Interesting Things I Read This Week
Battery chemistry & storage
Solid-state hype ≠ near-term scale.
Takeaway: China’s “solid-state rally” is real in markets, but the tech path is mostly high-silicon anodes and incremental wins—not lithium-metal at scale.
Move: De-risk roadmaps around hybrid solid-state cells for premium niches; keep mass storage on LFP/LMFP. voltrush.substack.com+1Yield beats chemistry.
Takeaway: Europe’s gigafactory prospects hinge more on manufacturing yield than on picking the “right” cathode.
Move: Tie incentives to first-pass yield, scrap rate, and OEE—not just nameplate GWh. christopherchico.substack.comSodium-ion’s edge is OpEx, not energy density.
Takeaway: For grid-scale, sodium’s capex/OpEx profile can beat LFP even with lower Wh/kg.
Move: Pilot Na-ion for daily-cycle BESS; reserve Li for mobility/space-constrained uses. cleantechies.substack.com+1China’s storage surge will set global price floors.
Takeaway: New energy-storage targets and booming shipments in China will compress global BESS pricing.
Move: Bake price-learning curves into tenders; consider shorter contract tenors to capture step-downs. voltrush.substack.com+1Policy contrarian: the U.S. is over-weight extraction, under-weight factories.
Takeaway: “Upside-down” strategy: billions to mining, too little to process/production know-how.
Move: Rebalance toward process tech + yield tooling + workforce; fund direct-recycling pilots tied to OEM take-back. christopherchico.substack.com
Robotics & embodied AI
Watch Amazon’s quiet robotics stack.
Takeaway: Amazon’s logistics bets point to where unit economics work first (manipulation + movement under controlled conditions).
Move: Prioritize pilots in structured warehouses and ports before street/public-space robots. ontheseams.substack.com
Non-U.S. AI stacks, sovereignty, and geopolitics
Definition — Countries adopting AI “end-to-end” without U.S. chips or hyperscalers.
What happened — Malaysia stood up a sovereign stack pairing DeepSeek with Huawei Ascend as a national service, a template many ministers are now eyeing. Developing Telecoms+1
Why it matters — It proves a full non-U.S. stack can be packaged as policy, not just a tech choice.
What to watch — Africa copycats: look for explicit naming of Ascend + DeepSeek in MOUs, not just “AI strategy” rhetoric.Definition — Freemium model access plus low inference cost to seed usage.
What happened — Bloomberg’s Africa coverage: Huawei Cloud pitches DeepSeek with very generous free daily token quotas and undercuts U.S. rivals on price. Video + feature recap the pitch startups are hearing. Bloomberg+1
Why it matters — For ministries and SOEs, the TCO story is suddenly compelling even before performance parity.
What to watch — Formal, public Africa region pricing pages and SLAs from Huawei Cloud to move from pilots to procurement.Definition — Use existing carrier + local DC footprint to sell AI platforms.
What happened — Johannesburg workshops and the Tech Wave DeepSeek roadshow market an “end-to-end stack powered by DeepSeek and R1.” Reuters underlines Huawei Cloud’s 16× SA growth since 2019 and the first local hyperscale DC. itweb.co.za+1
Why it matters — Faster enterprise sales when cloud sits near the data, with carrier relationships already in place.
What to watch — Kenya/Nigeria/ Ghana: carrier-backed AI days that mention DeepSeek by name.