Korea Robot Component Compare — SPG (Reducers, PER 110×) vs. Halla Cast (Lightweight Structural Parts, PER 62×). Both Expensive — Which Carries Less Risk?

Tue, 12 May 2026 22:00:00 +0900

📚 Korea robotics series — Part 2. Part 1: Korea Robotics Value Chain — Samsung / Hyundai Motor / LG Moving in Parallel Part 3: Robotis vs. Rainbow Robotics — Humanoid Whole-Robot Compare Hub: Korea Humanoid / Robotics Investment Hub

SPG appeared 5 times in the Korea-robotics value-chain piece — supplying reducers to Boston Dynamics, Rainbow Robotics, Samsung, and LG. Market cap ₩3.0tn, PER ~110×. Halla Cast more recently won a humanoid-component order from a “global AI automaker.” Market cap ₩620bn, PER ~62×. Both are tagged as “robot names” — but the position in the value chain, the price, and the shape of the risk differ. Which is the better setup — more precisely, which carries less risk?

TL;DR

SPG = the robot’s “joint.” Makes reducers and actuators — the most core motion components in the value chain. Only Korean company producing harmonic, planetary, and RV reducers all in-house. Market cap ~₩3.0tn, 2026E PER ~110×.
Halla Cast = the robot’s “skeleton.” Magnesium / aluminum die-casting for lightweight structural parts and thermal-management components. In 2025 won a ₩12.4bn humanoid-component order from a “global AI automaker.” Market cap ~₩620bn, 2026E PER ~62×.
Read. SPG is the more core robot asset, but the current price has already absorbed that core status aggressively. Halla Cast is less core, but its price burden is lower and there is still verification ahead. Both are expensive, but the reason and the risk shape differ.

1. First — why the parts they make are different

1.1 SPG reducers — the robot joint’s core

As explained in the Korea robotics value-chain piece, reducers convert fast motor rotation into slow, strong torque. Every robotic joint (shoulder, elbow, wrist, hip, knee, ankle) uses one. A humanoid contains dozens of reducers.

Why reducers are expensive and hard:

Precision: gears must mesh with sub-micron tolerance
Durability: joints must move tens of thousands of cycles without play
Variety: harmonic, RV, planetary — each application uses different types

Global #1: Japan's Harmonic Drive (decades of effective monopoly)
Korea's only in-house challenger: SPG (produces all three types)

SPG additionally produces actuators (motor + reducer + sensor integrated module). Its proprietary SDD (Smart Direct Drive) actuator targets a 5,000-unit sales goal in 1H 2026 — a strategy to step up from “reducer supplier” to “reducer + actuator integrated-solution supplier.”

1.2 Halla Cast structural parts — the robot skeleton’s lightweighting

Halla Cast makes precision metal parts by die-casting molten magnesium and aluminum. Originally focused on auto sensor housings, display frames, and EV inverter cases, the company is now expanding into humanoid structural frames, drive-mechanism peripherals, and thermal-management components.

Why a light skeleton matters:

Heavy arms, legs, torso for a robot:
→ require bigger motors (more power)
→ drain batteries faster (shorter runtime)
→ generate more heat (shorter component life)

Magnesium is 33% lighter than aluminum:
Magnesium density: 1.8 g/cm³
Aluminum density: 2.7 g/cm³
→ identical volume, significantly lighter part

Magnesium is also harder to work with:
→ oxidation / ignition risk, corrosion susceptibility, machining difficulty
→ holding yield in mass production is the core capability

Halla Cast’s technical edge is high-vacuum die-casting (eliminates porosity) and thin-wall die-casting (large-area parts at minimal thickness). The platform was validated in auto electronics; humanoid structural parts are now the extension.

1.3 In one line

SPG: the part that makes a robot move (joints = reducers)
Halla Cast: the part that keeps a robot light (skeleton = structural frame)

Without SPG, a robot does not move.
Without Halla Cast, a robot becomes heavier.

Both are needed, but "absolutely necessary" applies more strongly to SPG.
Hence SPG trades more expensively.

2. The numbers — how far apart

May 12 close basis.

Item	SPG	Halla Cast
Price	₩135,100	₩16,900
Market cap	~₩3.0tn	~₩620bn
2025 revenue	₩341.7bn	₩155.9bn
2025 OP	₩17.9bn	₩10.3bn
2025 OPM	5.2%	6.6%
2026E revenue	~₩418.6bn	~₩181.4bn
2026E OP	~₩27.1bn	~₩14.6bn
2026E OPM	~6.5%	~8.0%
2026E PER	~110×	~62×
2026E PSR	~7.2×	~3.4×
Value-chain position	Joints (reducers / actuators)	Skeleton (structure / thermal / lightweighting)
Mentions in value-chain piece	5× (most cited)	1-2×
Robot share of backlog	Undisclosed (assumed small)	₩12.4bn (1.1% of total)

2.1 What the table shows

One: SPG is 4.8× more expensive than Halla Cast. ₩3.0tn vs. ₩620bn. PER 110× vs. 62×.

Two: Halla Cast has the higher operating margin. SPG 5.2-6.5% vs. Halla Cast 6.6-8.0%. SPG’s mix blends fan motors (OPM 2-3%) and geared motors (OPM ~10%), pulling the consolidated number down. Precision robot reducers are still only ~3% of revenue.

Three: SPG’s premium is paid for “core-ness.” Reducers account for 30-40% of robot bill-of-materials, and SPG is the only Korean producer of all three reducer types. Halla Cast’s structural parts matter, but the level of “irreplaceable” is lower than reducers.

3. SPG — better robot asset, but a price problem

3.1 Why it’s core

SPG is the only company that appeared 5 times across the value-chain piece — Boston Dynamics (Spot, Atlas), Rainbow Robotics, Samsung Electronics, LG Science Park, all source reducers from SPG. Practically every major Korean robot platform.

The SDD-actuator launch (5,000-unit 1H26 target) is the strategic move from “reducer supplier” to “integrated drive-solution supplier.”

3.2 Why the price is heavy

The numbers:

SPG valuation:
Market cap: \~₩3.0tn
2025 OP: ₩17.9bn
2026E OP: \~₩27.1bn
PER: \~110×
PBR: \~11.6×

Cross-check:
Market cap ₩3.0tn ÷ OP ₩27.1bn = \~111×
→ \~111 years of current operating profit pre-paid

PER 110× embeds “Korean robot industry will explode and SPG’s earnings will multiply by tens.” That expectation has to deliver. If it doesn’t, the multiple compresses sharply.

Specifically, precision robot reducers are still ~3% of SPG’s revenue. The other 97% is industrial motors / reducers. The name “robot core-part play” is the marketing frame; the financials still read like an “industrial motor company.”

3.3 Analytical position

Good company, not good price. The reducer-core thesis is real. But the price has substantially front-run the “global humanoid orders → earnings explosion” scenario. Multiple compression resolves through one of two paths: (a) SDD-actuator real sales cleanly clearing the 5,000-unit guidance, or (b) a major global humanoid order officially confirmed. Without one, PER 110× is “supported by expectation alone.”

4. Halla Cast — less core, lighter price, still has verification

4.1 What’s driving the attention

In May 2025, Halla Cast won a humanoid-robot-component order for 3 SKUs totaling ₩12.4bn from a “global AI automaker.” An additional ~10 SKUs are reportedly under discussion, with full mass-production targeted for 2H 2026.

The market interprets “global AI automaker” as Tesla. But no public source confirms it is Tesla. The conservative analytical framing: not a “confirmed Tesla supplier” but a “humanoid-component option play tied to a global AI automaker.”

4.2 The backlog mix

Halla Cast’s end-2025 backlog is ₩1,134.6bn — large. But the composition is:

Item	Amount	Share
Display	₩641.9bn	56.6%
Autonomous driving	₩350.6bn	30.9%
EV inverter	₩95.6bn	8.4%
Robot	₩12.4bn	1.1%
Battery	₩11.2bn	1.0%
Other	₩22.9bn	2.0%

Robot is 1.1% of total backlog. Halla Cast is still a display / autonomous-driving auto-electronics supplier that has just begun adding humanoid parts. Robotics is small but optionality.

The option’s value depends on whether the ~10 SKUs under discussion convert. ₩12.4bn growing to ₩50.0bn+ would reclassify the company as a “robot component play.” Without that conversion, it stays “expensive auto-electronics supplier.”

4.3 Where Halla Cast scores better than SPG

Lower price burden. PER 62× is still rich but half SPG’s 110×.

Higher and improving operating margin. Halla Cast’s OPM trajectory: 6.6% (2025) → 7.8% (2026E) → 8.4% (2027F). SPG progresses more slowly from 5.2% to 6.5%.

Defensive base business. Halla Cast’s non-robot backlog of >₩1tn protects against scenarios where robotics doesn’t deliver. SPG also has industrial-motor base business, but the high stock price means “robotics not materializing → share-price drawdown.”

4.4 Where Halla Cast scores worse

Lower value-chain core-ness. Reducers are mandatory for a robot to move. Structural die-cast parts are important but more substitutable — another die-caster could in principle provide them.

No customer confirmation. SPG’s Rainbow / Boston Dynamics relationships are documented. Halla Cast’s “global AI automaker” designation has not been officially identified.

5. Side-by-side — which carries less risk

Item	SPG	Halla Cast	Edge
Robot core-ness	Reducers / actuators (joints)	Structure / thermal / lightweighting (skeleton)	SPG
PER	~110×	~62×	Halla Cast
PSR	~7.2×	~3.4×	Halla Cast
OPM	5.2-6.5%	6.6-8.0%	Halla Cast
Earnings growth (2026E)	+70%	+112%	Halla Cast
Value-chain mentions	5×	1-2×	SPG
Customer verification	Rainbow / BD confirmed	“Global AI automaker” (unconfirmed)	SPG
Base-business safety net	Industrial motors (exists but price-protective effect weaker)	Display / autonomous backlog >₩1tn	Halla Cast
Pricing already-in	Very heavy	Heavy but relatively less	Halla Cast

Relatively, Halla Cast carries less risk. SPG is the better robot asset, but its current price has absorbed that quality too aggressively. Halla Cast is less core but priced less full, with a thicker base-business safety net.

That said, “less risky” ≠ “cheap.” Halla Cast at PER 62× is rich for an auto-parts supplier. Without further robot-order expansion, the multiple is hard to defend.

6. Fair value — what’s adequate for each

6.1 SPG

Scenario	Logic	Applied PER	Implied price range
Current (over-anticipation)	Robot revenue 3%, multiple driven by expectation	110×	₩135,100 (now)
Reasonable entry	Confirmed global order + multiple compresses	70-80×	~₩86,000-₩98,000
Sharp-decline interest	SDD volumes + major order delays	50-60×	~₩61,000-₩74,000

At ₩135,100 the price embeds a meaningful share of the “global humanoid order → revenue explosion” scenario. Multiple normalization requires fundamental evidence — SDD-actuator real volumes or a confirmed major global customer order.

6.2 Halla Cast

12-month forward EPS (average of 2026E and 2027F) ≈ ₩333 basis:

Scenario	Applied PER	Implied price	Vs. current
Bear (robot orders stall)	35×	~₩11,700	-31%
Base (current order holds + base business grows)	55×	~₩18,300	+8%
Bull (robot SKU expansion + mass production)	70×	~₩23,300	+38%

Current ₩16,900 sits slightly below base ₩18,300. If robot backlog expands to ₩50.0bn+, the bull-case ₩23,300 opens.

7. What to verify next

7.1 SPG

SDD-actuator actual unit sales (target 5,000)
Confirmed global humanoid customer orders (currently Rainbow / BD anchored)
Precision robot-reducer revenue share (currently ~3% → potential 10%+)
2H 2026 results showing recognized robotics revenue

7.2 Halla Cast

Robot-backlog evolution (₩12.4bn → ₩50.0bn+ would shift classification)
~10 additional SKUs under discussion converting to firm orders
2H 2026 robot mass-production revenue recognition
Margin restoration (2025 6.6% → 2026 8%+)
New plant yield / utilization stabilization

7.3 Joint

Tesla Optimus mass-production timeline (Fremont line currently in “Construction”)
Samsung Electro-Mechanics 2H humanoid camera-module mass production → ecosystem-wide signal
Hyundai Motor Group 3Q RMAC operation → Boston Dynamics Atlas supply-chain acceleration

8. Bottom line

SPG is the “crossroads” of Korea’s robot value chain — the only supplier of reducers to Boston Dynamics, Rainbow Robotics, Samsung, and LG. But ₩3.0tn at PER 110× — for a company where robot revenue is still 3% — requires “everything to go right” for that price to feel justified.

Halla Cast is less core than SPG — structure / thermal / lightweighting rather than reducers. But ₩620bn at PER 62× is half the multiple, with a >₩1tn non-robot base-business safety net. If the robot backlog expands from ₩12.4bn to ₩50.0bn+, re-rating room opens.

Both are expensive. The reason for being expensive differs. SPG: “core part, so expensive.” Halla Cast: “option attached, so expensive.” SPG risk: “if expectations don’t realize, sharp drawdown.” Halla Cast risk: “if option doesn’t realize, reverts to auto-parts multiple.”

FAQ

Q: What’s the most important difference between SPG and Halla Cast? A: Their position in the robot value chain. SPG is the “joint” (reducers); Halla Cast is the “skeleton” (structural frame). Reducers are non-substitutable; structural parts can be sourced from other die-casters. That asymmetry is why SPG carries a higher multiple.

Q: Can SPG’s PER 110× normalize? A: Two paths. (1) Earnings grow fast enough that the multiple compresses naturally — current OP ₩27.1bn rising to ₩80bn+ by 2027-28 would bring PER toward 30-40× — possible if global humanoid mass-production accelerates. (2) Price drops to compress the multiple — if SDD or the major order is delayed. The timing of which path materializes first is the variable.

Q: Is the “global AI automaker” Tesla? A: The market reads it that way, but no public source confirms it. The company filing only says “global AI automaker.” Analytically: treat it as a “global-AI-automaker-tied option play,” not a “confirmed Tesla supplier.” Tesla-confirmation would add a re-rating leg on top of the current premium.

Q: If robotics is only 3% of SPG’s revenue, what’s the other 97%? A: Industrial motors and reducers — appliance motors (air-con / washer), industrial geared motors, fan motors. Stable but single-digit-growth base business. Robotics is the “growth accelerant” on top.

Q: Is Halla Cast’s ₩1tn backlog a real safety net? A: Partially. Display (56.6%) + autonomous (30.9%) = 87.5% of backlog is auto-electronics base business. That means “if robotics doesn’t deliver, the stock reverts to an auto-parts multiple.” Useful protection. But defending PER 62× requires the robotics option to work.

Q: Should one avoid both and look at other robot names? A: Reasonable alternative. The robotics value-chain piece covers other component suppliers (HL Mando, Samsung Electro-Mechanics, LG Innotek, Hyundai Mobis) and pure-play robotics names (Rainbow Robotics, Doosan Robotics). Pure-plays carry heavier price burden; large component suppliers carry tiny robot-revenue share, so robotics catalysts have limited share-price effect.

Q: What signals would justify owning both? A: Three simultaneously: (1) Tesla Optimus or Boston Dynamics Atlas formally entering mass production; (2) Korean component suppliers showing recognized robot revenue in their P&L (quarterly robot revenue ≥₩10bn); (3) global humanoid market-size guidance revised higher. All three together would re-rate Korea robot components broadly.

This article is for research and informational purposes only and does not constitute investment advice. SPG data references Samsung Securities, KB Securities, and Hana Securities reports. Halla Cast data references NH Investment & Securities, Korea Investment & Securities, and KB Securities reports. Prices reflect Maeil Business Newspaper data (May 12). Portions of SPG 2026 consensus and Halla Cast’s actual customer identity are not publicly verified. The market’s “global AI automaker = Tesla” inference is unofficial. Fair-value scenarios are analyst estimates and may be wrong. Korean robotics remains industry-early with uncertain mass-production timelines. Analysis can be wrong. Data cut: May 12, 2026 KST.

Disclaimer: For research and information purposes only. Not investment advice. Names cited are for analytical illustration; readers should perform their own due diligence and consult licensed advisors before any investment decision.

Die Casting on Korea Invest Insights