285H Real Sustained Wattage — Is Liquid Metal Worth It?

The Intel Core Ultra 9 285H realistically sustains about 80W before its cooler-and-firmware combination throttles it — not the 115W Intel’s own spec sheet allows, and not the 45W “base power” figure that undersells what the chip actually runs at. That 80W ceiling comes from the one laptop that currently ships this chip in a ROG chassis, the ASUS ROG Zephyrus G16 (2025), and on that laptop a Thermal Grizzly Conductonaut swap won’t raise it — because ASUS already applies genuine liquid metal at the factory.

Skip to the picks

The 285H isn’t in the Zephyrus G14

Worth clearing up first, because it changes which numbers below apply to you: the 2025 Zephyrus G14 (GA403) kept the AMD Ryzen AI 9 HX 370 it shipped with the year before. The Core Ultra 9 285H only shows up in the 2025 Zephyrus G16 (GU605) and in Strix-series ASUS laptops running Intel’s Arrow Lake-H platform, confirmed on Ultrabookreview’s rundown of the 2025 G14/G16 lineup and on ASUS’s own GU605 product page. If you’re chasing this chip in a 14-inch chassis, it doesn’t exist yet. Everything past this point is about the G16 pairing, and about non-ASUS 285H laptops (MSI, Lenovo, Gigabyte, HP) where the situation is genuinely different — flagged where it matters.

What Intel allows vs. what the laptop sets

Intel’s ARK listing for the 285H states a 45W Processor Base Power and a 115W Maximum Turbo Power. Neither number describes what happens in a real laptop. 45W is a conservative reference floor no gaming laptop actually runs at; 115W is a ceiling no thin chassis can sustain for more than a few seconds.

The 80W/100W figures come from ASUS’s own power-limit configuration on the GU605, as reported in coverage of the 285H trim of the Zephyrus G16 2025 review and in NotebookCheck’s stress-test data for the same chassis. That’s the realistic number: roughly 80W sustained, with a 100W window for the first 20-30 seconds of a demanding load before the CPU settles.

What 80W sustained actually looks like

In a 30-minute Cinebench R23 loop, NotebookCheck measured the G16’s Core Ultra 9 285H peaking at 96°C, with multi-core throughput dropping by roughly 9% from its initial score as the chip settled into its sustained power envelope. That’s a soft decline, not a cliff — the CPU isn’t dropping clocks dramatically, it’s trimming a modest amount to hold near 80W indefinitely. Under extended gaming load (Cyberpunk 2077), the same chip ran cooler, in the 78-82°C range, since games rarely pin all cores at once the way a synthetic loop does.

Intel raised the throttle ceiling for Arrow Lake and Panther Lake parts to as high as 105°C junction temperature, a change covered by WCCFtech when it was announced. A 96°C peak under a 30-minute synthetic loop means the G16’s cooler is already running close to that wall, not idling well below it. That’s the detail that matters for the liquid-metal question: this chip isn’t leaving obvious thermal headroom on the table.

Does Conductonaut raise the ceiling on this laptop?

No — and the reason is specific, not a general “liquid metal doesn’t help” claim. ASUS’s own spec page for the GU605 states its factory liquid metal reduces CPU temperatures by up to 13°C versus standard thermal paste, applied through automated dispensing equipment and a patented containment fence designed specifically to keep the compound off the aluminum sections of the chassis it would otherwise corrode. On the Strix G16 2025, ASUS names the compound directly as Thermal Grizzly Conductonaut Extreme.

Swapping in standard Conductonaut yourself means replacing a factory liquid-metal application with a hand-applied one — same base chemistry, worse containment, and a firmware power limit that has nothing to do with the interface material. The 80W PL1 figure is a number ASUS’s engineers chose in the power table, not a wall the stock paste is failing to clear. There’s no wattage to unlock by reapplying what’s already there.

If your 285H laptop didn’t ship with liquid metal

This is the real fork in the decision, and it’s a condition you can check yourself: open your OEM’s spec sheet or run HWInfo and look at the thermal interface note, or just check whether your laptop is an ASUS ROG model at all. Several other vendors — MSI, Lenovo, Gigabyte, HP — build 285H configurations that still ship with standard silicone-based paste. On those, a repaste can genuinely drop temperatures.

But the same firmware logic still applies: a cooler CPU doesn’t automatically mean a higher sustained wattage. Repasting only raises the effective power ceiling if the chip was thermally throttling below its firmware-set PL1 before you touched it. If it wasn’t — if it was already sitting comfortably under its power limit and temperature cap — a repaste buys you lower temperatures and quieter fans at the same wattage, not more performance. Raising the wattage itself requires a separate step: your OEM’s control-center Manual/Turbo mode, or Intel XTU where the vendor hasn’t locked it, to actually move the PL1 table.

Conductonaut, Conductonaut Extreme, or PTM7950

Thermal Grizzly’s own datasheet for standard Conductonaut lists a 73 W/mK thermal conductivity and an explicit warning: it must not touch aluminum, which will pit and corrode on contact with the gallium in the alloy. The Conductonaut Extreme datasheet carries the same warning and recommends application on nickel-plated copper specifically for long-term stability — which is exactly what ASUS’s factory fence design is built to guarantee and a hand application on a cramped laptop board is not.

Skip standard Conductonaut on any ASUS ROG 285H laptop — it’s already there, applied better than a DIY job can replicate. Consider Conductonaut Extreme only if you own a non-ASUS 285H laptop, are confident in liquid-metal application on a laptop-scale board, and have already confirmed you can raise the firmware power limit afterward — otherwise you’re taking on aluminum-corrosion and short-circuit risk for a temperature drop with nowhere to spend it. For most stock-paste 285H laptops, PTM7950 is the better repaste: as cooling-hub comparisons of the material note, its phase-change design resists the pump-out that degrades both stock paste and liquid metal in a laptop cooler over several years, without the spill risk that makes liquid metal a one-mistake repair.

What to check before you do anything

Confirm the exact OEM and SKU, not just “285H” — the chip is identical across vendors, but the power table and the stock thermal interface are not. Log PL1/PL2 and package temperature with HWInfo during a 20-30 minute load to see whether your unit is temperature-throttling below its own firmware limit or holding steady at it; that single log tells you whether a repaste or a power-limit change is the actual fix. And check your warranty terms before opening the chassis — ASUS’s liquid-metal application on the 285H trims is a factory process built around a containment fence that a home repaste cannot fully replicate, which is the concrete reason this specific swap doesn’t pay off on that laptop.