The U.S. AI-Biotech Rally: Why Proteina, GC Genome, and ST Pharm Need Separate Cases in Korea

On August 19, Moderna and Merck announced that their personalized mRNA cancer therapy met key endpoints in a Phase 3 trial. Through the October 1 close, Moderna shares had gained 156% over the past 60 trading sessions. Over the same period, Twist Bioscience rose 110%, Natera 49%, and Tempus 34%.¹²

There are links to explore in Korea as well. Proteina has an experimental-validation memorandum of understanding with Twist. GC Genome sells an AI-based blood test for cancer. ST Pharm has RNA-therapy manufacturing infrastructure. But these companies connect to the theme in different ways, and their profits will emerge on different timelines.

Proteina has the clearest business link, GC Genome combines liquid biopsy with recent buying flows, and ST Pharm has the clearest RNA-manufacturing results. Based on the evidence available, no candidate has verified business relevance, operating results, medium-term trend, and attractive valuation all at once. Seers remains a profitability benchmark for medical AI, while Lunit is a growth candidate in cancer AI platforms.

Prices and market capitalizations are as of the October 1, 2026 close. Korean financial results are first-half 2026 year-to-date figures. Business prospects are assessed over the next 6–12 months. Price returns are calculated separately for Korea’s 21 and 63 trading sessions and the U.S. 20 and 60 sessions. The candidates below were selected from a review of 20 companies. The list identifies research candidates and does not constitute a buy recommendation.

The U.S. rally is backed by clinical results, testing volumes, and demand for experiments

All four U.S. companies are grouped under AI and biotech, but they earn money in different ways.

Price returns compare the October 1 close with the September 2 and July 8 closes. They are not total returns including dividends. All four stocks rose over the recent periods but fell on October 1. Medium-term gains and short-term chase risk both matter.²

Moderna’s INTerpath-001 Phase 3 trial met its recurrence-free survival and distant metastasis-free survival endpoints. Overall-survival assessment and regulatory steps remain. The risk-reduction result from Phase 2b should not be read as a Phase 3 result. Lower clinical risk for one therapy does not prove that all AI drug-development companies will succeed.¹

Tempus reported second-quarter 2026 revenue of $382.5 million, up 22% year over year. Data licensing and modeling revenue rose 36%, and adjusted EBITDA was $8.04 million. Paid testing and data operations are translating into actual revenue.³

Twist reported June-quarter revenue of $118.4 million, up 23%, with a gross margin of 52.8%. On September 16, it announced an agreement to provide antibody-characterization services to users of Eli Lilly’s TuneLab. Its business has exposure to demand for experimentally validating candidates designed by AI.⁴⁵

Natera reported second-quarter revenue of $752.8 million, up 37.7% year over year. Oncology test volumes increased 57.2%, and gross margin was 64.5%. Repeat testing and insurance reimbursement are important growth drivers.⁶

These results suggest that clinical success, repeat testing, and demand for experimental infrastructure have all gained attention. They do not establish that Moderna’s rally caused the other three companies’ gains. Nasdaq’s announcement that Moderna would join the Nasdaq-100 on October 9 is also a company-specific U.S. flow event. There is no basis to assume the same capital will flow into Korean peers.⁷

Korean peers should be matched by product and monetization model

Instead of linking companies by name alone, compare which process and customer generate revenue.

MRD means minimal residual disease, which testing seeks to detect and monitor after treatment. MCED means multi-cancer early detection. They differ in target population, testing frequency, clinical evidence, and reimbursement pathway. A “Korean Natera” label alone does not justify applying Natera’s revenue multiple.⁸⁹

Proteina is linked to Twist, but paid orders are the next evidence

Proteina’s SPID platform observes protein-protein interactions at the single-molecule level. Antibodies designed by AI still need real experiments to confirm that they bind the intended target and work as designed. Proteina’s opportunity is to feed those validation data back into the next design cycle.¹⁰

On August 27, the company announced an MoU with Twist for large-scale experimental validation. The collaboration links Twist’s synthesis and expression capabilities with Proteina’s validation and data production. In July, Proteina also announced a joint technology-development and technology-transfer option agreement with Samsung Bioepis.¹¹¹²

The company says it plans to expand data-production capacity from 40,000 to 1 million antigen-antibody interactions per month by 2028. This is a capacity target. No minimum order volume, purchase consideration, revenue commitment, or contracted sales schedule has been disclosed. Samsung Bioepis’s options and milestones also depend on conditions being met.¹¹¹²

There is a wide gap between current results and market expectations. Revenue was KRW 2.95 billion in 2025 and KRW 520 million in the first half of 2026, down 74.4% from the prior-year half. The first-half operating loss was KRW 8.67 billion. The October 1 market capitalization of KRW 569.7 billion was about 193 times last year’s revenue.¹⁰¹³

Revenue mix matters too. SPID Systems equipment generated KRW 2.07 billion of the KRW 2.95 billion total in 2025. In the first half of this year, the company recorded KRW 100 million from equipment, KRW 130 million from PathFinder, and KRW 290 million from Landscape. Because equipment and project revenue can be lumpy, the overall decline does not by itself prove that demand for repeat services has collapsed. Conversely, the current scale does not prove that recurring monetization is sufficiently established.¹⁰

At the end of June, Proteina had KRW 2.09 billion in cash, plus KRW 14.26 billion in short-term deposits and about KRW 3.03 billion in fair-value financial assets. A liquidity assessment based only on the cash line would be incomplete. But the company spent KRW 3.42 billion to acquire an ABX stake in July, after the reporting date. June-end financial assets should not be treated as the October balance.¹⁰¹⁴

Proteina is a high-risk growth candidate with a strong business link to the theme. The next evidence should be paid validation contracts, repeat orders, revenue and profit per validation project, and a recovery in quarterly service revenue. Expanding internal drug development could increase both the platform opportunity and R&D spending.

GC Genome combines a testing business with favorable recent flows, but its tests differ from Natera’s

GC Genome has an existing hospital and testing network and sells ai-CANCERCH. The test looks for cancer signals in cell-free DNA in blood. Its purpose differs from Natera Signatera, which tracks minimal residual disease using tumor-informed testing.⁹⁸

First-half revenue was KRW 15.72 billion and operating profit was KRW 570 million. Operating cash flow was positive KRW 3.03 billion, and cash plus short-term financial instruments totaled about KRW 42.05 billion. The existing testing business and profitability are real. However, revenue grew about 7.5%, which is not the same stage as Natera’s rapid growth in oncology testing.¹⁵

The market capitalization of KRW 219.2 billion was about seven times last year’s revenue of KRW 31.53 billion. Profitability alone does not mean the stock is cheap. The next checks are paid ai-CANCERCH volumes and net receipts, repeat orders in Japan, and the clinical, regulatory, and commercial terms for PANC-CERCH in the U.S. An application for FDA Breakthrough Device designation is not an approval or reimbursement decision.¹⁵¹⁶

ST Pharm has the clearest operating results in RNA manufacturing

ST Pharm’s core business is contract development and manufacturing of oligonucleotide APIs for RNA therapies. If AI-discovered drugs advance through clinical trials and reach commercial production, demand for manufacturing may increase. But a larger pool of clinical candidates does not equal confirmed orders today.

First-half revenue was KRW 175.45 billion and operating profit was KRW 29.84 billion. They rose 45.4% and 114%, respectively, year over year. On September 18, the company disclosed an oligonucleotide API supply contract worth KRW 91.33 billion, with delivery due by November 22, 2027. The counterparty is undisclosed, so this cannot be attributed to a Moderna order.¹⁷¹⁸

FnGuide’s October 1 estimates put 2026 forward PER at 29.1x and next-twelve-month PER at 23.1x. These are based on estimated earnings. Valuation should be assessed alongside actual conversion of backlog into revenue and utilization of expanded capacity.¹⁹

First-half operating cash inflow was substantial at KRW 104.54 billion. However, the contribution from working-capital movements has not been fully decomposed, so this cash flow should not be assumed to recur every half-year.¹⁷

Seers shows profitability and cash generation; Lunit shows growth and a narrower loss

Seers’ MobiCare and ThynC businesses cover ECG analysis and patient monitoring. They are useful as a comparison for whether hospital AI can translate into spending and cash, rather than as direct cancer-genomics peers.

First-half revenue was KRW 60.9 billion, operating profit was KRW 26.8 billion, and operating cash inflow was KRW 15.65 billion. The operating margin was about 44%. However, accounts receivable rose from KRW 22.01 billion at year-end to KRW 32.02 billion at midyear. The gap between installations, revenue recognition, and cash collection needs monitoring.²⁰

Second-quarter revenue of KRW 28.38 billion was below the estimated first-quarter figure of KRW 32.53 billion. High year-over-year growth alone does not show that quarterly growth accelerated. FnGuide’s 2026 forward PER estimate is 18.5x, and the next-twelve-month estimate is 15.5x. The multiple is below ST Pharm’s, but the businesses do not have the same risk profile.²⁰²¹

Lunit should be assessed across INSIGHT imaging AI, SCOPE pathology and biomarkers, and the consolidated business that includes Volpara. First-half consolidated revenue was KRW 45.77 billion, up 23.4%, while the operating loss narrowed to KRW 28.96 billion. Total revenue should not be treated as SCOPE data-licensing revenue.²²

Lunit’s KRW 154.5 billion in cash at midyear came after capital raising. It was not cash generated by profitable operations. Operating cash outflow was KRW 21.83 billion. Dividing the KRW 826.5 billion market cap by a simple annualization of first-half revenue gives about 9x. This is not a fair-value estimate; it illustrates the valuation burden if loss reduction stalls.²²²³

The table below compares the scale and profitability of the core candidates. Amounts are in KRW 100 million units (one Korean eok equals KRW 100 million). ST Pharm, Seers, Lunit, and Curiox are consolidated figures; Proteina and GC Genome are standalone figures.

Curiox makes cell-analysis pretreatment automation equipment. It is adjacent to demand for AI experiments, but its product differs from Twist’s DNA-synthesis platform. Its market cap is about 198 times last year’s revenue of KRW 5.15 billion. Paid installations, OEM business, and recurring consumables need to expand materially to validate current expectations.²⁴²⁵

Price and flows show the chase risk in Proteina and GC Genome

Technical indicators use the completed October 1 close. The 20- and 60-day lines are simple moving averages. Investor net purchases are Kiwoom REST provider data for September 23 and the latest five sessions from September 28 to October 1. The table uses the provider’s investor-specific net share volumes, not accumulation inferred from trading volume.

KRX regular-session closes take priority. Proteina’s ordinary daily bar showed KRW 51,600, while the regular-session close was KRW 51,300; the latter is used. Core-candidate closes were checked against FnGuide. Returns may differ slightly from calendar-based one- and three-month returns because the starting dates differ.²⁶

Institutional buying and a medium-term rise are visible in Proteina, while foreign flows are in the opposite direction. Its premium to the 20-day average is 40.7%, higher than GC Genome’s. Without improved results or easing valuation pressure, the price trend alone does not show that the stock is attractive.

GC Genome has the best alignment between recent buying flows and a medium-term rise. At the same time, Wilder RSI14 is about 84 and the stock trades 36.7% above its 20-day average, indicating overbought conditions. Strong business evidence and attractive expected returns for a new buyer are separate questions.

ST Pharm has reclaimed its 20-day average but remains below its 60-day average near KRW 103,560. Seers has rebounded above its 60-day average but remains below its 120-day average near KRW 31,876. Lunit is also rebounding below its 120-day average near KRW 13,055, with sizable recent foreign selling.

The flow data has not been cross-checked against official KRX statistics. For most candidates, only nine of the latest 20 sessions were available, so the comparison uses the completed five-session volumes. Missing data was not filled with zero or combined with another provider’s series. Short selling, stock lending, and passive-fund contributions are unverified, so the movements are not attributed to short covering or ETF accumulation.

Other candidates first need evidence on commercialization, dilution, or business mix

The reasons for lower priority differ even within the same theme.

Vuno’s September 29 rights-issue announcement proposed 6.3 million shares, an October 20 record date, and an indicative issue price of KRW 4,980. The proposed new shares equal about 45% of the current share count. That ratio differs from the ownership dilution for a holder who does not participate. The final issue price and schedule may change.²⁷²⁸

Oncocross held only KRW 285 million in cash and cash equivalents, but also had KRW 21.8 billion in short-term financial instruments. Do not infer a liquidity crisis from the cash line alone. The actual convertibility and restrictions on investment assets need separate checking.²⁸

Pharmicell reported standalone first-half revenue of KRW 75.67 billion and operating profit of KRW 26.31 billion. Nucleosides contributed KRW 3.64 billion, or 4.81% of revenue, and PEG derivatives KRW 6.09 billion, or 8.05%. Other products, which include pharmaceutical, agrochemical, and electronic-material intermediates, accounted for 71.64%. Attributing all of Pharmicell’s strong results to mRNA demand would misidentify the growth drivers.²⁹

OliX is a separate RNA-therapy watchlist name. siRNA suppresses gene expression and differs from an mRNA cancer vaccine. The maximum potential value of Lilly’s license, USD 630 million, should not be treated as confirmed cash or revenue. First-half consolidated operating revenue was about KRW 6.2 billion, operating loss KRW 20.51 billion, and operating cash outflow KRW 21.15 billion. Cash of KRW 111.48 billion includes about KRW 100.2 billion raised through a share issue during the first half.³⁰

Paid revenue, cash, and clinical evidence—not announcements—will change the ranking

After identifying a promising theme, specify what evidence would change the company’s value.

Moderna announced that it plans to present detailed Phase 3 data at ESMO on October 24. This may be a sentiment catalyst for Korean peers, but the presentation itself is not a revenue event for a Korean company. This schedule is based on the company’s September 21 announcement.³¹

The strongest counterargument is that U.S. industry gains can take time to reach Korean-company earnings. Reimbursement, clinical-data scale, pharmaceutical contracts, and access to capital differ by company. A Korean candidate is not more attractive just because it has risen less than its U.S. peers.

Scope and calculation notes

We compared the technical indicators of 17 of the 20 Korean candidates after excluding LabGenomics, Gencurix, and Clinomics because of price-data quality issues. Core-candidate closing prices and market caps were checked against public prices on October 1. Financials use OpenDART first-half year-to-date fields. Quarterly versus cumulative values, consolidated versus standalone statements, and cash versus short-term financial instruments are kept separate.

Korean returns, moving averages, and five-session net purchases were calculated from collected daily prices and Kiwoom REST data. Long-term returns versus the KOSDAQ and biotech indices, full corporate-action adjustments, and exchange-level verification of investor flows were not completed. The data therefore do not establish market outperformance or investor intent. Forward PER estimates and simple revenue multiples are not fair value or new price targets.