Key findings
— H1 2026 R&D spend totalled KRW 3.08tn across 252 companies, up +15.3% among the 204 companies comparable year on year.
— Over 2024–2025, 260 companies spent KRW 11.14tn on R&D; the same period produced 1,869 approvals, 507 Korean trial authorisations, 231 global trial starts and 1,082 publications.
— Spend is concentrated: the top 10% (26 companies) account for 61.1% of R&D, 65.8% of global trial starts and 23.3% of approvals.
— Output takes different forms by sector: per KRW 10bn, approvals are highest in Medical devices (4.7), Korean trials in Pharma & CDMO (0.64) and publications in Medical AI (9.3).
1. H1 2026 R&D spend — the top ten
The ten largest spenders account for KRW 1.31tn, 42.5% of the total. Celltrion, Inc. leads with KRW 284.3bn.
| Rank | Company | H1 2026 (KRW bn) | H1 2025 (KRW bn) | Change |
|---|---|---|---|---|
| 1 | Celltrion, Inc. | 284.3 | 229.7 | +23.8% |
| 2 | LigaChem Biosciences Inc. | 137.1 | 77.3 | +77.4% |
| 3 | Hanmi Pharm. Co., Ltd. | 125.5 | 106.2 | +18.2% |
| 4 | SK Bioscience Co.,Ltd | 122.4 | 68.0 | +80.0% |
| 5 | YUHAN CORPORATION | 122.2 | 107.3 | +13.8% |
| 6 | DAEWOONG PHARMA | 115.7 | 106.6 | +8.5% |
| 7 | Chong Kun Dang Pharmaceutical Corp. | 113.5 | 83.1 | +36.6% |
| 8 | SAMSUNG EPIS HOLDINGS Co., Ltd. | 106.0 | — | — |
| 9 | SK biopharmaceuticals Co., Ltd. | 97.3 | 81.6 | +19.2% |
| 10 | GC Biopharma Corp. | 85.9 | 82.7 | +3.8% |
"—" marks companies whose consolidation scope changed (e.g. the November 2025 demerger), so no like-for-like comparison is possible.
2. Fastest growth
Companies with at least KRW 2.5bn of R&D in H1 2025, ranked by growth multiple.
| Rank | Company | H1 2025 (KRW bn) | H1 2026 (KRW bn) | Multiple |
|---|---|---|---|---|
| 1 | SAM CHUN DANG PHARM CO. LTD | 4.4 | 21.7 | 4.9x |
| 2 | Mezzion Pharma Co., Ltd. | 6.9 | 15.8 | 2.3x |
| 3 | Voronoi, Inc. | 21.4 | 41.7 | 2.0x |
| 4 | FutureChem Co., Ltd. | 2.5 | 4.8 | 1.9x |
| 5 | SK Bioscience Co.,Ltd | 68.0 | 122.4 | 1.8x |
| 6 | LigaChem Biosciences Inc. | 77.3 | 137.1 | 1.8x |
| 7 | JEIL PHARMACEUTICAL CO.,LTD | 21.3 | 35.9 | 1.7x |
| 8 | Qurient Co., Ltd. | 10.0 | 16.5 | 1.6x |
| 9 | Orum Therapeutics, Inc. | 21.7 | 35.2 | 1.6x |
| 10 | SillaJen, Inc. | 5.2 | 8.4 | 1.6x |
3. R&D intensity by sector
Median R&D-to-revenue ratios: medical AI 31.5%, biotech 29.0%, pharma 5.9%. Sectors with more pre-revenue developers run higher ratios.
Medians of company-disclosed ratios; values above 1,000% (near-zero revenue) are excluded.
4. Two years of spend and disclosed output, by sector
Pharma & CDMO (99 companies) accounts for 60.0% of 2024–2025 R&D. Per KRW 10bn, the leading sector is Medical devices for approvals, Pharma & CDMO for Korean trials, Medical AI for global trials, Medical AI for publications and Medical AI for patents.
| Sector | Companies (n) | R&D (KRW bn) | approvals (n) | Korean trials (n) | global trials (n) | publications (n) | patents (n) |
|---|---|---|---|---|---|---|---|
| Pharma & CDMO | 99 | 6,684.1 | 1,091 | 431 | 173 | 384 | 1,043 |
| Biotech & drug discovery | 74 | 3,110.1 | 178 | 59 | 37 | 319 | 917 |
| Medical devices | 68 | 1,113.8 | 525 | 14 | 14 | 198 | 977 |
| Medical AI | 12 | 182.1 | 49 | 0 | 6 | 169 | 170 |
| Healthcare infrastructure | 7 | 50.2 | 26 | 3 | 1 | 12 | 83 |
| Sector · per KRW 10bn | approvals (n) | Korean trials (n) | global trials (n) | publications (n) | patents (n) |
|---|---|---|---|---|---|
| Pharma & CDMO | 1.63 | 0.64 | 0.26 | 0.57 | 1.71 |
| Biotech & drug discovery | 0.57 | 0.19 | 0.12 | 1.03 | 3.27 |
| Medical devices | 4.71 | 0.13 | 0.13 | 1.78 | 9.65 |
| Medical AI | 2.69 | 0.00 | 0.33 | 9.28 | 11.57 |
| Healthcare infrastructure | 5.18 | 0.60 | 0.20 | 2.39 | 15.11 |
| All | 1.68 | 0.46 | 0.21 | 0.97 | 3.15 |
Sector leaders are chosen among sectors with at least 10 companies — small denominators swing. Patents are published 18 months after filing, so 2023–2024 filings are divided by 2023–2024 R&D.
5. Concentration
The top 10% of spenders (26 companies) account for 61.1% of two-year R&D. They also hold 65.8% of global trial starts and 40.8% of Korean trial authorisations — the costlier the stage, the more concentrated. Approvals (23.3%) and patents (22.6%) are broadly spread.
6. Approvals and trials per KRW 10bn
Companies with at least KRW 5bn of two-year R&D, ranked by approvals (top) and Korean trial authorisations (bottom) per KRW 10bn. This is not an efficiency measure — a generic and a new drug, or a Class 2 certification and a Class 4 approval, each count as one. It shows what form of public record the spend took.
| Rank | Company | Two-year R&D (KRW bn) | approvals (n) | Per KRW 10bn (n) |
|---|---|---|---|---|
| 1 | SINSIN PHARMACEUTICAL Co., LTD. | 5.3 | 24 | 44.9 |
| 2 | DongBang Medical Co., Ltd. | 5.9 | 24 | 40.8 |
| 3 | Precision Biosensor Inc. | 6.2 | 20 | 32.4 |
| 4 | KOREA ARLICO PHARM CO.,LTD. | 10.1 | 30 | 29.8 |
| 5 | CMG Pharmaceutical Co., Ltd. | 17.6 | 52 | 29.6 |
| 6 | SHINIL PHARMACEUTICAL CO., LTD. | 8.6 | 20 | 23.2 |
| 7 | Coreline Soft Co., Ltd | 6.1 | 14 | 23.1 |
| 8 | Yuyu Pharma, Inc. | 8.2 | 15 | 18.3 |
| 9 | DIO Corporation | 9.3 | 17 | 18.2 |
| 10 | DONGSUNG PHARMACEUTICAL CO.,LTD | 8.3 | 14 | 16.8 |
| Rank | Company | Two-year R&D (KRW bn) | Korean trials (n) | Per KRW 10bn (n) |
|---|---|---|---|---|
| 1 | VIVOZON PHARMACEUTICAL CO., LTD. | 6.9 | 10 | 14.5 |
| 2 | KOREA ARLICO PHARM CO.,LTD. | 10.1 | 13 | 12.9 |
| 3 | THERAGEN ETEX CO.,LTD. | 10.4 | 8 | 7.7 |
| 4 | Yuyu Pharma, Inc. | 8.2 | 6 | 7.3 |
| 5 | HLB PHARMACEUTICAL CO., LTD. | 11.4 | 8 | 7.0 |
| 6 | SEOUL PHARMA CO.,LTD | 5.4 | 3 | 5.6 |
| 7 | REYON PHARMACEUTICAL CO., LTD | 7.7 | 4 | 5.2 |
| 8 | SINSIN PHARMACEUTICAL Co., LTD. | 5.3 | 2 | 3.7 |
| 9 | DONGSUNG PHARMACEUTICAL CO.,LTD | 8.3 | 3 | 3.6 |
| 10 | Daehan Nupharm Co.,Ltd. | 16.8 | 6 | 3.6 |
7. Publications per KRW 10bn
On the same basis, PubMed-indexed publications per KRW 10bn. Medical AI companies feature prominently.
| Rank | Company | Two-year R&D (KRW bn) | publications (n) | Per KRW 10bn (n) |
|---|---|---|---|---|
| 1 | 3billion Inc. | 7.7 | 64 | 83.5 |
| 2 | Optipharm.CO.,LTD | 8.9 | 27 | 30.2 |
| 3 | Future Medicine Co., Ltd. | 6.7 | 19 | 28.5 |
| 4 | Coreline Soft Co., Ltd | 6.1 | 16 | 26.4 |
| 5 | NEUROPHET Inc. | 12.7 | 30 | 23.6 |
| 6 | VUNO, Inc. | 11.2 | 22 | 19.7 |
| 7 | HEM Pharma Inc. | 7.1 | 10 | 14.2 |
| 8 | Oncocross Co.,Ltd. | 6.3 | 8 | 12.8 |
| 9 | ORGANOIDSCIENCES LTD. | 11.0 | 13 | 11.9 |
| 10 | DEEPNOID Inc. | 12.3 | 14 | 11.4 |
8. Companies with no disclosed output
10 companies with KRW 58.7bn of two-year R&D had no approval, trial, publication or patent disclosed over the same period (Pharma & CDMO 3 · Medical devices 3 · Medical AI 2 · Healthcare infrastructure 1 · Biotech & drug discovery 1). Many are early-stage or file through overseas partners, so they are not named.
Method and limits
R&D spend — the "total R&D expense" line of half-year and annual filings, before government grants, consolidated where available. Only values that pass a cross-check (e.g. against the disclosed ratio) are used: a verified census, not a sample.
Non-healthcare companies excluded — 19 companies whose company-wide R&D is mostly unrelated to healthcare: AHNLAB,INC. (security software) · Chunbo Co., Ltd. (battery & display materials) · CJ CHEILJEDANG CORP. (food) · COSMAX BTI, INC. (cosmetics holding company) · COSMAX NBT, INC. (dietary supplements) · HLSCIENCE Co., LTD (dietary supplements) · HYUNDAI ELEVATOR CO.,LTD (elevators) · INTOPS CO., LTD. (handset & appliance parts) · KOLMAR KOREA CO.,LTD (cosmetics contract manufacturing) · Kwangdong Healthbio Co., Ltd. (dietary supplements) · NOVAREX Co., Ltd. (dietary supplements) · NUON CO., LTD. (dietary supplements) · PAKERS CO., Ltd. (printer parts & lighting) · RN2 Technologies Co., Ltd. (telecom components) · S&D Co., Ltd (supplement ingredients) · SBSUNGBO Co., Ltd. (agrochemicals) · SK Innovation Co., Ltd. (refining & batteries) · WonIk Co.,LTD (quartz & semiconductor parts) · WOOREE BIO CO., LTD (dietary supplements).
Holding companies — 10 holding companies that consolidate listed subsidiaries are excluded to avoid double counting: DAEWOONG CO.,LTD · CHONGKUNDANG HOLDINGS CORP. · GC Corp. · JW HOLDINGS CORPORATION · JEIL PHARMA HOLDINGS INC. · ILDONG HOLDINGS CO.,LTD · Huons Global Co., Ltd. · Dong-A Socio Holdings Co.,Ltd · Hanmi Science Co.,Ltd · SK Chemicals Co., Ltd.. Samsung Epis Holdings, whose subsidiary is unlisted, is included.
Year-on-year comparison — only companies with both half-year values on the same consolidation basis. Samsung Biologics (scope changed by the November 2025 demerger) and Samsung Epis Holdings (no prior-year value) are excluded from the comparison.
Output — approvals: MFDS Class 2 device certifications, Class 3–4 device approvals and drug approvals, plus US FDA 510(k) and drug approvals. Class 1 devices, which are only notified without review (242 in the period), are not counted as approvals. Korean trials: MFDS trial authorisations. Global trials: ClinicalTrials.gov start dates. Publications: PubMed author affiliation. Patents: KIPRIS filings, domestic and foreign. Window 2024–2025; patents 2023–2024.
Matching and subsidiaries — records are matched by business registration number or exact corporate name; similarity matching is never used. Approvals, trials and patents filed by subsidiaries whose control is confirmed in filings are counted under the listed parent. Publications are counted by the listed company's own author affiliation only and exclude subsidiary-affiliated papers — except Samsung Epis Holdings, a holding company whose research is carried out entirely by its wholly owned subsidiary Samsung Bioepis, which is counted by that affiliation.
Not yet linked — patents filed by Samsung Bioepis, the wholly owned subsidiary of Samsung Epis Holdings, are not included in this edition. KIPRIS query limits allow only a monthly collection; they will be linked in the October 2026 refresh. Sector patent totals are correspondingly understated.
Direction of error — approvals filed by overseas affiliates or partners, CRO-sponsored trials, records of unconfirmed subsidiaries and papers published under a subsidiary's affiliation are not captured. Every known error undercounts output.
Verification — figures for ranked companies were re-queried against the source. Reported errors are recomputed and corrected with a visible history.