Uthando Nethemba

Manufacturing, supply chain and materials bottlenecks (the outcome layer): where the lab cell meets the gigawatt-hour — why LFP won, and why the next generation stalls at scale

—

by

itembalabantu
in

Evidence-first notes on bioscience and deep tech, at the edge of the lab and the market. Information only — not investment advice. All market-share, capacity and cost figures are attributed to their agency or company source (IEA, USGS, BNEF, or the company itself), and “measured/operational (shipped)” figures are separated from “announced/projected” ones. Because this piece rests on agency statistics and trade/think-tank reporting rather than peer-reviewed papers, unverified items are labeled inline. The Northvolt bankruptcy and the China-concentration data are sensitive facts, described strictly neutrally and only from public court, company and agency sources.

The 30-second version

  • What. Below the headline cell chemistry sits the “outcome layer”: whether a lab-validated cell can be produced at gigawatt-hour (GWh) scale, meet a real-world $/kWh, yield and safety bar, and be fed by a secure materials supply. On this layer the winner is decided by manufacturing execution and supply-chain concentration, not cell density. LFP (lithium iron phosphate) won on manufacturability — abundant, low-cost elements (iron, phosphate; cobalt/nickel-free) that can be produced at GWh scale — not on energy density, and it now sits at roughly 90% of grid deployments (Part 0 inheritance).
  • So what. The materials substance of that moat is Chinese vertical concentration. Per the IEA: battery cells ~80%, cathode ~85% (LFP cathode/cells >98%), anode/graphite refining >90%, phosphate ~75%, manganese sulphate ~95% — structurally isomorphic to computing-power’s TSMC/Taiwan apex single-point-of-failure (SPOF). Graphite refining/graphitization is a chokepoint isomorphic to neon (Ukraine ~50%) in semiconductors and to HALEU in nuclear. Lithium resource itself is not scarce (USGS: ~115 Mt); the bottleneck is mid/downstream concentration, not reserves.
  • Now what. Northvolt is the textbook proof that execution — not chemistry — decides the winner: Chapter 11 (November 2024), ~$5.8B debt, ~$30M cash, less than 1% of the planned 16 GWh delivered, roughly 10% utilization — the canonical “announced capacity ≠ shipped product” case. One refuted claim: “Western reshoring resolves China dependence at announced capacity.” And one unresolved skepticism: BNEF’s pack price of $115/kWh (2024, −20%) → $108/kWh (2025) — is it structural learning-curve progress or margin compression from a 3.1 TWh oversupply (about 2.5x demand)? Not adjudicated.

The five-minute read

What “manufacturable at scale” means — and why headline cell specs do not survive it

Between a lab cell (coin-cell / single cell) and a GWh production article sits the outcome-layer gate. “Manufacturable at scale” demands four things at once: yield (defect rate below the economic threshold on large-area continuous lines — early lines with low yield multiply per-cell cost); process compatibility (meshing with existing roll-to-roll coating, calendering, electrolyte filling and formation equipment — a solid-state ceramic separator or dry electrode needs a new process, resetting the equipment/yield learning curve from scratch); materials supply (cathode active material, graphite anode and electrolyte salt sourced stably at GWh scale — the upstream rate limiter); and qualification (automotive-grade thousands of cycles, temperature, fast charge and nail/crush safety, at volume and reproducibly). A lab cell optimizes a single variable (density or charge speed); mass production must satisfy yield x cost x safety x supply simultaneously. LFP is the one chemistry that passed all four; the next generation’s strong lab properties mostly stall in front of that gate.

Why LFP won — manufacturability beat density

LFP trails NMC on energy density, yet it dominates the outcome layer (~90% of grid deployment). The reason is a combined manufacturing/cost/supply advantage: iron- and phosphate-based, cobalt/nickel-free, so exposure to critical-mineral price and geopolitics is lower (the elements are more abundant and cheaper than nickel or cobalt, even though the IEA notes China supplies ~75% of phosphoric acid and ~95% of manganese sulphate); long cycle life and low thermal-runaway risk; and, per BNEF, LFP adoption was a core driver of the 2024 pack price fall to a volume-weighted $115/kWh (−20% year-over-year, the largest drop since 2017). LFP’s win is not “a better cell” but “a cell you can build cheaply, reliably and safely at GWh scale.” The moat’s material substance, however, is Chinese: the IEA records that more than 98% of LFP cathode active material and LFP cells are produced in China. “LFP won” and “China holds LFP” are two sides of the same fact.

Stage China share (2024) Source character
Battery cell production ~80% IEA (agency, primary)
Cathode active material ~85% (LFP >98%) IEA (agency, primary)
Anode active material (mostly graphite) >90% IEA (agency, primary)
Graphite refining >90% (mining ~80%) IEA / EIA (mixed secondary)
Graphitization capacity ~98% Single trade source — unverified
Phosphate (for LFP) ~75% IEA (agency, primary)
Manganese sulphate ~95% IEA (agency, primary)
The outcome-layer winner is decided by manufacturing execution and supply-chain concentration, not cell chemistry. All shares are attributed to the IEA (Global EV Outlook 2025 / Global Critical Minerals Outlook 2025) except graphitization ~98%, which rests on a single trade source and is not cross-confirmed (unverified). Graphite refining is the deepest single point — China holds >90% and has controlled electrode-grade graphite exports since 2023 — structurally isomorphic to computing-power’s neon (Ukraine ~50%) and to nuclear’s HALEU. These are neutral, source-attributed facts, not buy/sell implications for any country or security.

Deep dive

1. Background — the outcome layer, not the chemistry

Parts 1–3 of this series covered the cell properties of each chemistry (solid-state; Na-ion/LFP; long-duration energy storage, LDES). Part 4 looks at the layer beneath: whether a lab-validated cell is produced at GWh scale, meets real-world $/kWh, yield and safety, and is supported by materials supply. The firm’s recurring thesis — “the real bottleneck is elsewhere” — is at its sharpest here. In the storage domain the cell-performance headline is only a starting point; the winner is decided in manufacturing execution and supply-chain concentration. Three demonstrations carry the thesis: (1) LFP won on manufacturability, not density; (2) the next generation stalls not on lab performance but on yield, process compatibility and materials supply; (3) Northvolt’s collapse demonstrates execution risk — buying a good cell design is a separate problem from running a GWh line at yield.

2. What this establishes — Chinese vertical concentration (measured, agency-primary)

Principle: every market-share, capacity and cost figure is reported as in the source and attributed to its agency or company; “measured/operational” is separated from “announced/projected”; and none of this implies a security or country judgment. Per the IEA (Global EV Outlook 2025 / Global Critical Minerals Outlook 2025), China holds ~80% of cell production, ~85% of cathode active material (LFP cathode/cells >98%), >90% of anode active material and graphite refining (mining ~80%), ~75% of phosphate and ~95% of manganese sulphate. Graphite is the deepest single point: China holds refining >90% and has controlled electrode-grade graphite exports since 2023 (EIA/IEA) — a chokepoint that is not a chemistry or physics problem but single-country materials-refining concentration, forming a separate rate-limiting step. This is structurally isomorphic to computing-power’s neon (Ukraine ~50%) upstream chokepoint, and to the HALEU bottleneck in advanced nuclear — a triple isomorphism.

3. Lithium reserves and supply — resource is not the bottleneck

Per the USGS Mineral Commodity Summaries 2025, world lithium measured-plus-indicated resources are about 115 Mt, sharply higher as exploration expands; low prices in 2024 delayed or cancelled many projects, but Argentina, Chile, China and Zimbabwe expanded capacity. US lithium import sources (2020–23) were Chile 50%, Argentina 47%, other 3% — while refining and downstream remain a separate, China-concentrated rate limiter (resource reserves ≠ refining/cell supply chain). The key distinction: lithium resource itself is not scarce; the bottleneck is concentration in refining, cathode and cell manufacturing, not in reserves. “Midstream/downstream concentration” is a more accurate frame than “critical-mineral shortage.” This holds even for sodium-ion (Na-ion): the IEA notes that while upstream (soda ash, caustic soda, biomass) is also supplied by the US and Europe, the downstream — cells, cathode, hard-carbon anode — remains China-dominated. Elemental abundance does not by itself equal a de-China supply chain.

4. Northvolt — the demonstration of manufacturing-execution risk

Northvolt (Sweden) was Europe’s flagship for domestic cell independence, and it collapsed — not because the chemistry was wrong, but on manufacturing execution and capital (cited only from public court, company and think-tank facts). Timeline: it filed for US Chapter 11 on 21 November 2024, followed by bankruptcy of the Swedish parent in March 2025 (the largest in modern Swedish industrial history), per TechFundingNews, Bruegel and energy-storage.news. Financials at bankruptcy: debt of about $5.8B and cash of about $30M (roughly one week of operations; secondary sourcing). The core failure was yield/ramp: production delays at Skellefteå (Ett) were reported from September 2022, and by end-2023 less than 1% of the 2024-planned 16 GWh had been delivered, with actual operation at roughly 10% of installed capacity — the textbook demonstration of “announced capacity ≠ shipped product.” Commercial blows followed (a reported €2B BMW contract cancellation in June 2024; a halt to Ett expansion and ~1,600 layoffs — about a quarter of staff — in September 2024). A structural irony: though Northvolt championed European independence, it relied heavily on Chinese cathode materials and equipment and reportedly needed Chinese personnel to run the lines — suggesting Western reshoring did not bypass the upstream concentration. Firm reading: even a good cell design and abundant capital (over $10B raised cumulatively) do not make ramping a GWh line at yield any less of a distinct, hard problem. (Generalizing a single case to “the West cannot manufacture” is blocked by the skeptic — see section 6.)

5. Why the next generation stalls at scale — not lab performance, but production and supply

The next-generation chemistries of Parts 1–3 (solid-state, Na-ion, LDES) had strong lab/pilot properties yet commonly stall in front of the outcome layer — and the stall point is not performance. Solid-state stalls on the large-area manufacturing yield of ceramic separators, interfaces and dendrites; QuantumScape sits at the B-sample / pilot-line stage (February 2026) and has not reached automotive-qualified GWh production (Part 0 refuted, inherited). Na-ion is near production-ready on properties, but hard-carbon anode and cathode downstream are China-concentrated, and early volumes, real-world $/kWh and 10,000-cycle measurements are deferred (Part 2). LDES (iron-air, flow) has unverified round-trip efficiency and early commercial track record, with $/kWh still a vendor target (Part 3). The common pattern: the substance of “always 5 years away” is not that the lab cell fails but that the GWh line, qualification and materials supply do not follow — a state of not clearing the section-1 gate (yield, process compatibility, materials, qualification).

6. The skeptic’s bottom line (mandatory caveats)

  • Verdict: proceed-with-caveats (conditional). The outcome-layer facts (China cells ~80%, LFP cathode >98%, graphite refining >90%; BNEF pack $115 → $108/kWh; Northvolt bankruptcy and ~10% utilization) are confirmed from agency-primary (IEA, USGS, BNEF) and public sources. But the following caveats must be surfaced at Tier-1 level.
  • Chemistry-split pack prices (LFP vs NMC $/kWh) are secondary citations — unverified (the BNEF original was not accessed directly).
  • Graphitization ~98% rests on a single trade source, not cross-confirmed against other IEA tallies — unverified.
  • Northvolt is a single case — do not generalize to “the West / Europe cannot manufacture” (capital, timing and management were compounding factors).
  • Cost-curve skepticism: whether BNEF’s $115 → $108/kWh fall is structural learning-curve progress or temporary margin compression from a 3.1 TWh oversupply (about 2.5x 2024 demand) is not adjudicated; the 2025 fall (−8%) also has “continued Chinese oversupply” as one axis (BNEF/ess-news). Do not assert “structural cost-curve progress.”
  • Western reshoring: announced capacity vs operational capacity must be distinguished — this piece counts only operational (measured) capacity as the outcome layer.
  • Refuted (1): “Western reshoring/diversification resolves China dependence at announced capacity” is refuted by the Northvolt failure, downstream China concentration and announced ≠ operational. Announcements are abundant, but operation, yield and upstream materials remain tied to China (isomorphic to computing-power Part 5’s diversification frame).

7. What to watch (falsifiable)

  • P1 — cost-curve decomposition: if BNEF pack prices keep falling after oversupply clears (3.1 TWh down), a “structural learning curve” is confirmed; if they rebound as utilization recovers, “temporary margin compression” is confirmed (isomorphic to computing-power Part 3 cost skepticism).
  • P2 — reshoring operation: if by 2028 non-China cell/cathode operational (not announced) capacity reaches a meaningful share (say >20%), “concentration easing” is confirmed; if announcements stay abundant while operation and upstream materials remain China-bound, “concentration persists” is confirmed (isomorphic to computing-power Part 5’s TSMC-diversification frame).
  • P3 — next-gen production gate: if solid-state or Na-ion clears yield, $/kWh and qualification on a GWh line, the section-5 stall has broken; if they stay at pilot, the “lab performance ≠ manufacturability” thesis strengthens (settling with Parts 1–2).
  • Also watch: whether Western reshoring converts announced capacity into shipped product, and whether graphite export controls tighten the graphite refining chokepoint further.

References

Disclosure

This post is for information only and is not investment advice.

COI note: this post describes listed companies (CATL 300750.SZ, BYD, QuantumScape QS, Samsung SDI 006400.KS, LG Energy Solution 373220.KS, Panasonic) and a private/bankrupt company (Northvolt) in a descriptive, neutral, manufacturing- and supply-chain context. Every market-share, capacity and cost figure is attributed to its agency or company source (IEA, BNEF, USGS, EIA, or the company itself), and “announced/projected” capacity is separated from “operational/shipped” capacity — announced does not equal operational. The Northvolt bankruptcy is cited only from public court, company and think-tank sources, and the China-concentration data only from agency statistics; both are sensitive facts described strictly neutrally and are not extended into negative implications for any security or country. Some figures (chemistry-split pack prices, graphitization ~98%) rest on secondary or single trade sources and are labeled unverified. Quantitative claims are attributed to the vendor, agency or source. Competitive and concentration statements are factual, neutral descriptions and are not buy/sell implications for any security. The author holds no position in, and has no financial interest in, the companies named.

Comments

Leave a comment