Safety Is No Longer a Spec, It's a Survival Condition — A BMS Developer's Observation

When you build BMS for a living, you feel two forces pulling in opposite directions every day.

On one side is the customer. Customers generally want a product that's cheap, fast, and just clears the regulatory bar. I've lost count of how often I've been asked, "So it just needs to pass KC 62619, right?" Shorter lead times, lower costs, and a spec trimmed to whatever's written in the contract. From the buyer's seat, it's a pretty reasonable ask. If the product doesn't sell, there's no opportunity to think about safety in the first place.

On the other side are researchers and academia. Battery fires often happen without a clear party to blame — tiny defects inside a cell, minor external impacts, the long slow accumulation of aging. So researchers keep trying to design safer cells, and to catch anomalies at the BMS level a step earlier. Papers on flame-retardant electrolytes keep pouring in, gas-sensor-based early warning is being discussed seriously, and new versions of AI-based thermal runaway prediction models show up every year.

Between these two forces, the product a BMS developer ships leans a little more to one side each day. In my experience, that direction has almost always been downward. The spec document still has the word "safety" written somewhere, but what gets trimmed first in practice is usually the safety margin hiding behind it.

Lately, though, I've felt this dynamic starting to tilt back. At some point, the market, the regulations, and the economics of fire all started pointing in the same direction.

The Gravity of Cheap LFP

The most visible force is the gravity coming from below.

As of 2025, CATL took 39.2% of the global EV battery market, and BYD took 16.4%. The two together add up to 55.6%. Over the same period, Korea's three majors slid from a combined 16% down to around 12% (SNE Research). When a low-cost LFP offensive drags cell prices down, BMS cost expectations stacked on top of those cells naturally come down with them.

That's why the first line of an RFQ these days is almost always, "What can you do at this spec?" After that comes the spec sheet, and at the end, a single line: "compliance with relevant regulations." What's compressed into that one line is the minimum required to stamp a KC 62619 certificate on the product once. Words like functional safety level, redundant architecture, or independent monitor channel rarely appear in the body of the RFQ.

Under these conditions, the thing that gets cut first in a BMS quote is the safety margin no one sees. You merge separate protection circuits into one. You drop the redundant voltage measurement channel down to a single one. You simplify the SOC estimation algorithm. None of these decisions show up as changes anywhere in the delivery spec. The product keeps selling. Until it catches fire.

The Regulatory Floor Rises Every Year

Look at the other side and the story is quite different.

In Korea, KC 62619 Ed 2.0 was published in March 2023 (KTL notice). The 2019 first edition focused mostly on basic cell-level safety, but this revision adds thermal runaway propagation testing, system-level functional safety (system lock), and software safety assessment. The propagation test is literally this: force a single cell into runaway, and it must not spread to adjacent cells to pass. That's not a problem you solve by swapping cells — it's a problem that only unwinds when the pack and BMS are designed together.

Europe goes a step further. The EU Battery Regulation (EU) 2023/1542 came into force in February 2024. Due diligence obligations and CE marking kicked in in August 2024; EV battery carbon footprint declarations landed in February 2025; and in February 2027, the battery passport becomes mandatory. The passport isn't just traceability. Article 14 specifies that the BMS must provide SoH and remaining useful life data in a form third parties can read (EUR-Lex). The very definition of a BMS shifts — from a closed measurement and safety device to a data provider that interacts with external systems.

The EV side is more concrete. UN ECE R100-03 became mandatory for all vehicle categories in September 2025, making thermal runaway propagation testing a legal requirement. UN R156, since July 2024, has brought BMS firmware OTA updates into the scope of type-approval impact assessment. Pushing a single software update now goes through an approval process.

Put in one line: the short phrase "compliance with relevant regulations" is getting heavier every year. A design that squeaked over last year's bar is likely to miss it next year. The low-cost market and the regulation-compliant market are pulling apart fast.

The Arithmetic of a Single Fire

The third force is the economics of an accident.

The battery market has this property where a single incident can wipe out an entire supply contract. It sounds abstract when you describe it in words, but looking at the numbers makes it sharper.

The Samsung Galaxy Note 7 recall alone cost 5.3 trillion won in 2016, and roughly 31 trillion won in market cap evaporated (CNBC). A single battery design flaw discontinued an entire flagship line. Hyundai's Kona EV recalled 81,701 units between 2020 and 2021. The total cost of 1.4 trillion won was split 3:7 between Hyundai and LG Energy Solution (Megaeconomy). In the GM Bolt recall, LG reimbursed GM up to 1.9 billion dollars — roughly 2.5 trillion won (CNBC). A market where one cell defect comes back as a two-trillion-won liability.

The Mercedes EV fire in Cheongna, Incheon, in August 2024 showed this structure more vividly. A fire that began from a single car in an underground parking lot burned 87 vehicles to the ground and scorched another 783; 480 households were displaced. Estimated damage came to 100 billion won, with an 11.2 billion won fine from the Fair Trade Commission (Kyunghyang Shinmun). The onboard battery was from the Chinese maker Farasis, and after this incident, non-price barriers against Chinese batteries went up noticeably among Korean consumers. In November 2025, Tesla voluntarily recalled 10,500 Powerwall 2 units over fire risk (Electrek).

Line these numbers up and it's not hard to work out how a decision to shave a few dollars off a BMS comes back to corporate books in units of hundreds of billions of won. The upside on what you can save in a BMS has a ceiling; the downside on what a fire can cost has no floor. That asymmetry is probably the closest thing to the real essence of the battery market.

Academia Is Already Playing a Different Game

The final current is, for a developer, actually the most welcome one.

For anyone stuck between downward and upward pressures, what offers at least a little relief is that the tech coming out of academia is maturing faster than expected.

The center of gravity in early thermal runaway detection research has shifted over the past few years — from lagging signals like voltage and temperature toward leading indicators. Detecting electrolyte vapors like DMC and EMC with semiconductor gas sensors can raise an alarm minutes to tens of minutes before a voltage anomaly does (Frontiers in Chemistry, 2025). Real-time impedance tracking via Rapid EIS, and multimodal frameworks like T-RUNSAFE that layer spatiotemporal transformers such as ST-Former, are now right on the edge of commercialization (Nature Scientific Reports, 2025).

Hardware trends are moving in a similar direction. Lock-step dual-core MCUs like NXP S32K3, Infineon AURIX TC3xx, and Renesas RH850 have now become the default in BMS reference designs. Stacking independent monitoring ICs like TI BQ79616, ADI ADBMS6830, and NXP MC33775A on top of those to build redundant voltage and current channels is gradually taking hold as a standard. A combination that only lived in premium automotive BMS three or four years ago is now a reference you can pick straight from a catalog. Designs targeting ASIL-D functional safety are landing at reasonable prices at the component level.

Cell-side innovation is mostly on the BMS's side too. Flame-retardant electrolytes (Nano-Micro Letters, 2024), sulfide-based solid-state batteries (Samsung SDI and Toyota are both targeting mass production in 2027), and structural innovations like BYD's Blade LFP are changing the onset conditions of thermal runaway itself. As cells get safer, the BMS can drop the excess margin and allow faster charging and higher-energy operation. We're entering a phase where safety technology isn't holding performance back — it's lifting the ceiling.

A Market Where Safety Becomes the Moat

Step back and the four forces line up fairly cleanly in the same direction.

Below, the low-cost pressure is still strong. Customers still want the KC 62619 bare minimum, and Chinese LFP cell prices keep dropping. But above, the regulations are rising every year. KC 62619 Ed 2.0, EU Battery Regulation, UN R100-03, R156 stack up layer by layer, pulling the height of the bar up fast. Off to the side, a single fire creates costs in the hundreds of billions to trillions of won, and after the Cheongna fire alone Chinese batteries have run into a visible non-price barrier in the Korean market. And from behind, academia is handing developers far more refined safety technology, which will be part of reference designs within a few years.

The vector sum of those four forces is pretty clear. A high-safety BMS is no longer a premium option — it's quietly becoming a survival condition. Companies building their product line around regulation-minimum BMS today are likely to find it increasingly hard to get into the RFQ rooms of major customers a few years from now.

The battery market's peculiarity — "one fire and you're done" — is, from an engineer's point of view, closer to an advantage than a disadvantage. It's one of the rare industries where the safety axis maps directly onto competitive advantage. It's not a software race where shipping the first feature wins, and it's not a typical manufacturing race where a 1% cost difference decides the winner. In the battery market, "nothing catches fire" itself is a differentiator. And a fairly lucrative one.

When a customer says, "Just make it cheap, fast, and pass the regulatory minimum," what a BMS developer has to protect, quietly, is the contract three years past that request. The safety margin being shaved off today is what decides who's still standing in three years. Academia, regulation, and the market have already started moving in that direction together.