Analysts Say YTL’s Seven Turbines Could Back 5.25 GW for AI Power
Broker research published around September 18, 2026 argued that YTL Power International’s seven reserved Siemens Energy SGT-9000HL gas turbines—three earlier units plus four newly reserved with Ganda Power—could together support about 5.25 gigawatts of combined-cycle capacity for Malaysia and the wider region, The Star reported. CGS International estimated potential annual profit in the seven-hundred-fifty to eight-hundred-fifty-million-ringgit range if projects land, and sketched a multi-billion-ringgit valuation uplift, while noting commissioning might land in late 2030 or early 2031 after roughly four-year builds.
Filed under Infrastructure and dated September 21, 2026, this AI4Malaysia briefing treats the analyst math as Malaysian AI-power news distinct from the turbine-booking announcement alone. Hong Leong Investment Bank said the first three turbines, about 2,250 megawatts, are expected before 2030 and the remaining four, about 3,000 megawatts, afterward, arguing government planners should welcome capacity as data-centre load rises. YTL’s own AI and data-centre ambition remains pegged around 2.4 gigawatts, making generation reservations a gating item for hyperscale campuses in Johor and beyond.
Why it matters: Malaysian AI factories cannot run on press releases—they need firm megawatts. Turbine scarcity can decide who builds on time—but reservations are not power-purchase agreements.
What it means in practice
Malaysian energy and DC developers should map which campuses need CCGT-backed firm power; demand clarity on PPA and grid-connection timelines; assign an owner for environmental and fuel-risk reviews; run time-boxed scenarios on 2030 delivery slips; and prefer offtake structures that do not strand households with cost shifts. Read the analysis beside YTL’s turbine bookings and the Sedenak campus plan.
Caveats come first. Research targets are not awarded plants; IRRs assume favourable contracts; and gas-fired AI power faces transition politics. AI4Malaysia therefore presents the 5.25-gigawatt thesis as directional infrastructure context until PPAs and site approvals appear.
What to watch next: first PPA announcements; delivery schedules for the initial 2,250 megawatts; and how Firmus–OpenAI compute bookings pull power demand forward. Readers can continue on the AI4Malaysia homepage, or browse the Newsroom for additional briefings.
Bottom line: treat this update as orientation, not instruction. Malaysian AI growth is increasingly a power-generation story and remains contingent. Organizations that benefit most will secure real megawatts, keep humans on grid and community risk, and refuse to confuse turbine reservations with finished AI campuses.