WCEF 2026 Gandhinagar / Accelerator post-read
From silicon
to system
This ring is a chipset's life. Right now it leaks. Close the loop and watch what stays in the economy.
Five modules, about an hour
Pick up where the room left off
Each module is one skill from the session and one thing to try. A dot in the top bar marks the ones you've opened.
Module 1 / Systems thinking
One chipset, four phases
Flip between the line we run today and the loop we're aiming for. Tap any phase to see what goes in, what comes out, and where value slips away.
Ripple test
Pull one lever, watch it travel
Pick an intervention. Each phase shows what shifts, good and bad. Notice which ones leave the loop open.
Look below the surface
The iceberg behind e-waste
Tap each layer, top to bottom. The deeper you intervene, the more of the system changes.
Close to home
Gujarat's chip corridor
India's first commercial fab and three packaging plants sit within a couple of hours of Gandhinagar. Tap a site. Each one is a place where the four phases become real decisions.
Sources: India Semiconductor Mission (ism.gov.in); Blitz India Media, 2026; Anantam IAS, 2026. Status as of mid-2026.
Deep dive
Two loops that run the system
Every system is driven by reinforcing loops that amplify and balancing loops that stabilise. Open each one.
Falling cost per transistor makes devices cheaper. Cheaper devices get replaced sooner. More replacements mean more demand for new chips, which funds the next cost drop. The loop amplifies itself, and e-waste rides along with it.
In numbers: e-waste is rising five times faster than documented recycling (UNITAR, 2024).
As critical minerals get scarcer or geopolitically risky, prices rise, and recovering them from waste starts to pay. Scarcity stabilises the system by making the urban mine competitive with the real one.
This is why gold in circuit-board scrap, at 40 to 800 times ore grade, is starting to look like a resource rather than a problem.
Donella Meadows ranked places to intervene in a system. Changing numbers (a tax, a target) is the weakest. Changing rules, information flows and goals is far stronger.
For chips: a collection target is a number. A digital product passport changes information flows. Product-as-a-service changes the goal from selling units to keeping them in use.
Watch
Three videos for systems thinking
Three videos: the product, the thinking tool, the big idea. Each opens on YouTube in a new tab.
Optimising one phase alone can make the whole system worse. The design decision in phase one decides what's recoverable in phase four.
Module 2 / LCA and material flow analysis
Follow the data, not the hunch
Make your guess first. Then find the input that's tiny by volume and huge by impact.
Step 1
Where does most of a chip's carbon sit?
Most of that comes from electricity. But inside the fab sits a smaller source that punches far above its weight. Try it below. Source: imec, 2025
Step 2
Find the hidden giant
Pick a process gas and a leak size. The right side shows its true climate weight.
GWP100 factors from IPCC AR5. Gases like these are used in plasma etching and chamber cleaning.
The method
How an LCA actually runs
Four stages, set out in ISO 14040 and 14044. Tap each to see it applied to a single chip.
Material flow
What it takes to make 2 grams
Inputs to produce one 32 MB memory chip (Williams, Ayres and Heller, 2002). Log scale, so every step is ten times bigger. Tap a row.
Tap a row to compare it with the chip.
Three scopes
Where a chipmaker's emissions come from
These figures come from different studies with different boundaries. Read them side by side, not added together.
Follow the waste
62 Mt, two ways
Switch between weight and value. Same waste stream, very different story.
The second hidden giant
How thirsty is a fab?
Defaults match Dholera's planned 50,000 wafers a month. Move the sliders to see how water recycling changes the draw on local supply. An estimate built on industry ranges.
UPW use of 4.5 to 7 L per cm² (ITRS roadmap); 1,400 to 1,600 L of municipal water per 1,000 L of UPW (IDE Tech, 2026); one US household ≈ 1,150 L a day (US EPA, via WEF).
fresh water drawn per day
Deep dive
Why two LCAs of the same chip disagree
Three choices that quietly change the answer. Worth asking about in any LCA you're shown.
Primary data is measured at your own fab or supplier. Secondary data comes from databases and averages. Most chip LCAs mix both, and the mix matters.
Ask: which phases used measured data, and which used averages?
With Scope 2 driving most chip emissions, the same process on a coal-heavy grid and a hydro grid can differ severalfold. The ozonated-clean study moved from a 55% to a 63% cut just by switching to hydro power.
Ask: which grid mix, and which year?
A 2002 DRAM chip and a 2025 N2 logic wafer are not comparable. Advanced nodes use more energy per wafer, but pack far more transistors into it.
Ask: per wafer, per chip, or per transistor? The functional unit changes the story.
Watch
Three videos on measuring impact
Start with the method, then see where the impact comes from. Each opens on YouTube in a new tab.
Volume and impact are not the same thing. You cannot close a loop you cannot see, and you cannot see it without measuring it.
Module 3 / Circular product design
Climb the ladder, don't skip to recycling
Ten strategies, ranked by how much value they keep. Read each scenario and tap the rung it belongs on.
Design levers
Redesign a phone
Switch design choices on and off and watch the parts and scores respond. Scores are illustrative, to show direction rather than measured values.
Inside the chip
Why chiplets waste less silicon
One defect ruins a whole die. Split the same area into chiplets and a defect only costs one small tile. Scatter defects and compare.
One big die
Four chiplets per die
Pre-verified chiplets can also be reused across products, cutting design waste as well as silicon waste (yieldWerx).
Copper over gold
A 2026 packaging LCA found copper bonding carries a significantly lower carbon and water-scarcity footprint than gold. Where a package is made matters too: the local grid changes the result.
Designed in, not bolted on
Substrate, bonding metal, glue or screws: each is fixed at the drawing board, years before a recycler ever sees the product.
Case study
What a 10/10 phone looks like
The Fairphone 5 is one of the few phones iFixit rates 10 out of 10 for repairability. Tap each feature to see which lever it pulls.
Fairphone 5
A Dutch social enterprise's phone, built so the owner can fix it with a screwdriver.
Sources: iFixit; Fairphone, 2026; The Register, 2023. The honest trade-off: iFixit notes it uses a mid-range chip, and its board is less modular than the Fairphone 4's.
Deep dive
Four ways to design for the loop
The principles behind every toggle in the teardown above.
Reversible fasteners, labelled connectors, no adhesive where a clip will do. If it can’t come apart, it can’t be repaired or recovered at grade.
Fairphone 5: the back cover unclips without heat or tools.
Plan for the part that fails first. Make it the easiest to swap, and keep software support alive as long as the hardware lasts.
Software support is often what ends a phone’s life, not the hardware.
Pick materials that are lower impact, already in circulation, or easy to separate. Mixed materials bonded together are what make recycling low grade.
Copper bonding over gold; recycled aluminium over virgin.
Common screws, connectors and chiplet interfaces let parts move between products and generations, which keeps them on the higher rungs of the ladder.
Pre-verified chiplets reused across products cut design and silicon waste.
Watch
Three videos on circular design
Why recycling alone falls short, and what better design looks like. Each opens on YouTube in a new tab.
Recycling sits near the bottom. Every rung you climb keeps more of the energy and material already spent, and modular design alone can cut waste by about half.
Module 4 / Business innovation, data and digital tools
Make it pay, make it traceable
Flip the models, run the numbers, then build a product passport and watch it fill up.
Three models
Tap a card to see how it earns
Why durability pays
Sell it, or serve it?
One customer seat over six years. Slide how long the product lasts and watch profit and units made shift. Illustrative numbers, listed on the left.
Sell model: ₹60,000 price, ₹40,000 to make, customer replaces every 3 years.
Service model: ₹1,800 a month fee, ₹300 a month upkeep, maker owns and replaces the unit when it wears out.
Run the numbers
What does remanufacturing save?
Remanufacturing saves an estimated 80–90% of material and energy versus making new (arXiv, 2022). A planning estimate, not a quote.
saved per unit, at the conservative end
The urban mine
What's inside a pile of old phones?
Recovery per million phones from the US EPA. Gold and copper valued at spot prices on 10 September 2026; silver and palladium shown by weight.
Circuit-board scrap can hold 40 to 800 times more gold than the same weight of mined ore (UNITAR, 2024).
gold and copper value recovered
Build it
Your digital product passport
The data that has to travel with a product for take-back and remanufacturing to work. The EU's ESPR is making these mandatory, starting with batteries in 2027.
Untitled product
Pick who is scanning the passport.
Compliance and circularity
Hitting the target isn't closing the loop
India's e-waste EPR collection targets rise each year under the E-Waste (Management) Rules, 2022. Tap a year.
Certificates say it's handled
EPR certificates cover 17 million tonnes of plastic packaging recycled since 2022, roughly four to five years of what India generates. Yet the plastic is still around us.
Sunita Narain, Down To Earth
Where did the material go?
One client met every non-ferrous scrap target with registered recyclers. Traced end to end, recovered copper wiring was shredded and exported as low-grade scrap instead of refined in India. The certificate was earned; the value still left the country.
Deep dive
How the passport rules actually work
Why an Indian exporter should care, even though the rules are European.
The Ecodesign for Sustainable Products Regulation (EU) 2024/1781 entered into force on 18 July 2024. It sets passport requirements product group by product group through delegated acts, rather than all at once.
The EU battery passport is fixed in law and applies from 18 February 2027. Electronics and other groups follow as their delegated acts are adopted.
Batteries are a preview of what electronics suppliers will face.
Obligations apply when a covered product is placed on the EU market, wherever it was made. An Indian component maker supplying an EU-bound product will be asked for passport data.
Gujarat’s new packaging plants in Sanand ship into global supply chains. Their material data will travel.
Each product carries a machine-readable data carrier, such as a QR code, linked to its record, with a central EU registry behind it. Different readers see different fields, as in the role switcher above.
Watch
Three videos on value and traceability
The passport in two minutes, in depth, and the urban mine up close. Each opens on YouTube in a new tab.
A model that sells use instead of units earns more when the product lasts. The passport is what makes getting it back operationally possible.
Module 5 / Change mindset
Swap the chip for your product
The four phases don't change. Pick your industry, then commit to three small moves this week.
Where you stand
A two-minute readiness check
Rate each statement for your organisation, 1 for not at all and 5 for fully. Answers stay in this browser.
Expect pushback
Who loses when you succeed?
Every circular move shifts money or habits for someone. Pick a move to see who resists, and one way to bring them along.
Make it concrete
Write your commitment
Four lines. Copy the card and share it with your team or the session group.
This week
Three small moves
Ticks are saved in this browser, so you can come back and finish.
Plan it
Your 90-day roadmap
Pick one move for each month. The plan builds below as you choose.
Deep dive
Three traps that stall circular programmes
Each has a tell and a fix.
Targets met, certificates filed, nothing changed on the ground. Tell: nobody can say where the material went after collection.
Fix: trace one material stream end to end, once, before scaling anything.
A cleaner fab or a lighter pack, while the product still ends up as waste. Tell: every KPI sits inside one department.
Fix: give one person ownership of the whole loop for one product.
A take-back pilot runs for a quarter and quietly ends. Tell: the pilot has a champion but no P&L line.
Fix: put recovered value on the same P&L as the cost, from day one.
Watch
Three videos to take it further
For sharing with your team after the session. Each opens on YouTube in a new tab.
Read
The sources behind the numbers
Each opens in a new tab.
Compliant on paper can still be linear underneath. The question that keeps a loop honest is simple: where does the material actually go?