The Road to Abundance

Search for the substance a better world needs

Designing materials

New batteries, catalysts, membranes, and electronic materials become easier to discover and improve.

The unlock

AI proposes candidates, predicts selected properties, and guides experiments. Better quantum many-body methods could extend what is computationally tractable.

The hard part

A predicted material is not a manufactured product. Stability, synthesis, durability, supply chains, and performance at scale determine practical value.

How to read these scenarios

These are positive possibilities, not promises or timeline forecasts. Lists put the most intriguing ideas first: an editorial judgment about interest and transformative potential, not likelihood or arrival date. The labels describe what must happen next, not numerical probabilities.

Products are useful tools and services; cures include potential treatments and prevention, not claims of proven cures; insights are discoveries that help us understand the world.

Positive possibilities

Most intriguing first

5 scenarios to explore

ProductBreakthrough-dependent

Ambient-condition superconductors

If suitable materials exist, low-loss components could transform some magnets, electronics, and power equipment.

Why AI helps
Better models of interacting electrons could guide searches that are currently difficult to calculate.
What still has to happen
AI cannot guarantee a practical room-temperature, ordinary-pressure superconductor exists; synthesis and useful current capacity are additional hurdles.
Read details
Full details: Ambient-condition superconductors

Designing materials · Product

Ambient-condition superconductors

Breakthrough-dependent

If suitable materials exist, low-loss components could transform some magnets, electronics, and power equipment.

A superconductor has distinctive electrical and magnetic properties below its operating limits. A practical material that worked near room temperature and ordinary pressure could remove some cooling requirements and enable new designs for selected devices.

The central uncertainty is whether a suitable material exists and can be made reliably. Even a convincing superconductivity result does not establish useful wire: critical current, magnetic-field tolerance, mechanical properties, and manufacturing all matter.

What this could look like

If a candidate material is discovered, independent laboratories reproduce its superconducting behavior. Engineers then investigate whether it can carry useful currents in a workable component under relevant conditions.

Who benefits, and how

  • Potentially less demanding cooling for some superconducting devices.
  • New options for magnets, sensors, and electrical components if practical properties align.

What would make it real

  1. Reproduce both electrical and magnetic evidence under the claimed temperature and pressure conditions.
  2. Establish usable current capacity, field tolerance, durability, and manufacturability.

AI cannot guarantee a practical room-temperature, ordinary-pressure superconductor exists; synthesis and useful current capacity are additional hurdles.

Explore the evidence

Background, not proof this scenario is available.

Permanent link · opens in a new tab
ProductPlausible next step

Batteries better suited to everyday life

Vehicles and stationary storage gain safer, longer-lived, or less resource-intensive battery chemistries.

Why AI helps
AI can prioritize promising electrodes, electrolytes, and combinations for testing.
What still has to happen
Energy density, cycle life, safety, charging rate, cost, and manufacturability must work together.
Read details
Full details: Batteries better suited to everyday life

Designing materials · Product

Batteries better suited to everyday life

Plausible next step

Vehicles and stationary storage gain safer, longer-lived, or less resource-intensive battery chemistries.

A battery is a system of interacting components. A promising electrode can fail when paired with a particular electrolyte, or work well in a laboratory cell but degrade during fast charging. AI could help prioritize combinations and understand failure patterns.

There is rarely a single best chemistry. A stationary storage system, a car, and a wearable device place different weights on weight, volume, lifetime, safety, materials availability, and cost.

What this could look like

A team designing stationary storage searches for a chemistry with long cycle life and abundant inputs, even if it stores less energy per kilogram than a premium vehicle battery.

A better battery depends on the job

Design goals and the trade-offs to test
Desired improvementWhat can pull the other way
More energy per kilogramMust also satisfy safety, lifetime, and practical packaging requirements.
Faster chargingCan increase heat or degradation unless chemistry and thermal management support it.
Longer operating lifeMay require different materials or a more conservative operating window.
Abundant, inexpensive inputsStill need suitable performance and a scalable manufacturing process.
Qualitative design tradeoffs, not measured performance claims. Improvements can sometimes be combined, but one laboratory metric is not enough to rank a battery.

Who benefits, and how

  • Battery designs better matched to their intended use.
  • More informative experiments before expensive scale-up.

What would make it real

  1. Test full cells under realistic operating and aging conditions.
  2. Demonstrate a reproducible manufacturing process and acceptable system-level economics.

Energy density, cycle life, safety, charging rate, cost, and manufacturability must work together.

Explore the evidence

Background, not proof this scenario is available.

Permanent link · opens in a new tab
ProductPlausible next step

Catalysts that reduce industrial energy use

Useful chemicals are manufactured with less energy or fewer unwanted byproducts.

Why AI helps
Models can screen reaction pathways and candidate catalyst surfaces.
What still has to happen
Activity in a model or laboratory must survive real feedstocks, long operating periods, and industrial economics.
Read details
Full details: Catalysts that reduce industrial energy use

Designing materials · Product

Catalysts that reduce industrial energy use

Plausible next step

Useful chemicals are manufactured with less energy or fewer unwanted byproducts.

A catalyst can change the rate or selectivity of a chemical reaction without being consumed in the overall reaction. Discovering a useful one requires balancing activity with stability, availability, and behavior under actual operating conditions.

AI can guide the search by comparing candidate structures, reaction pathways, and experimental results. A catalyst that looks excellent with clean laboratory inputs may lose performance when exposed to impurities or repeated use.

What this could look like

Researchers investigate catalysts for making an important chemical at less demanding conditions, measuring not only reaction rate but also unwanted products and performance over time.

Who benefits, and how

  • Potentially lower energy demand for selected chemical processes.
  • Less waste when reactions produce more of the desired product.

What would make it real

  1. Validate activity, selectivity, and lifetime under representative process conditions.
  2. Assess the full process, including catalyst production, recovery, and replacement.

Activity in a model or laboratory must survive real feedstocks, long operating periods, and industrial economics.

Explore the evidence

Background, not proof this scenario is available.

Permanent link · opens in a new tab
ProductPlausible next step

More effective water-treatment membranes

Water systems remove selected contaminants or salts with improved energy use and durability.

Why AI helps
AI can search material structures for permeability and selectivity tradeoffs.
What still has to happen
Fouling, maintenance, disposal, pressure requirements, and system costs still matter.
Read details
Full details: More effective water-treatment membranes

Designing materials · Product

More effective water-treatment membranes

Plausible next step

Water systems remove selected contaminants or salts with improved energy use and durability.

A membrane separates substances by allowing some to pass more readily than others. Researchers must balance throughput, rejection of the target contaminant, operating pressure, and resistance to fouling.

AI could help identify promising structures and connect laboratory measurements to operating conditions. A good result with a clean test solution is only a first step because real water contains mixtures that interact with the membrane.

What this could look like

A water-treatment team tests candidate membranes for removing a particular contaminant, then runs them on representative feed water to measure rejection, flow, fouling, and cleaning requirements.

Who benefits, and how

  • More selective treatment for particular water problems.
  • Potentially lower energy or maintenance needs in well-matched applications.

What would make it real

  1. Measure performance over time on realistic water mixtures.
  2. Account for cleaning, membrane replacement, and disposal of concentrated contaminants.

Fouling, maintenance, disposal, pressure requirements, and system costs still matter.

Explore the evidence

Background, not proof this scenario is available.

Permanent link · opens in a new tab
InsightPlausible next step

Reusable design rules for chemistry

Scientists learn which structural features produce valuable properties, allowing discoveries to transfer between material families.

Why AI helps
AI can propose relationships across calculations and experiments that researchers then test.
What still has to happen
A predictive pattern must survive new compositions and conditions before it becomes a dependable design rule.
Read details
Full details: Reusable design rules for chemistry

Designing materials · Insight

Reusable design rules for chemistry

Plausible next step

Scientists learn which structural features produce valuable properties, allowing discoveries to transfer between material families.

A predictive model may identify relationships between a material's structure and a useful property. Scientists can investigate whether those relationships reflect mechanisms that remain valid beyond the examples used to train the model.

Failed experiments are informative too: they help define where a rule stops working. The ambition is a design principle that guides a new search, rather than a model that only recognizes close relatives of known materials.

What this could look like

A team proposes that a particular structural feature improves ion transport. It deliberately tests materials that separate that feature from other correlated properties, refining the explanation when predictions fail.

Who benefits, and how

  • More transferable knowledge from each experiment.
  • Better search strategies across related material families.

What would make it real

  1. Test prospective predictions on new compositions and operating conditions.
  2. Explain boundary cases and include negative results rather than only successful examples.

A predictive pattern must survive new compositions and conditions before it becomes a dependable design rule.

Explore the evidence

Background, not proof this scenario is available.

Permanent link · opens in a new tab

Evidence & further reading

Background for the capabilities and scientific constraints above. These sources do not establish that every scenario will happen.

Solving math Modeling cells Robotics All possibilities →
Possibility in focus