UMC · United Microelectronic Corp. · Stock research

Current Thesis

UMC spent the first half of 2026 being repriced from commodity mature-node foundry into an AI-adjacent silicon-photonics supplier, and the tape ran further than the story: +222.6% YTD, a 52-week range of $6.56 to $28.96, and a market cap near $55.9B against $7.53B TTM revenue. That leg has now cracked. The ADR closed $24.92 on 2026-07-15 after a 4.5% gain, then gave back 4.67% on 2026-07-17 to $21.25 roughly 27% below the July high inside a handful of sessions, and back beneath the $21.63 level it traded at on 2026-06-12 before the final vertical push. Underneath, the operating trend is genuinely improving: June revenue NT$23.12B (+22.85% YoY, reported 2026-07-06) marked a third consecutive month of accelerating growth after April's +10.8% and May's +17.78%, with H1 sales NT$129.77B (+11.28%). What has changed is who owns the narrative. Macquarie reinstated coverage at Outperform on 2026-07-14 and BNP Paribas moved off Underperform to Neutral in the same window the sell-side catch-up that the earlier read flagged as the saturation marker. Buying a fourth-quartile foundry at 35.7x trailing earnings, ten days ahead of a Q2 print, after the Street has started publishing the thesis, is a materially worse setup than it was in May.

Bullish and bearish views on United Microelectronic Corp.

The model's bull view on United Microelectronic Corp. (UMC), in brief: June 2026 revenue NT$23.12B, +22.85% YoY (reported 2026-07-06), the third straight monthly acceleration (Apr +10.8% → May +17.78% → Jun +22.85%); H1 sales NT$129.77B, +11.28% YoY. The bear view: The July blowoff has already been sold. From $24.92 (2026-07-15) to $21.25 (2026-07-17) is -14.7% in two sessions, and -26.6% from the $28.96 high. A parabola that gives back a quarter of its range while volume runs 20.8M shares is distribution, and the June breakout shelf is… Both cases follow in full.

Bull Case

• June 2026 revenue NT$23.12B, +22.85% YoY (reported 2026-07-06), the third straight monthly acceleration (Apr +10.8% → May +17.78% → Jun +22.85%); H1 sales NT$129.77B, +11.28% YoY. The recurring monthly catalyst is still trending in the right direction.
• SILITH 1.6T silicon-photonics platform reached mass production out of UMC's 12-inch Singapore fab (announced 2026-07-14), taken from development to volume in 18 months and qualified by a named-as-leading cloud infrastructure customer. This is revenue-bearing AI datacenter exposure rather than a roadmap slide.
• Roadmap extended from 200G/lane to 400G/lane optical interconnects (2026-07-14) following the 200G commercialization the follow-on socket is already scoped, and UMC's own 12-inch photonics platform opens to customer development in 2027.
• Q1 2026 EPS NT$1.29 versus NT$0.85 consensus on revenue NT$61.04B (+5.5% YoY, reported ~2026-05-06), with record 22nm revenue. EPS more than doubling on mid-single-digit revenue growth is mix and utilization leverage.
• H2-2026 wafer price increases flagged on the Q1 call to fund $1.5B capex returning pricing power reverses the 2024–25 mature-node price war that capped the multiple.
• Macquarie reinstated at Outperform 2026-07-14; BNP Paribas upgraded to Neutral from Underperform the first real institutional re-rate of the coverage base, with consensus targets ($8.60 to $16.24 depending on the panel) still miles below spot and therefore mechanically due for revision higher.

Bear Case

• The July blowoff has already been sold. From $24.92 (2026-07-15) to $21.25 (2026-07-17) is -14.7% in two sessions, and -26.6% from the $28.96 high. A parabola that gives back a quarter of its range while volume runs 20.8M shares is distribution, and the June breakout shelf is the next thing at risk.
• Q2 earnings land 2026-07-29 (investor conference 17:00 Taipei) with consensus at $0.15 EPS on $2.06B revenue versus $0.12 on $2.01B a year ago. That is a +4% top-line estimate against a stock that has tripled the print has to deliver gross-margin expansion, not just revenue, and the bar is now set by the tape rather than by the model.
• Valuation left the foundry comp set behind. 35.7x trailing earnings on $0.12 EPS, consensus rating "Sell" from the 4-analyst panel with a $16.24 average target implying ~24% downside. Beta 1.57 means this unwinds faster than the group on any semis risk-off day.
• Zero-coupon convertible raise approved 2026-06-03 (~NT$16B, dual tranche) funds the capacity build but leaves a dilution overhang, and convertible-arb desks short the underlying to hedge, which adds mechanical supply into exactly this kind of drawdown.
• The revenue base is still commodity mature node. Display driver, RF and power silicon at 28/22nm, with SMIC and Hua Hong adding capacity in China. Silicon photonics is real but small; if the H2 wafer price hikes do not hold, margin compression resumes regardless of the AI headline.
• Dividend support is gone the $0.412 payout with a 2026-07-08 ex-date now yields ~1.5% after the ADR tripled, versus 5–6% a year ago. There is no income floor beneath the growth re-rate.

Setup & Price Structure

Price $21.25 (2026-07-17) against a $28.96 52-week high and $6.56 low. The structure that mattered a steady advance from the June $21–23 zone into a July vertical has been broken by the two-day reversal, and the stock now sits below where it traded on 2026-06-12. That puts it in the awkward middle: too far below the high to call the trend intact, too far above the June base to call it supported. The June breakout shelf near $19 is the level that decides which it is; a weekly close beneath it turns the entire July leg into a failed extension. Above, reclaiming $24.92 on volume would re-establish the advance and make the drawdown a shakeout ahead of the print. Entering between those two poles, ten days before earnings, with a 1.57 beta, is paying for both outcomes at once.

Catalyst Calendar (next 30 days)

• 2026-07-29 Q2 2026 results and investor conference (17:00 Taipei). Consensus $0.15 EPS on $2.06B revenue. Gross margin, Q3 utilization guidance and confirmation of the H2 wafer price increases are the three numbers that matter.
• ~2026-08-07 (est.) July monthly Taiwan revenue print. The recurring catalyst; the question is whether the Apr→Jun acceleration extends to a fourth month or rolls over off a hard June comp.

Elapsed catalysts

• Ongoing through August silicon-photonics customer news flow following the 2026-07-14 SILITH mass-production and 400G/lane roadmap announcements; incremental cloud-customer qualifications would be the first evidence the photonics narrative carries revenue weight in 2027 guidance. _(passed 5d ago)_
• Elapsed 2026-07-08 $0.412 dividend ex-date. Already passed; no longer a support factor. _(passed 11d ago)_

What Would Change Our Mind

The bear read gets retired if the 2026-07-29 print delivers gross-margin expansion alongside the +22.85% June sales growth and management quantifies silicon-photonics revenue as a 2027 line item rather than a platform milestone that converts a multiple-expansion story into an earnings story and justifies holding through the drawdown. A weekly close back above $24.92 with the July high in reach would say the two-day flush was a positioning event rather than a top. Conversely, a weekly close below $19.00 takes out the June breakout shelf and confirms the $28.96 print as the high of the cycle; from there the $16.24 consensus target stops being a joke and becomes a magnet. A Q2 print where revenue meets but gross margin compresses the mature-node price war reasserting itself under an AI headline is the specific fundamental break to watch for.

Correlation Notes

UMC trades as high-beta foundry-cycle exposure with an AI-photonics option layered on top, not as the income ADR it was through 2025. Correlated cluster: TSM (the sector's directional anchor), and the mature-node peers SMIC and Hua Hong, whose China capacity additions are the direct competitive threat. On the photonics side the read-through runs to COHR, LITE and the optical-interconnect complex a bid there supports UMC's newest narrative leg, while weakness undercuts it. Broader semis beta comes via SOXX/SMH; on the memory and storage side MU and STX move the sentiment tape but not UMC's revenue. Given beta 1.57 and a stock 220% higher YTD, any general semiconductor de-rate hits UMC harder than the index, and the July reversal already showed that asymmetry working.