehicles are essentially electric by 2086. Medium-duty trucks largely electric. Heavy-duty trucks mostly electric with some hydrogen for long-haul. Maritime shipping has transitioned partially to ammonia, methanol, and other alternatives, with some residual heavy fuel oil. Aviation is the hardest case: long-haul aviation in 2086 probably still uses liquid hydrocarbons, but increasingly synthetic fuels (made from CO2 and hydrogen using clean electricity) rather than petroleum-derived.

Petrochemical feedstock continues to consume oil at significant volumes. The economics of producing plastics from oil are hard to beat, even at high carbon prices, because the carbon goes into the product rather than the atmosphere.

Total oil consumption in 2086 is probably 25-40% of current levels. The industry has shrunk dramatically but hasn't disappeared. The geographic distribution has shifted: the US shale industry has largely declined; OPEC production has consolidated around the lowest-cost producers (Saudi Arabia, UAE); Russian production has shrunk significantly; new production in places like Guyana has filled some gaps. The petrostate model is mostly dead by 2086. The states that built their fiscal base on oil rents have either diversified successfully or collapsed economically.

Carbon capture as the wildcard I should have addressed

Carbon capture and storage (CCS) at point sources, and direct air capture (DAC) from the atmosphere, are technologies that fundamentally change the fossil fuel question if they work at scale. By 2086, CCS for industrial point sources (cement, steel, ammonia production, gas turbines) is mature technology, currently being demonstrated and scaling now. DAC is at much larger scale than today, possibly 1-5 billion tonnes per year of CO2 removal capacity.

What CCS does to the fossil fuel projection: it allows specific applications to persist that would otherwise be untenable in a carbon-constrained world. Gas peaking with CCS. Coal with CCS in jurisdictions that committed to it. Industrial heat with CCS. The cost is significant (CCS adds 30-80% to the generation cost of the underlying fuel) but it's economic if the carbon price is high enough or if the regulatory framework requires it.

By 2086, a significant fraction of remaining fossil fuel use is paired with CCS. This isn't enthusiasm for fossil fuels; it's pragmatism about the cost of transitioning specific applications. The result is that fossil fuels persist in a smaller, more expensive, and largely carbon-neutralized form. Whether this counts as "phased out" depends on how you frame it: the molecules are still being burned, but the carbon isn't reaching the atmosphere.

DAC at scale also changes the math by making it possible to compensate for residual emissions from applications where decarbonization is uneconomic. A grid that's 95% clean with 5% unabated gas can be made effectively carbon-neutral by paying for DAC to remove the equivalent CO2. This is expensive but feasible by 2086 if DAC follows the cost-decline curves that solar did.

Revised 2086 generation mix

Taking all this into account, the realistic 2086 global electricity mix:

Nuclear fission (Gen IV + SMRs): 20-30%

Solar (terrestrial + space-based): 25-35%

Fusion: 5-20%

Wind: 10-15%

Geothermal: 5-10%

Hydro: 3-5%

Gas (including for peaking and balancing): 10-20%, partly with CCS

Coal: 3-10%, mostly with CCS in jurisdictions where it persists

Other (biomass, tidal, etc.): 2-5%

This is a different picture than my previous answer implied. Fossil fuels are still 15-30% of electricity generation in 2086. The fossil share is much smaller than today, but it's not negligible. Gas in particular plays a larger role than the clean-energy-dominant picture suggested, because its dispatchability is genuinely valuable for renewable-heavy grids.

For total primary energy (including transport, heating, industry, chemical feedstock), the fossil share is probably 25-40% in 2086 — down from current ~80%, but still significant. Most of this is gas and oil for non-electricity uses. Coal is largely gone outside specific Asian jurisdictions.

Political and geographic implications for the worldbuilding

The fossil fuel transition by 2086 produces a markedly different political geography than current:

The petrostates are mostly diminished. Saudi Arabia, Russia, Iran, Venezuela, Nigeria, Iraq: most of these have lost most of the rent base that funded their political systems. Some have adapted (UAE most successfully); most haven't. The Middle East in 2086 is a different region politically because the oil wealth that shaped the 20th century has largely evaporated.

US energy politics in 2086 looks completely different. The fracking boom of the 2010s-2020s is a memory. The oil-and-gas-state political block (Texas, Oklahoma, Louisiana, North Dakota, Wyoming, West Virginia) has shrunk in economic and political importance. The energy-producing states are now the renewables-rich states (sun belt, wind corridor) and the nuclear-host states.

Australia, Canada, and Norway as major fossil exporters are diminished. Australia's coal export industry is largely dead. Canada's tar sands are uneconomic. Norway has transitioned successfully because they prepared early through the sovereign wealth fund and offshore wind investment.

China's energy politics is the dominant global story. China is the largest renewable energy producer, the largest nuclear builder, the largest grid storage deployer, and possibly the first major operator of space-based solar power. The Chinese energy system in 2086 is the model that the rest of the developing world has substantially copied. Whether this is good or bad depends on what China has become politically by 2086, which we've already established is unpredictable.

India is the second-largest energy story. India has built more renewables than current models project but also retained more coal than ideologically clean models predict. India in 2086 has the largest absolute energy demand of any country and a generation mix that's roughly 60-70% clean, 20-30% fossil with CCS, with significant ongoing transition.

For your fiction this matters because the energy backdrop of 2086 is messier and more politically textured than a clean-transition story would imply. The world isn't powered by clean energy with fossils as a vestigial memory; it's powered by a mixed system where the clean technologies dominate but the fossils persist in specific applications, in specific jurisdictions, with significant political fights still ongoing over the remaining decarbonization. The petrostate decline is one of the major political stories of the century, with substantial geopolitical aftermath. The gas peaker plant with CCS attached, providing the firm balancing capacity for a high-renewables grid, is a recognisable 2086 institution. The Indian coal plant with CCS, providing baseload to support continued development, is too. The petrochemical complex in Texas, now producing materials feedstock rather than transport fuel, still exists and employs people.

The clean-energy advocacy of the 2020s was right that the transition would happen and wrong about how complete it would be by 2086. The transition is real and dominant. It's also incomplete in important ways that produce specific political and economic textures worth representing in fiction set in the period.

Good catch. The previous answer was too clean. The fossil fuel role in 2086 is smaller than today but still real, and the persistence has its own political and economic logic that's worth taking seriously.

Lars

I note there was no mention of battery technologies (not as power generators) or tidal as a source of power. Here in Australia, at least, batteries (both domestic and industrial) as well as pumped hydro have been an enabler of clean energy transition. Do you believe that the combination of fission and fusion will reduce their importance by 2086?

Claude

Direct answer first: no, storage's importance increases by 2086 rather than decreasing, and fission plus fusion don't change that conclusion. The reasoning is structural and worth working through, because the intuition that abundant firm power makes storage unnecessary is wrong in specific ways.

Why storage importance grows

Even with nuclear fission at 25-35% and fusion adding 5-20%, the remaining 50-60% of generation still comes substantially from solar, wind, and other variable sources. The solar plus wind share alone is 35-50% of generation in the realistic 2086 mix. That's an enormous intermittent fleet that needs smoothing across timescales from seconds to seasons. Storage is what makes this intermittent capacity function as effective dispatchable power.

Nuclear doesn't help with this in the way intuition suggests. Both fission and fusion are designed for steady-state baseload operation. Load-following with nuclear is technically possible (French reactors do it routinely) but economically suboptimal. You've built capital-intensive infrastructure to run at 90%+ capacity factor, and underutilizing it raises the levelized cost of the electricity it does produce. The economically optimal architecture is nuclear running flat-out with storage handling the load variation. So even nuclear-dominated grids want significant storage capacity.

Grid resilience is the second structural pressure. The 2086 grid faces threats that the 2024 grid doesn't face at the same intensity: more extreme weather from continued climate change, cyber and physical attacks on infrastructure, cascading failure risk from increased grid complexity, and the strategic dependency that comes with concentrating generation at a smaller number of large facilities. Distributed storage is the primary resilience layer. A grid that can island into local segments and continue operating during major disruptions requires storage at every node.

Data centers want storage that goes well beyond the grid average. An AGI training facility experiencing a power interruption mid-run can lose weeks of compute and millions of dollars in capital costs. The 2086 hyperscale facility has multiple layers of backup: local battery storage for short interruptions, dedicated nuclear for primary power, possibly hydrogen-based long-duration backup for extended events. The storage spending per data center in 2086 may exceed the storage spending per equivalent residential load by an order of magnitude.

Electrified transport creates a massive distributed battery fleet. By 2086, with vehicle electrification essentially complete in developed economies, the global EV fleet represents storage capacity in the tens of terawatt-hours. With vehicle-to-grid (V2G) integration mature, this fleet becomes a major grid asset, with parked vehicles contributing storage and discharge capacity in response to grid signals. The economics of V2G work because vehicles are parked 95% of the time and their batteries are otherwise idle; the grid pays for the use of capacity that owners already have. By 2086 V2G is plausibly 20-40% of total grid storage capacity, distributed across hundreds of millions of vehicles.

Long-duration storage covers the rare but critical events. Both renewables and nuclear have outage scenarios: a cloudy windless week across a continent, a refueling outage at multiple reactors, an extended period of low geothermal output. The 2086 grid has long-duration storage capacity sufficient to ride through these events. The technology mix that makes this work isn't batteries; it's hydrogen produced by electrolysis during surplus periods and burned or fuel-celled when needed, plus synthetic methane and ammonia for similar applications. Round-trip efficiency is poor (30-40% for hydrogen) but this doesn't matter much for storage that's used a few times per year and where the generation feeding it is essentially free surplus.

The integrated effect: storage capacity in 2086 is comparable in capital scale to generation capacity itself. The energy system is roughly half generation, half storage and grid infrastructure. This isn't because the technologies are inefficient; it's because the system is operating at much higher reliability standards than current grids, with more variable inputs, supporting more critical loads, and the economics make storage the right answer at scale.

The storage technology mix in 2086

Different storage durations want different chemistries and physics. The 2086 system stratifies into layers.

Short-duration (seconds to hours). Sodium-ion batteries dominate stationary grid storage because they're cheap, contain no rare materials (sodium is everywhere), have long cycle life, and are safe. Lower energy density than lithium-ion, which doesn't matter for stationary applications where space and weight aren't constraints. Lithium-ion persists primarily in transport and high-density applications. Iron-air batteries provide medium-duration (50-100 hour) storage at low cost — the chemistry is unimpressive on cycle efficiency but the materials cost is negligible. Flow batteries (vanadium, iron, or organic chemistries) provide a different short-to-medium duration option, particularly for very large stationary installations where scale economies favor the flow architecture.

Medium-duration (hours to days). Pumped hydro continues to be the dominant medium-duration technology where geography permits. Snowy 2.0 in Australia, despite its execution problems, is the template: gigawatt-scale, hundreds of GWh of storage, very high cycle efficiency, decades of operational life. By 2086 essentially every viable pumped hydro site globally has been exploited. Compressed air energy storage (CAES) in salt caverns provides similar duration at sites where pumped hydro isn't available. Gravity storage in mineshafts and purpose-built towers fills smaller-scale needs. Thermal storage in molten salt is the natural complement to concentrated solar thermal and operates as a medium-duration component.

Long-duration (days to seasons). Hydrogen produced by electrolysis during surplus generation periods, stored underground in salt caverns or depleted gas fields, and burned or fuel-celled when needed. Round-trip efficiency is poor but the infrastructure is essentially geological: once the cavern is set up, the storage is durable on geological timescales. Synthetic methane and ammonia provide alternative long-duration storage media with the advantage of being transportable. By 2086, large-scale underground hydrogen storage exists in jurisdictions that committed to it: northern Europe (German salt formations), Texas, parts of Australia.

Distributed (V2G and behind-the-meter). The vehicle fleet plus residential and commercial batteries provide distributed storage at the network edge. Coordinated by grid management software (which by 2086 is AI-driven and continuous), the distributed layer handles substantial fractions of load balancing without requiring centralized infrastructure.

Australia as the lead case

Australia is, more than anywhere else, the laboratory for what this all looks like deployed at scale. The reasons are structural to the country.

The solar resource is exceptional. Northern Australia gets among the highest solar irradiance on Earth. The land available for large-scale generation is abundant and largely empty. The population is concentrated in coastal cities that can be served by long-distance transmission. The grid is operated by AEMO, which has become one of the more globally innovative grid operators in integrating high renewable shares. Rooftop solar penetration is the highest in the world (more than 30% of detached homes have solar panels) and home battery adoption is accelerating. The Tesla Big Battery at Hornsdale was the global proof-of-concept for grid-scale lithium batteries providing frequency support and arbitrage. Snowy 2.0 will be one of the largest pumped hydro facilities globally when complete.

By 2086 Australia's trajectory plausibly leads to its becoming a major energy exporter. The Sun Cable / Australia-Asia Power Link proposal (solar generation in northern Australia exported via HVDC undersea cable to Singapore and eventually further into Asia) is the first move in this direction. By 2086, with mature multi-thousand-kilometer HVDC technology, Australia exports clean electricity across the region. The economic position is roughly analogous to what Saudi Arabia occupied in oil in 1980, but with renewable rather than depletable resources.

The technology stack that makes this work is overwhelmingly storage-dependent. Solar generation in northern Australia is variable on multiple timescales: diurnal, weather-driven, seasonal. The HVDC export systems work best at high capacity factor, which means storage at the Australian end smooths the generation profile before transmission. The receiving end also has storage for further smoothing. The transmission system itself becomes the test case for what continent-scale clean energy infrastructure looks like.

For Australia by 2086, storage isn't a supporting technology; it's a primary national infrastructure category, comparable in importance to the road network or the telecommunications grid. The companies and institutions that emerged from this (Tesla's grid storage business, the various Australian pumped hydro operators, the manufacturers of sodium-ion and iron-air batteries at scale) are major industrial players globally. The Australian energy export economy in 2086 looks substantially different from the current coal-and-LNG export model, with clean electricity, hydrogen, and ammonia as the major products and storage technology as a key component of the value chain.

The lesson from Australia for the global picture: high-renewables grids are storage-intensive grids, and the storage isn't a cost to be minimized but an infrastructure category to be invested in deliberately. Jurisdictions that build storage as primary infrastructure rather than as an afterthought to renewables end up with more reliable and cheaper energy systems. The Australian model becomes the model that other sun-rich and wind-rich countries follow through the 2030s and 2040s.

Tidal honestly assessed

You're right that I dismissed tidal too quickly. The realistic 2086 picture for tidal is more interesting than the cursory mention suggested.

Tidal has the property that other intermittent renewables don't: it's predictable. The tides are calculable years in advance, accurate to seconds. This is firm power on a known schedule, which is uniquely valuable for grid planning. The penalty is that the schedule isn't aligned with demand (generation peaks at the tide cycle, not at consumption peaks) but with storage layered in, this becomes a feature rather than a bug. A grid operator planning a week ahead knows exactly how much tidal generation will be available and when.

The technology has split into two approaches. Tidal barrage (dam an estuary, run turbines as tides fill and drain it) has the environmental impact of damming the estuary, which is significant enough that most proposed sites have been blocked. The Rance estuary in France (operating since 1966) and the Sihwa Lake in South Korea are the main commercial barrages. The technology is mature; new deployments are politically constrained rather than technically constrained.

Tidal stream (turbines placed in tidal current flows without damming) is the more deployable approach. The MeyGen project in Pentland Firth (Scotland), various pilots in the Bay of Fundy, projects in the Channel Islands and the Cook Strait, Korean and Chinese developments. Tidal stream is mature technology now and continues to improve.

For Australia specifically, the resource is significant. The Kimberley coast has some of the largest tidal ranges in the world: up to 11 meters in some places, with massive water flows through specific channels. Banks Strait between Tasmania and the islands has strong tidal currents. Various locations along the northern and western coasts have viable tidal stream resources. By 2086, Australian tidal generation is plausibly 2-5% of national electricity, concentrated in specific regional grids near the resource.

Globally, the realistic 2086 tidal contribution is 1-3% of total electricity: meaningful but not dominant. The contribution is concentrated in coastal regions with strong tidal resources: northwest Europe, eastern Canada, parts of Korea and China, southern Argentina, parts of Australia. In these regions tidal can be a major local contributor (10-20% of regional generation). Globally it's a niche technology that has finally been deployed at the sites where it makes sense.

Wave and other ocean energy

Wave energy is the perennial promise that hasn't delivered. The marine environment is harsh, the energy density per device is modest, the maintenance costs are high. By 2086 wave generation has been deployed at sites where it's competitive but remains a marginal contributor: under 1% of global electricity. Some specialized applications (powering coastal infrastructure, complementing tidal generation) have grown, but wave hasn't broken through.

Ocean thermal energy conversion (OTEC) (exploiting the temperature differential between warm surface water and cold deep water in tropical oceans) has tropical applications and provides some local generation in island nations and coastal facilities. Not a global contributor. By 2086, OTEC is a specialty technology in tropical maritime applications.

Osmotic power (salinity gradient at river mouths) remains largely theoretical. Some pilot projects, no commercial scale.

The combined ocean energy contribution (tidal + wave + OTEC + osmotic) in 2086 is probably 2-5% of global electricity. Significant in coastal regions, modest globally.

The integrated picture

Storage is the under-appreciated half of the 2086 energy system: not just an enabler of renewables, but the integrating layer that makes the entire grid work as a coherent system. The grid of 2086 has roughly half its capital invested in generation and roughly half in storage and grid management infrastructure. This is a different shape from the current grid, which is overwhelmingly generation-weighted, and the shift is one of the major capital reorientations of the mid-century.

Australia is the lead case for what the high-storage clean energy economy looks like, and the lessons from Australia diffuse globally through the 2030s and 2040s. Australia's energy export position by 2086 (selling clean electricity, hydrogen, and ammonia into Asian markets) is one of the more significant economic transformations of the period. The political economy of energy in 2086 has new geographic centers, and Australia is one of the largest.

Tidal is real but niche. The good sites get developed; the global contribution is modest. For Australia specifically the resource is meaningful and gets used.

The combined answer to your question: fission and fusion don't reduce storage's importance because storage isn't primarily about compensating for intermittency; it's about grid resilience, load matching, distributed reliability, data center backup, V2G integration, and the rare-event coverage that no generation technology provides on its own. By 2086, storage is at least as important as generation in the energy system architecture, and Australia is one of the countries that figured this out first and benefited most from the early commitment.

Lars

Now let's get into the esoteric — Zero Point Energy sources, micro black holes in cavity devices (like the Lachlan Trilogy), anything else that jumps out.

Claude

The credibility ladder

Worth starting with the discipline I'll apply, because the esoteric territory rewards specificity. Real physics matters even in speculation: the speculative energy sources that are interesting fictionally are the ones grounded in actual physics with extreme engineering challenges, not the ones that require physics to be wrong. The ones that require physics to be wrong are usually less interesting than they sound because they have no engineering grammar to fail in. Fiction set in a world where physics is wrong has to invent its own consistency rules, and the world ends up not feeling like a world.

So: a ladder, from real physics with hard engineering at the top, through real physics with engineering on the edge of possible, to physics that might be wrong, down to pure fiction. The interesting fiction lives in the middle bands.

Tier 1: Real physics, plausible by 2086 in some form

Aneutronic fusion. D-T fusion (the kind ITER pursues) produces high-energy neutrons that activate reactor walls and require heavy shielding. Aneutronic fusion uses fuel cycles that produce mostly charged particles instead: proton-boron (p-B11) being the most studied. TAE Technologies in California, HB11 Energy in Australia, and a handful of others are pursuing this. The physics is real and known. The engineering challenge is that p-B11 requires temperatures roughly ten times higher than D-T fusion: around a billion kelvin instead of a hundred million. Plasma confinement at those temperatures is genuinely harder.

By 2086, aneutronic fusion is plausibly demonstrated at commercial scale, possibly in operation at small scale, possibly a meaningful contributor in jurisdictions that pursued it. The advantage if it works: cleaner waste (no activated structural materials), direct electrical conversion (charged particles can be slowed by electric fields, producing electricity without a thermodynamic cycle), potentially higher efficiency. The risk: it requires plasma physics breakthroughs that may take another fifty years rather than another thirty.

Not exotic in the metaphysical sense, but genuinely outside the conventional fusion roadmap. Worth including in the 2086 picture.

Deep mantle geothermal. Current geothermal taps the hot dry rock 3-7 km below the surface. The deeper version (Quaise Energy and others working on millimeter-wave drilling) proposes wells 10-20 km deep, accessing temperatures of 400-500°C. The drilling technology uses gyrotrons (high-frequency microwave generators originally developed for fusion plasma heating) to vaporize rock rather than crush it. If it works, the cost curve for deep geothermal collapses, and essentially every location on Earth becomes a viable geothermal site.

By 2086 deep geothermal at this level is plausible. The implications would be substantial: geothermal would scale from the 5-10% I projected to potentially 20-30% of global generation, displacing nuclear in some markets and complementing it in others. True baseload, continuous, with no fuel cycle, no waste, no proliferation risk. Hot enough to drive supercritical steam cycles at high efficiency.

Real physics. Engineering challenge real but not insurmountable. By 2086 either this is a major contributor or it's a technology that almost worked and didn't quite scale. Both outcomes are plausible.

Antimatter for niche applications. Antimatter annihilation gives 100% mass-to-energy conversion, far better than fusion's 0.7% or fission's 0.1%. The constraint is production: every gram of antimatter currently costs orders of magnitude more energy to produce than it yields when annihilated. The energy economics fail badly for general power.

By 2086 antimatter production costs have come down (possibly by orders of magnitude) but probably not enough to make it competitive as a general power source. The applications that survive are ones where energy density per kilogram matters more than cost per joule: space propulsion (where carrying fuel mass dominates the rocket equation), possibly some military applications, possibly some research applications. Antimatter isn't powering the grid in 2086 and probably isn't powering it in 2186 either. It's a specialty technology in trace amounts.

Tier 2: Real physics, engineering on the edge

This is where micro black hole power lives, and it deserves serious treatment because the physics is real and the Lachlan Trilogy is working in legitimate territory rather than soft-physics territory.

Micro black hole power as Hawking radiation engines

The physics is genuine. Black holes radiate via the Hawking mechanism: quantum effects at the event horizon produce particle pairs, with one falling in and one escaping. The temperature of the radiation is inversely proportional to the black hole's mass. A small black hole is hot and radiates intensely. A large black hole is cold and radiates almost nothing.

The energy economics are extraordinary. Mass falling into a black hole is converted to radiation at efficiencies approaching 100% over the BH's lifetime: far better than fusion. A 10^9 kg black hole (about a million tonnes, with a Schwarzschild radius smaller than a proton) would radiate at around 10^14 K and produce output power of about 10^16 watts: ten petawatts, roughly a thousand times current global electricity demand. Its lifetime would be on the order of 10^12 years if undisturbed, so it's stable on any human timescale. Fed continuously, it converts the input mass to energy at near-perfect efficiency.

So why is this Tier 2 rather than Tier 1? The engineering challenges are catastrophic.

Creating the BH. You need to concentrate mass into a Schwarzschild density region. For a 10^9 kg BH, the Schwarzschild radius is about 10^-18 meters: smaller than the diameter of a proton. The energy density required at that scale exceeds anything currently achievable by particle accelerators by perhaps fifteen orders of magnitude. Speculative creation pathways include extreme gravitational collapse engineered via concentrated electromagnetic radiation (laser implosion approaches), use of cosmic strings or topological defects if they exist, harvesting of primordial black holes if they exist in the dark matter halo, or beyond-Planck-scale physics we don't yet understand.

For 2086: not realistic. The energy required to create a usable BH is far beyond what 2086 humanity can muster. This is 22nd or 23rd century technology if it ever exists. The Lachlan Trilogy's positioning of it in its specific timeline is the right kind of fictional choice: it places the technology in a frame where the prior engineering work has been done off-page.

Containing the BH. The BH has mass; it falls under gravity. You can't put it on a shelf. Containment requires either:

Electromagnetic levitation. Give the BH a net electric charge by feeding it charged matter, then suspend it in an electromagnetic field. Plausible, but you have to manage the charge: every uncharged particle you feed in dilutes the charge.

Magnetic levitation if the BH has magnetic moment from infalling charged particles. Harder.

Orbital containment: keep the BH in a stable orbit around the center of a containing structure, with the structure rotating to provide centripetal force. Mechanically demanding.

Bringing the cavity to the BH rather than the BH to the cavity: building the cavity around a BH that's been put into a free-fall trajectory. This is closer to what the Lachlan Trilogy implies with the cavity device.

The cavity device approach is good engineering thinking. A magnetically and electromagnetically configured cavity, with the BH suspended in the center, mass fed in through controlled channels, Hawking radiation absorbed by the cavity walls and converted to usable energy through thermodynamic or direct-conversion processes. This is recognizably engineering, with recognizable failure modes.

Feeding the BH. The BH consumes mass and turns it into energy. To maintain steady output, mass has to be fed in at a controlled rate. The accretion cross-section of a small BH is tiny (its Schwarzschild radius determines this); accretion of free particles is slow. The realistic approach is to direct charged-particle beams or atomic beams at the BH, with the particles passing close enough to be captured. This requires precise beam control. Failure modes include overfeeding (BH grows uncontrollably) and underfeeding (BH evaporates faster than it's replenished, eventually catastrophically as it gets smaller and hotter).

Extracting the energy. The Hawking radiation from a usable BH is in extreme gamma-ray or higher frequencies. Conventional photovoltaic conversion doesn't work at those energies. The realistic conversion process is thermal: the radiation heats the cavity walls, which run a thermodynamic cycle. The wall material has to survive extreme radiation flux without degrading; this is itself an enormous materials science challenge. Alternative: direct conversion via particle deceleration if much of the radiation is in charged particles (pair production becomes significant at high temperatures, giving you electron-positron pairs that can be slowed in electromagnetic fields for direct electricity generation).

Failure modes.

Containment failure is the catastrophic scenario. A loose sub-stellar BH on Earth would not consume the planet instantly: its accretion cross-section is too small. It would fall through the Earth's interior, oscillating through the core, slowly accreting matter as it passed through dense regions. Over geologic timescales it would grow, eventually consuming the planet. The civilization-ending event would be slow rather than dramatic.

For smaller BHs, the failure mode is the opposite: if it loses mass faster than it's fed, it gets smaller and hotter, accelerating its own evaporation. Eventually it evaporates violently: a burst of high-energy radiation releasing the remaining mass-energy in a short time. A 10^6 kg BH at the end of its life would release energy comparable to a nuclear weapon over the course of a few seconds. The very smallest BHs evaporate in microseconds, with the final burst comparable to a strategic nuclear weapon.

So the safety calculation for a BH power plant is: containment must be reliable on geological timescales for the long-lifetime case, or the BH must be small enough that an evaporation event is survivable. These are different design philosophies and both have their problems.

The Lachlan Trilogy's positioning of this technology (as something that exists, is functional, has its specific failure modes recognized and managed) is good Lem-school speculative fiction. The physics is real; the engineering is at the edge; the failure modes are catastrophic and require ongoing institutional commitment to manage. This is the kind of energy source whose existence reshapes the civilization that has it. Worth marking as the right fictional treatment of speculative technology.

For 2086 specifically: not realistic as a deployed technology. Possibly being researched in deep theoretical and experimental form. The first proof-of-concept might be 22nd century. Commercial deployment 23rd. The Lachlan Trilogy timeline would have to extend further than 2086 for this to be plausible.

Penrose process around larger black holes. A rotating (Kerr) black hole has an ergosphere: a region outside the event horizon where spacetime itself is dragged by the BH's rotation. Mass entering the ergosphere and splitting, with one fragment falling into the BH along a specific trajectory, can have the other fragment exit with more energy than the original mass possessed. The extracted energy comes from the BH's rotational energy.

Real physics, computed by Roger Penrose in 1969. The energy available is enormous: up to about 29% of the BH's total mass-energy can theoretically be extracted this way. The engineering challenge: requires access to a black hole. Stellar-mass BHs are at minimum tens of light-years from Earth. Primordial BHs in our solar system, if any exist, would be candidates but haven't been detected.

By 2086: not deployed. A 23rd or 24th century technology if humanity reaches the point of routine interstellar access or domestic primordial BH harvesting. Worth knowing about as a far-future power source.

Tier 3: Speculative physics that might be wrong

Zero-point energy. This is the one that needs the cleanest dismissal.

The vacuum has non-zero energy density due to quantum field theory: the lowest-energy state of a quantum field still has fluctuations that carry energy. This is real physics and observable through effects like the Casimir force (attractive force between conducting plates due to mode suppression of vacuum fluctuations between them). The total vacuum energy density predicted by naive QFT is enormous, though it disagrees with observed cosmological dark energy by 120 orders of magnitude (the cosmological constant problem).

So ZPE exists. The question is whether it can be extracted as useful work.

The standard physics answer is no, and the reasoning is straightforward. Energy extraction from a system requires the system to have a lower-energy state to transition into. The vacuum, by definition, is already in its lowest-energy state. There's nothing to extract. Any device claiming to extract ZPE is either fraudulent or doing something other than what it claims (drawing from some other source while attributing the output to vacuum).

There are speculative proposals that try to find loopholes: dynamic Casimir configurations that extract energy from moving plates, claims about extracting energy from the difference between vacuum states under different boundary conditions, various fringe theoretical frameworks. None of these have produced a working device or survived serious physics scrutiny.

Honest assessment: ZPE is not a credible energy source for 2086, 2186, or 2286. The thermodynamic argument is solid. The fictional treatment of ZPE as a power source is in the same category as perpetual motion: fun for fiction, not physics-grounded.

If you wanted to use it in fiction, the productive move is to use a fictional name for a related concept that has the right narrative properties without the false-physics baggage. Or to make it a clear failure mode in the story: the civilization that thinks it's tapping the vacuum is actually tapping something else, and the discovery of what they're actually doing is the revelation.

Vacuum metastability and false vacuum decay. Our universe may be in a metastable vacuum state: not the true ground state of the underlying field theory, but a long-lived local minimum. If this is true, a transition to the true vacuum could release enormous energy. It would also propagate at the speed of light and destroy the physics we know, replacing it with whatever physics the true vacuum supports.

This isn't a power source; it's an extinction event with no human application. Worth mentioning for fictional treatment of what if you accidentally triggered it, which is a different story than power generation.

Tier 4: Things worth knowing about

Magnetic monopole catalyzed proton decay. Magnetic monopoles are theoretical particles that carry magnetic charge. They're predicted by various grand unified theories but have never been detected. If they exist and could be captured, they would catalyze the decay of protons via the Rubakov effect: converting protons directly into mesons and energy at significant rates.

A captured monopole becomes a power source that converts ordinary matter to energy with high efficiency. The economics flip if monopoles can be found and captured (deep ocean searches and Antarctic ice searches have been ongoing for decades without confirmed detection). If they're out there at all, they're very rare.

By 2086: probably still undetected. If they're detected and capturable, this becomes a transformative power source. Worth keeping in the speculative-but-grounded category. Real physics if the particles exist; pure fiction if they don't.

Strange matter conversion. If strange-quark matter is stable at zero pressure (current physics is uncertain), then ordinary matter brought into contact with a strangelet would be converted to strange matter, releasing the binding-energy difference. Very high energy yield per gram converted.

The catastrophic problem: the conversion process might be autocatalytic. Contact between strange matter and ordinary matter converts the ordinary matter to more strange matter, which contacts more ordinary matter, and so on. An Ice-9 scenario where a strangelet on Earth converts the planet to strange matter over some timescale.

This was one of the safety concerns raised before the Relativistic Heavy Ion Collider began operations — the worry that high-energy collisions might produce a stable strangelet. Subsequent analysis suggested strange matter is probably unstable at zero pressure and the concern was overblown. But the underlying physics remains an open question.

For fiction this is rich: the catastrophic-energy-source-that-might-end-the-world theme. For 2086 reality: not in production.

Solar statite power collection. A statite is a hypothetical solar-powered structure that uses radiation pressure from the sun to hold position rather than orbiting: sun-pointing solar sails that hover. This allows positions near the sun, at much higher solar flux than Earth orbit. By 2086 with advanced space infrastructure, the first proof-of-concept statite could exist. By 2186 statite networks beaming power to Earth could be transformative: the proto-stages of a Dyson swarm.

Real physics, hard engineering. Possible by late 21st or early 22nd century.

Cherenkov radiation harvesting and pyroelectric/triboelectric nanogenerators. Not primary power sources but ubiquitous low-power applications by 2086: self-powered sensors, energy-harvesting clothing, ambient-power IoT infrastructure. Not transformative for the grid but transformative for what the grid doesn't have to power.

Tier 5: The pure-fiction frontier

Everything past this point is in the category of might be interesting in fiction, has no physics grounding: tachyon energy, cosmic string mining, brane intersection power, wormhole stellar energy import, extra-dimensional power transfer, vacuum scalar field harvesting, quantum coherence energy, consciousness-mediated energy extraction. All of these have appeared in serious-seeming SF and none of them have a physics basis.

The honest fictional discipline: if your story uses one of these, make it clearly fictional rather than dressing it up as plausible. The Lem-school approach is to take real physics to its extreme rather than to invent new physics, because the real physics under extreme engineering produces stranger and more rigorous fiction than invented physics does. The Lachlan Trilogy's micro black hole approach is the right kind of speculative move precisely because the physics is real; the engineering is the impossible part, and the engineering has specific failure modes that produce dramatic possibilities.

What jumps out as worth adding

A few candidates I'd flag for fictional development that haven't been much used:

Engineered primordial black hole harvesting. If primordial black holes exist in the dark matter halo at small-asteroid masses (an active area of current research), they would be naturally formed Hawking radiation sources of exactly the right size for power generation. A 22nd-century civilization with adequate space capability could survey and capture them. This sidesteps the BH creation problem entirely: you find them rather than make them. By 2086 this is research and possibly first-detection territory. By 2186 it could be deployed.

Directed nuclear isomer release. Nuclear isomers are excited nuclear states that can persist for long periods before releasing energy by gamma emission. Hafnium-178m2 is the famous example: claims have been made (and disputed) that its energy release can be triggered by X-ray stimulation, which would enable a controllable nuclear battery with extremely high energy density. The physics is contested; the experimental claims haven't been replicated reliably. If real, this is a deployable technology by 2086 in specialized applications. If the original claims were wrong (which is the consensus view), nothing happens.

Quantum vacuum thrust without energy extraction. Distinct from ZPE energy extraction: there are proposals for thrust devices that interact with the quantum vacuum to produce propulsion without expelling propellant. The EmDrive was one such claim and has been largely discredited. Mike McCulloch's quantized inertia theory is another. If any of these work, they're propulsion technologies rather than power sources, but they would reshape what's possible in space. For 2086: probably still speculative; either confirmed or rejected by then.

Bose-Einstein condensate energy storage. Not generation but extreme-density energy storage. By 2086 BEC manipulation is mature; whether it produces practical energy storage applications is uncertain.

The fictional discipline

For your work specifically, the productive territory is the boundary between Tier 2 (real physics, extreme engineering) and Tier 3 (physics that might be wrong). This is where the Lem-school sensibility lives: speculation grounded enough that the engineering grammar produces real story problems, speculative enough that the consequences are transformative.

The Lachlan Trilogy's micro black hole power is exactly the right kind of move. Real physics. Catastrophic engineering challenges. Specific failure modes. The civilization that has the technology is reshaped by having it. The technology has a backstory of accumulating engineering achievement that the story doesn't have to explain in detail because the engineering grammar is recognizable.

For 2086 specifically — the timeframe of the broader conversation — the realistic exotic energy picture is:

Aneutronic fusion possibly contributing meaningfully

Deep mantle geothermal possibly transformative if it scales

Antimatter in trace amounts for niche applications

Micro black hole power still theoretical, decades from any prototype

Statite power collection in earliest prototype form

Primordial BH detection campaigns ongoing

ZPE permanently in the won't-work category

The world of 2086 has solved its energy problem with conventional and near-conventional technologies — nuclear fission and fusion, solar at scale, geothermal where viable, storage and grid management making intermittents work. The exotic sources are research programs and aspirations rather than infrastructure. The 22nd century is where the truly exotic technologies arrive, if they arrive at all.

For fiction set in 2086, the right move is to treat exotic energy as the next-horizon technology that the protagonists are aware of but isn't yet deployed. The Lachlan Trilogy's positioning of micro black hole power is the template: it's the technology whose existence implies a civilization that has continued to advance beyond the energy abundance of the late 21st century, into a different relationship with the laws of physics itself. The civilization that runs on Hawking-radiation engines is recognizably descended from the civilization of 2086, and recognizably different from it. That's where the speculative frontier productively lives.

For a longer response, send a reply →

← All conversations