TRT

Microcap semi back-end burn-in test services pivoting into AI GPU + auto-semiconductor reliability testing. Q3 FY2026 (2026-05-14) confirmed the pivot: rev $16.51M +124% YoY, Back-End Solutions $13.1M +141%, 9-month $47.67M / $165K net income vs prior-year loss. Capacity response: 104k sqft Malaysia lease earmarked for AI testing. $10M Apr raise (~12% dilution) priced into the parabola; customers undisclosed.

• •2026-05-14 — Q3 FY2026 print: revenue $16.51M (+124% YoY); Semiconductor Back-End Solutions segment $13.1M (+141% YoY); GAAP net loss $38K (near-breakeven vs $495K loss prior year). 9-month FY26: rev $47.67M, net income $165K — turnaround from prior-year loss
• •2026-05-14 — Capacity expansion concrete: signed lease on an additional 104,000 sqft in Malaysia, earmarked for AI-related testing services. Funded by the Apr 27 $10M raise
• •2026-05-15 — Stock reaction: ranged $13.23 → $20.71, closed ~$20.27 on 3.74M shares (~4× average volume of 925K). Mcap re-rated to ~$184M
• •2026-02-13 — Q2 FY2026 reported +82% YoY revenue, attributed to AI + EV semi reliability test demand
• •2026-03-04 — $2.5M production burn-in order from a leading automotive IDM (customer not disclosed)
• •2026-03-17 — $5.3M burn-in board order for a "next-generation AI GPU platform" + $2.5M follow-on (per Q3 release) = $7.8M cumulative AI-GPU burn-in board orders
• •2026-04-24/27 — Registered direct offering priced + closed: $10M, 1,052,632 shares at ~$9.50, ~12% dilution on the ~9M-share base. Proceeds: AI + auto semi capacity expansion (Malaysia lease is the first concrete spend)
• •2026-04-15/17/21 — Insider Form 4 filings; direction not yet pulled — material to the read on whether management treated the spike as distribution
• •Snapshot 2026-05-15: spot ~$20.27, mcap ~$184M, Q3-annualized revenue ~$66M (P/S ~2.8× on run-rate) or TTM ~$53M (P/S ~3.5×). Net cash ~$26.5M post-raise vs ~$3M debt. Gross margin 19.7% (slim), GAAP near-breakeven. Compared to 2026-05-05 snapshot (price $12.54, mcap $126M): fundamentals caught up to the multiple rather than the multiple compressing