Solar, Storage, and the Cheap-Energy Flywheel

The Most Important Graph

There's a graph that energy analysts return to again and again. It shows the cost of solar electricity over time—a line that starts in the upper left corner and plunges toward the lower right. In 1976, solar panels cost over $100 per watt. Today, they cost around $0.20 per watt. That's a 99.8% decline.¹

No other energy technology has experienced anything like this. Coal plants today cost roughly what they cost thirty years ago (inflation-adjusted). Natural gas plants are somewhat cheaper due to efficiency gains, but nothing transformative. Nuclear has actually gotten more expensive in Western countries.²

Solar just keeps getting cheaper. And it shows no signs of stopping.

This is not a story about subsidies, though subsidies helped create early markets. It's not primarily a story about scientific breakthroughs, though those have contributed. It's a story about learning curves—the consistent, predictable phenomenon where manufacturing costs fall as cumulative production increases.

Solar panels are manufactured goods. Like semiconductors, like lithium-ion batteries, like LED lights, their costs fall with scale. Double the cumulative production, and costs fall by a predictable percentage. This relationship—Swanson's Law for solar, analogous to Moore's Law for chips—has held for decades.³

The implications are profound. If solar continues on its trajectory, electricity becomes radically cheaper. And cheap electricity changes everything.

2026 Snapshot — Solar and Storage Today

Solar

Global solar capacity exceeds 2,200 gigawatts—more than double what it was just three years ago.⁴ China alone has over 800 GW installed, more than the rest of the world combined a decade ago.

Costs vary by region but continue falling. Utility-scale solar in favorable locations can produce electricity for under $20 per megawatt-hour—cheaper than the operating costs of many existing coal and gas plants, let alone building new ones.⁵

Manufacturing is concentrated in China, which produces over 80% of global solar panels.⁶ This concentration creates supply chain vulnerabilities and trade tensions but has also driven costs down through massive scale.

Technology continues improving:

Commercial silicon panels have reached 22-24% efficiency, up from ~15% a decade ago
Perovskite and tandem cells promise higher efficiencies and lower costs
Bifacial panels capture light from both sides
Building-integrated PV turns facades and windows into generators

Deployment challenges:

Interconnection queues delay projects—in the US, the average wait is 5+ years⁷
Permitting adds time and cost
Land use and community opposition affect siting
Grid upgrades lag behind generation additions

Storage

Battery costs have fallen 97% since 1991, following a learning curve similar to solar.⁸ Lithium-ion batteries dominate both electric vehicles and grid storage.

Grid storage deployment is accelerating. Global capacity exceeded 100 gigawatt-hours in 2023, with projections for 400+ GWh by 2030.⁹ California, Australia, and China lead deployment.

Current applications:

Frequency regulation: Batteries respond instantly to grid fluctuations
Peak shaving: Storing cheap energy and discharging during expensive peaks
Renewable integration: Smoothing solar and wind output
Backup power: Providing reliability during outages

Limitations:

Lithium-ion is cost-effective for 2-4 hours of storage; longer duration remains expensive
Battery manufacturing has supply chain constraints (lithium, cobalt, nickel)
Recycling infrastructure is underdeveloped
Fire risks require careful system design

The Integration Challenge

Solar produces power when the sun shines—typically peaking at midday—while demand often peaks in evening. The mismatch creates challenges:

The "duck curve": In high-solar regions, net demand (demand minus solar production) creates a duck-shaped graph—low during the day, then ramping sharply as the sun sets. Managing this ramp requires flexible generation or storage.¹⁰

Curtailment: When solar production exceeds what the grid can absorb, panels must be turned off—wasting potential clean energy. California curtailed over 2.4 million MWh in 2023.¹¹

Value deflation: As solar penetration increases, midday electricity becomes worth less (there's more of it), while evening electricity becomes worth more. This makes solar's economic case harder at very high penetrations.

Storage addresses these challenges—storing midday solar for evening use—but the amount of storage needed grows as solar penetration increases.

Notable Players

Solar Manufacturers

LONGi Green Energy is the world's largest solar manufacturer by market capitalization, producing both silicon wafers and finished panels.

JA Solar, Trina Solar, JinkoSolar, and Canadian Solar are major Chinese manufacturers with global reach.

First Solar (US) produces thin-film panels using cadmium telluride technology, avoiding silicon supply chain dependencies.

Meyer Burger (Switzerland/Germany) and other European manufacturers are attempting to rebuild Western production capacity.

Perovskite developers including Oxford PV and various startups are racing to commercialize next-generation cell technology.

Storage

CATL and BYD (China) dominate battery manufacturing, supplying both EVs and grid storage.

Tesla manufactures batteries at its Gigafactories and deploys grid storage systems (Megapack).

Fluence (Siemens/AES joint venture) is a leading grid storage integrator.

Long-duration storage startups include Form Energy (iron-air batteries), ESS Inc. (iron flow batteries), Energy Vault (gravity storage), and various compressed air and hydrogen approaches.

Grid and Software

Enphase and SolarEdge dominate solar inverters and power electronics.

AutoGrid, Stem, and others provide software for optimizing distributed energy resources.

Virtual power plant operators aggregate residential solar, batteries, and flexible loads into grid-scale resources.

Solar Breakthroughs: The Next Decade

Efficiency Gains

Current commercial silicon panels convert roughly 20-24% of sunlight to electricity. Laboratory cells have exceeded 26%. Theoretical limits for single-junction silicon are around 32% (near the Shockley-Queisser limit of ~33% for ideal cells).¹²

Paths to higher efficiency:

Tandem cells stack multiple materials, each capturing different parts of the solar spectrum. Perovskite-on-silicon tandems have exceeded 33% efficiency in the lab and are approaching commercialization.¹³

Perovskites are a class of materials that can be manufactured cheaply (potentially through printing) and can achieve high efficiencies. Stability and durability have been challenges, but progress is rapid.

Concentrated solar uses lenses or mirrors to focus sunlight on high-efficiency cells, achieving over 47% efficiency in laboratory settings.¹⁴ Cost and complexity have limited deployment.

Trajectory: Near-term likely for commercial 25%+ efficiency panels. Plausible for 30%+ tandems becoming mainstream within a decade. Higher efficiencies mean more power from the same land area—important as the best sites fill up.

Manufacturing Innovation

Solar cost reductions have come primarily from manufacturing scale, not efficiency. Further cost reductions will likely continue this pattern.

Larger wafers: The industry is moving to larger silicon wafers (182mm and 210mm), producing more power per manufacturing step.

Thinner wafers: Using less silicon per cell reduces material costs.

Automation: Highly automated factories reduce labor costs.

Alternative approaches: Thin-film technologies (like First Solar's cadmium telluride) use less material. Perovskites could potentially be manufactured through printing rather than expensive vacuum processes.

Geographic diversification: Western countries are investing in domestic production (US Inflation Reduction Act, European initiatives), which may increase costs in the short term but reduce supply chain risk.

Trajectory: Continued cost declines of 5-10% annually are plausible. Solar at $10/MWh or below is achievable within a decade in favorable locations.

Building-Integrated PV

Solar doesn't have to be just panels in fields or on rooftops. It can be integrated into building materials:

Solar facades: Glass or other building materials with embedded photovoltaics

Solar roof tiles: Products like Tesla's Solar Roof that replace conventional roofing

Transparent solar: Windows that generate electricity while transmitting light

Solar roads and pavements: Experimental but theoretically possible

Most of these are currently more expensive than conventional solar on a cost-per-watt basis. But they capture otherwise unused surface area and can offset building material costs.

Trajectory: Plausible for building-integrated solar to become significant, particularly in new construction where it can replace rather than supplement building materials.

Storage Breakthroughs: The Next Decade

Lithium-Ion Improvements

Lithium-ion batteries will continue improving:

Higher energy density: More energy stored per kilogram, enabling longer-range EVs and more compact grid storage.

Faster charging: Reducing EV charging time and enabling faster grid response.

Longer cycle life: More charge/discharge cycles before degradation.

Lower costs: Continued manufacturing scale and process improvements.

Chemistry variations: LFP (lithium iron phosphate) batteries avoid cobalt and offer longer life but lower energy density. Different chemistries will serve different applications.

Trajectory: Near-term likely for continued 5-10% annual cost declines. Lithium-ion will dominate short-duration storage (4 hours or less) for the foreseeable future.

Solid-State Batteries

Solid-state batteries replace the liquid electrolyte in conventional lithium-ion with a solid material, potentially offering:

Higher energy density
Faster charging
Improved safety (less fire risk)
Longer life

Status: Multiple companies (QuantumScape, Solid Power, Toyota, Samsung) are pursuing solid-state batteries. None have achieved mass production at competitive costs.

Challenges: Manufacturing solid-state batteries at scale has proven difficult. The interfaces between solid components degrade in ways liquid electrolytes don't.

Trajectory: Plausible for solid-state batteries to enter production for EVs within a decade. Whether they'll be significantly cheaper than improved lithium-ion is uncertain.

Alternative Battery Chemistries

Sodium-ion batteries use abundant sodium instead of lithium. CATL and others have begun production. Lower energy density but potentially lower cost.¹⁵

Iron-air batteries (Form Energy) use rust chemistry for very long duration (100+ hours) at potentially very low cost. Suitable for multi-day storage, not for EVs.

Flow batteries store energy in liquid electrolytes that can be scaled independently of power capacity. Good for long duration; less suitable for mobile applications.

Trajectory: Near-term likely for sodium-ion to capture some market share. Plausible for iron-air and flow batteries to address long-duration storage needs.

Beyond Batteries

Some storage approaches don't use batteries at all:

Pumped hydro remains the dominant form of grid storage globally. Water is pumped uphill when power is cheap, released through turbines when power is valuable. Limited by geography.

Compressed air stores energy by compressing air into underground caverns. A few commercial plants exist.

Gravity storage (Energy Vault) uses cranes to lift and lower heavy blocks, storing energy as gravitational potential.

Hydrogen can store energy indefinitely and in large quantities. Electrolysis splits water into hydrogen and oxygen; the hydrogen can later be burned or run through fuel cells. Currently inefficient (round-trip efficiency ~40%) and expensive, but potentially valuable for seasonal storage.

Thermal storage heats or cools a medium (sand, molite salt, rock) for later use. Particularly relevant for industrial heat applications.

Trajectory: Pumped hydro will continue where geography allows. Hydrogen may become important for very long-duration and seasonal storage. Various other approaches will find niches.

The Cheap-Energy Flywheel

As solar and storage costs fall, entirely new applications become economic. This creates a flywheel effect: new applications drive more deployment, more deployment drives further cost reductions, further cost reductions enable even more applications.

AI-Managed Grids

Today's grids are managed by human operators using decades-old software. AI enables:

Real-time optimization: Continuously balancing supply and demand across millions of resources

Predictive maintenance: Identifying equipment failures before they cause outages

Coordinated distributed resources: Managing millions of rooftop solar systems, batteries, and EVs as a unified resource

Automated trading: Optimizing electricity purchases and sales across markets

Example: A grid with millions of EVs could charge them during solar peaks (absorbing excess generation) and discharge them during evening peaks (providing backup power)—all coordinated automatically.

Trajectory: Near-term likely for increasing AI involvement in grid operations. The complexity of high-renewable grids may make AI essential rather than optional.

Desalination and Water

Desalination is energy-intensive: producing fresh water from seawater requires roughly 3-4 kWh per cubic meter.¹⁶ At current electricity prices, this limits desalination to wealthy, water-scarce regions.

Cheap solar changes the equation: In sunny coastal deserts—precisely where water is scarcest—solar is cheapest. If electricity falls to $10/MWh, desalination costs drop by half or more.

Implications:

Water scarcity becomes a solvable problem in coastal regions
Agriculture becomes possible in currently barren areas
Geopolitical conflicts over water could diminish

Trajectory: Near-term likely for solar-powered desalination expansion in the Middle East, North Africa, and Australia. Plausible for cheap desalination to transform coastal water economics globally.

Industrial Heat and Hydrogen

Many industrial processes require high temperatures currently provided by fossil fuels. Cheap electricity enables alternatives:

Electric arc furnaces can produce steel from recycled scrap using electricity rather than coal.

Green hydrogen can replace coal in steelmaking (direct reduced iron process) and provide high-temperature heat for other processes.

Electric furnaces can provide heat for some processes directly.

The constraint has been cost: electricity from fossil fuels is expensive, and renewable electricity has been limited. Very cheap renewable electricity changes the calculus.

Green hydrogen economics: Electrolysis converts electricity to hydrogen. At $50/MWh electricity, green hydrogen costs roughly $5-6/kg. At $15/MWh, it drops to $2-3/kg—potentially competitive with fossil-derived hydrogen.¹⁷

Trajectory: Plausible for green hydrogen to become cost-competitive for industrial applications within a decade in regions with excellent solar resources.

Direct Air Capture

Removing CO2 from the atmosphere could offset emissions and eventually draw down accumulated greenhouse gases. Current direct air capture (DAC) costs $400-600 per ton of CO2—far too expensive for gigatonne scale.¹⁸

The energy requirement: DAC requires roughly 2,000-2,500 kWh per ton of CO2 captured (mostly heat, some electricity).

At current costs: $100/MWh electricity means $200-250 just for electricity, plus capital and other costs.

At future costs: $15/MWh electricity means $30-40 for electricity. Total costs could potentially fall below $100/ton.

At $100/ton, DAC becomes plausibly affordable at scale. At $50/ton, it becomes a serious climate solution.

Trajectory: Plausible for DAC to reach $100-150/ton within a decade if solar and electrolysis costs continue falling. Scaling to gigatonne capacity would still require massive investment.

Computing and AI

Data centers consume enormous amounts of electricity—currently 1-2% of global electricity, growing rapidly as AI scales.¹⁹

Cheap solar enables:

Locating data centers in solar-rich regions
Powering AI training with clean energy
Running energy-intensive computations when power is cheapest
Potentially enabling more AI development overall

The irony: AI accelerates solar development (through materials discovery, grid management, etc.), and cheap solar enables more AI (through affordable compute). The two technologies may reinforce each other.

Second-Order Effects

Energy Geopolitics Transforms

Fossil fuels have shaped geopolitics for a century. Oil exporters (Saudi Arabia, Russia, Iran) have wielded enormous influence. Oil importers (US, China, Europe, Japan) have structured foreign policy around energy security.

Solar changes this equation:

Sunlight is distributed globally; it can't be embargoed
Manufacturing can be located anywhere (though supply chains matter)
Energy independence becomes achievable for most countries
Petrostates face declining influence and revenue

The transition will be turbulent: Countries dependent on oil revenue must diversify. Energy importers gain leverage. New dependencies (on battery materials, manufacturing) replace old ones.

Trajectory: Near-term likely for continued erosion of fossil fuel influence. The pace depends on how quickly transportation and industry electrify.

Distributed vs. Centralized

Historically, energy has been centralized: large power plants, long transmission lines, utility monopolies. Solar enables a different model:

Rooftop solar lets homes and businesses generate their own power.

Batteries let them store it.

Smart inverters and software let distributed resources participate in grid services.

The result: Potential shift from passive consumers to active "prosumers." Utilities' business models must adapt.

Implications:

Grid defection becomes possible (going fully off-grid)
Utility revenues decline as customers self-generate
Grid costs must be recovered from fewer customers
Equity questions arise (who can afford solar vs. who subsidizes remaining grid costs)

Trajectory: This is already happening in sunny regions. The pace and extent of decentralization will vary by geography and policy.

Climate and the Carbon Budget

The world has emitted roughly 2,500 gigatons of CO2 since the industrial revolution. To limit warming to 1.5°C, the remaining carbon budget is roughly 400-500 gigatons at current emission rates—less than a decade's worth.²⁰

Solar's role: Even aggressive solar deployment won't change the near-term trajectory enough. But by 2035-2040, if solar and storage scale sufficiently:

Most new electricity could be renewable
Transportation could be substantially electrified
Industrial heat could begin switching to hydrogen
Emissions could be falling rather than rising

The critical variable is speed. The technology is ready. Deployment must accelerate.

Risks and Guardrails

Supply Chain Concentration

Solar manufacturing is concentrated in China. Battery manufacturing depends on minerals from a few countries (lithium from Australia and Chile, cobalt from Congo, nickel from Indonesia).

Risks:

Trade disputes disrupting supply
Geopolitical leverage over clean energy
Single points of failure (e.g., one factory fire disrupting global supply)

Mitigation: Diversifying manufacturing (IRA in US, EU programs), developing alternative chemistries (sodium-ion, solid-state), and recycling infrastructure.

Grid Reliability

High renewable penetration creates reliability challenges:

Variability: solar and wind fluctuate with weather
Inertia: spinning generators provide stability; inverters must be designed to replace this function
Extreme events: extended periods of low renewable generation require backup

Examples of failure: California's 2020 rolling blackouts occurred during a heat wave when solar output dropped in evening hours. Texas's 2021 failures included frozen wind turbines (though gas system failures were more significant).

Mitigation: Storage, transmission, dispatchable backup (gas, nuclear), demand response, and better grid management.

Resource Constraints

Scale-up requires resources:

Silicon: Abundant, not a constraint
Silver: Used in solar cells; alternatives are being developed
Lithium: Abundant but production must scale; prices have been volatile
Cobalt: Concentrated in DRC; ethics concerns; substitution underway
Nickel: Supply must expand; deep-sea mining is controversial
Land: Large solar farms require significant land area

None of these are absolute barriers, but each requires attention and investment to avoid becoming bottlenecks.

Equity and Access

The benefits of cheap solar and storage may not be evenly distributed:

Wealthy homeowners install rooftop solar; renters and apartment dwellers don't
Developing countries may lack capital for solar deployment
Legacy fossil fuel workers and communities face economic disruption

Policy responses can address equity—but they're not automatic. The clean energy transition can increase or decrease inequality depending on choices made.

The Path Forward

Near-term likely (5-7 years):

Solar continues to be the cheapest new electricity source in most markets
Battery costs continue falling 5-10% annually
Grid storage deployment accelerates
Some transmission bottlenecks are addressed; others persist
Interconnection queues remain a significant constraint

Plausible (7-15 years):

Solar reaches $10-15/MWh in favorable locations
Long-duration storage (10+ hours) becomes cost-effective
Green hydrogen becomes competitive for some industrial applications
Solar-powered desalination expands significantly
Grids operate with 60-80% renewable penetration in leading regions

Wild (speculative):

Electricity becomes "too cheap to meter" in some contexts
Energy abundance transforms industrial economics
Large-scale direct air capture becomes affordable
Fossil fuels become stranded assets faster than expected
Energy ceases to be a significant constraint on economic activity

The solar-storage flywheel is already spinning. Each revolution makes the next cheaper. The question is not whether this transformation will happen but how fast—and whether it will happen fast enough.

Endnotes — Chapter 10

NREL and Our World in Data track solar cost history. The decline from >$100/W in 1976 to ~$0.20/W today represents one of the fastest cost declines of any energy technology.
Lazard's Levelized Cost of Energy Analysis documents cost trends across energy sources. Nuclear costs in Western countries have generally increased over time.
Swanson's Law describes the relationship between solar costs and cumulative production. The "learning rate" (cost decline per doubling of production) has been approximately 20-25%.
IEA and IRENA track global solar capacity. Total capacity exceeded 1,500 GW in 2024.
Utility-scale solar PPA prices in the US Southwest and other favorable regions have reached $15-25/MWh. Global average costs are higher but declining.
Chinese dominance in solar manufacturing has been documented by IEA and various industry analyses.
Lawrence Berkeley National Laboratory tracks US interconnection queue data. Average wait times have increased significantly as renewable project submissions have grown.
BloombergNEF tracks battery price declines. Pack prices have fallen from >$1,100/kWh in 2010 to ~$140/kWh in 2023.
BloombergNEF and IEA project grid storage deployment. Capacity is growing 20-30% annually.
CAISO originated the "duck curve" term and concept to describe net load patterns in high-solar systems.
CAISO publishes curtailment data. Solar curtailment has grown as penetration has increased.
The Shockley-Queisser limit (~33% for single-junction cells under standard conditions) is a fundamental thermodynamic constraint. Practical limits for silicon are somewhat lower.
NREL efficiency charts track record cell efficiencies. Perovskite-silicon tandems have exceeded 33% in laboratory settings.
Multi-junction cells under concentration have exceeded 47% efficiency in laboratory settings.
CATL and BYD have announced sodium-ion battery production. Energy density is lower than lithium-ion but costs may be lower.
Desalination energy requirements vary by technology. Reverse osmosis typically requires 3-4 kWh/m³.
Green hydrogen cost projections depend heavily on electricity prices and electrolyzer costs. IRENA and BloombergNEF publish various scenarios.
Direct air capture costs are estimated by companies (Climeworks, Carbon Engineering) and researchers. Current costs are $400-600/ton; projections vary widely.
IEA estimates data center electricity consumption at 1-1.5% of global electricity, with growth driven by AI.
IPCC Sixth Assessment Report provides carbon budget estimates for various temperature targets.