The overall pace is slowed down, with poe2 trade a greater emphasis on active dodging, strategic positioning, and understanding complex boss mechanics. Power creep is being reined in to ensure encounters remain challenging and engaging, rather than being trivialized by overwhelming numbers. Every skill used, every resource spent, and every movement is intended to feel more impactful and consequential.

Within this framework, the original Hateforge, with its ability to completely subvert mana costs into a free, constantly regenerating offensive resource, would be fundamentally at odds. It would undermine the very mana economy GGG is striving to establish, allowing players to bypass a core design constraint and potentially trivializing content again.

Given its potential to destabilize the new mana economy, Hateforge might simply not be carried over into PoE2. Its core mechanic might be deemed irreconcilable with the new design goals. This is the more intriguing and likely scenario if GGG wishes to preserve the legacy of such an iconic unique. A re-imagined Hateforge would almost certainly feature:

Instead of 100% conversion, it might grant "X% of Mana Spent as Rage, up to a maximum of Y Rage per second" or "per skill use." This would allow for burst Rage generation but prevent constant, unbounded uptime. Rage generation might be tied to specific conditions, such as "Gain Rage on Mana Spent while at full Mana," "on Mana Spent while channelling a skill," or "when an enemy is killed shortly after mana was spent."

The 100% conversion is almost certainly out. It might be a much lower percentage, making mana management still important. To balance its power, it could come with severe penalties, such as "Reduced Maximum Mana," "Reduced Mana Regeneration," or "Increased Damage Taken while gaining Rage." As a Vaal Greatsword, it might now heavily synergize with buy POE 2 Currency Orbs two-handed attack skills that have inherently high mana costs, making it a niche item for specific attack builds rather than a universal spellcaster enabler.