Sodium-ion batteries trade cheaper, abundant inputs for lower energy density. This asks whether China still deploys them at grid scale while lithium is cheap.
50%likely
Well sourcedNext check
Watching
Why it is different
The bet is that China builds them for supply-security reasons even while lithium is cheap, against the cost logic Western analysts rely on.
Why it matters
At scale, the chemistry would diversify grid storage away from lithium supply and change the strategic value of domestic sodium inputs.
Next check · 31 Oct 2026
Does CATL’s Q3-2026 disclosure confirm that commercial sodium-ion grid-battery deliveries actually began in September 2026 as guided?
How we scored it ·
a judgement call
If right, how big3/5How overlooked4/5Upside vs downside2/5Source trust5/5Independent sources4/5
The exact wording we score
By 31 Dec 2028, China has commissioned at least 20.0 GWh cumulative of grid-scale stationary storage whose installed cell chemistry is sodium-ion. Announced orders, factory capacity and projects converted to LFP do not count. Resolves YES at 20.0 GWh or above; NO below 20.0 GWh using CNESA project-level deployment records, cross-checked against vendor disclosures. Lithium price affects the edge narrative but is not an outcome condition.
PROOF
Evidence and provenance
The links below are the evidence recorded when this forecast was scored. Each note states source quality or bias; a citation is not an endorsement.
Recorded sources · 5 citations / 4 independent origins
2026-07-28 | sell-side: low lithium prices dominate the economic view; mainstream press: sodium-ion is commonly framed as delayed or niche; priced market: no direct contract identified; specialist: Chinese tender and industry data support strategic scaling despite weak lithium-cost logic | E4
Revision history4 entries
Probabilities and interpretations are never silently overwritten. This is the append-only record of what changed and why.
The canonical claim is now brought into line with the prior log: one 20.0 GWh cumulative commissioned-capacity boundary, explicit exclusions for orders/factories/LFP conversions, and exhaustive YES/NO outcomes. The title no longer makes low lithium a hidden second condition.
(2nd pass) — P 66→50 on a threshold defect, and a self-contradiction removed. (1) The claim required >=20 GWh cumulative by end-2028 while the falsifier triggered on sub-10 GWh annual 2028 deployment — different quantities, so a 2028 with 9 GWh annual on 14 GWh prior cumulative would satisfy the claim AND trigger the falsifier simultaneously. Even read as cumulative, the 10-20 GWh band resolved neither. P=66 was almost exactly the midpoint of P(>=20 GWh)~0.50 and P(not falsified, >=10 GWh)~0.82 — what happens when a claim and its falsifier are scored as one statement. Both thresholds are now 20 GWh cumulative; P=50. (2) The edge argument's whole point is that China scales Na-ion regardless of the lithium spread, yet the kill criteria listed 'a lithium price that never rises to help Na-ion' — making the claim's own central premise a falsifier, so it could not lose cleanly on its own mechanism. Removed. The lithium checkpoint is retained but reframed: it updates EDGE, not probability — if lithium rallies through RMB 150k/t, deployment gets more likely while the edge collapses, because Western analysts would then re-adopt Na-ion on exactly the cost logic this file says they wrongly dismissed. Being right for the consensus reason is not edge.
v2 rescore. P 60→66: this file's framing ('despite low lithium prices') is now stale. Lithium carbonate traded at RMB 146,500/t on 2026-07-27, up ~98% over twelve months, with SMM reporting H1-2026 prices up >130% YoY. That directly erodes the IEA bear anchor this signal was written against ('current lithium prices are not yet high enough for sodium-ion to undercut LFP'). CATL's H1-2026 report (2026-07-24) contained no sodium-ion or ESS-chemistry disclosure, so the September delivery guidance is neither confirmed nor contradicted. Note for indepsources: the two IEA citations are one origin, not two.
created from fresh (non-power) research sweep. IEA used as the bear anchor against its own domain — the disagreement is the signal. Watch deployed GWh, not the cost debate.
Method challenge
Adversarial review record
Conditionally admitted
Structured internal role review tied to this frozen claim. It is not independent human peer review. Independent specialist review not yet performed
systems or deployment specialist
measurement and data reviewer
policy or geopolitical reviewer
forecasting-method reviewer
Strongest specialist challenge
Vendor agreements and factory capacity are not commissioned grid storage, and hybrid projects can contain only a small sodium-ion share. Cheap LFP, hard-carbon constraints, and project conversion can leave a large order book with little deployed sodium capacity.
Outside view
Challenger chemistries with commercial orders often take longer than vendor forecasts to reach multi-gigawatt-hour field deployment. China's unusually fast deployment system raises the base rate, while the exact 20 GWh threshold and short clock lower it.
Causal rival
Strategic procurement or cold-climate niches can generate announcements and demonstrations while LFP remains dominant on cost, bankability, and supply chain. The technology can succeed technically yet miss the cumulative grid-scale volume threshold.
Measurement risk
CNESA coverage may not expose a complete public project-level chemistry split, and vendor disclosures can double-count framework orders or hybrid capacity. Only commissioned sodium-ion cell capacity may count, with converted LFP and announced projects excluded.
Residual risk
The parent is not fully reproducible until the project ledger exists. Conditional admission is therefore appropriate: the binary and threshold are clear, but the evidence collection system must be built before resolution rather than reconstructed afterward.
Admission condition
Maintain a public project-level ledger with commissioned date, sodium-ion MWh, location, vendor, and independent corroboration before the 2028 parent adjudication; announcements and factory capacity cannot enter the total.
COLDTELL