Path of Exile 2

There is a particular kind of studio arrogance worth respecting: the kind that looks at a genre defined by dopamine drip-feeds and says, no, you’re going to earn this. Grinding Gear Games has been running that playbook since the original Path of Exile ate Diablo III’s lunch by being denser, darker, and free. Path of Exile 2 is that thesis escalated to an almost confrontational degree.

It is not a remaster and not really a sequel in the narrative sense. Set generations after the first game on the same ruined continent of Wraeclast, it’s a ground-up rebuild of the engine, the combat model, and the design philosophy — while inheriting the systemic maximalism that made the original a cult object. A long multi-act campaign, an enormous variety of environments, monster types and bosses, with nearly every zone terminating in a real encounter rather than a loot piñata. A full class roster, each with multiple Ascendancy specializations, producing a wide range of distinct endgame identities before you’ve touched the passive tree.

Market position is the interesting part. PoE 2 doesn’t compete with Diablo IV on comfort; it competes on ceiling. Blizzard sells a seasonal ritual you can complete in a weekend. GGG sells a second job. The Steam review split reflects exactly that: a solid-but-divided aggregate with recent sentiment trending warmer as successive Early Access patches sand down the worst edges. That’s not a mediocre game. That’s a polarizing one, and the aggregate does a poor job of hiding the fight underneath it.

The business model deserves its own line. PoE 2 is a paid release ($29.99 on Steam), and its monetization sticks to cosmetics and stash tabs. In a genre currently drowning in battle passes and “accelerators,” that restraint is close to a competitive moat.

• The Shared Passive Tree with Dual Specialization: A single sprawling constellation of hundreds of nodes that every class enters from a different point, now supporting two simultaneous specialization paths. It is the most legible expression of “your build is yours” in any ARPG, and the most intimidating first-hour screen in gaming.
• Skill and Support Gems That Socket Into Skills, Not Gear: A large library of skill and support gems, decoupled from your equipment. Your damage identity lives in the gems; your stats live in the gear. Changing your entire playstyle no longer requires re-farming an armor set.
• Deliberate, Combo-Driven Combat: Dodge-rolling, positioning, animation commitment, and cross-skill combos replace the genre’s usual hold-left-click-and-vibrate. Attacks have weight and recovery frames. So do the monsters’.
• The Atlas Endgame: A procedurally seeded map of biomes driven by Waystones, with its own separate passive trees layered on top of your character build. This is the actual game; the campaign is the tutorial.
• Deep Crafting and a Player Economy: A huge catalog of equipment base types with unique and corrupted variants, feeding a currency-based trade economy that has no gold — items are the money.
• Co-op, Hardcore, and Solo Self-Found Ladders: Play it social, play it permadeath, or play it in total economic isolation. Each is a legitimately different game.

The Gameplay Loop

The first thing PoE 2 does is slow you down, and the second thing it does is punish you for not noticing.

Combat here operates on a fundamentally different clock than its peers. Skills have real wind-up. Monsters telegraph, and the telegraphs mean something. The dodge-roll is not a movement convenience — it’s your primary defensive tool, with an invulnerability window tight enough that mistiming it reads as a mistake rather than bad luck. Bosses, in particular, are built like they wandered in from a different genre entirely: multi-phase, pattern-driven, demanding you learn a moveset rather than out-stat it.

When this works, it’s genuinely excellent. The best encounters in PoE 2 rank with the best boss fights in the genre. There’s a tactile satisfaction to freezing a pack solid and detonating it, or stunning a rare into a heavy-attack window, that the genre’s usual screen-clearing power fantasy simply cannot produce.

When it doesn’t work, it’s because the game has quietly decided your build is invalid and declined to tell you. Difficulty spikes are the single most consistent complaint in the review data, and they’re legitimate. The curve isn’t a curve; it’s a staircase with occasional missing steps. You can be comfortably clearing content and then meet a boss with an unavoidable one-shot mechanic that your defensive layers were never designed to absorb, with no in-game diagnostic explaining what went wrong. The game’s answer is “figure it out.” For a certain player, that’s the appeal. For most, it’s a brick wall with no handholds.

The Build System

The passive tree is the reason people stay for years. It’s also the reason the onboarding funnel leaks.

Decoupling skills from gear is the smartest structural decision GGG made. In the original game, your six-link chest piece was your build, and losing it was catastrophic. Here, gems socket directly into skills, which means your character concept is portable and your gear hunt is about stats and defenses rather than about a single lottery ticket. Dual specialization on the tree pushes this further, letting you commit to two identities — a damage archetype and a defensive or utility one — without the tree feeling like a series of taxes.

The cost is information density that borders on hostile. There is no meaningful in-game guidance for tree planning. The realistic path for a new player involves a third-party planner, a build guide, and a browser tab open on a second monitor. That’s not a flaw in the depth; it’s a flaw in the interface layer that should surface it. GGG has built a system with genuine expressive range and then outsourced its documentation to the community.

The Endgame, and the Death Penalty Problem

The Atlas is where PoE 2 becomes a hobby. Waystones seed maps, maps feed the economy, the economy funds your next build, and separate endgame passive trees let you shape what the world itself generates. The structural design is excellent — a genuine meta-progression layer rather than a difficulty slider.

Then you die once, and the map is gone.

Losing a Waystone map on a single death is the most contested design decision in the game, and the criticism is fair. It isn’t that the penalty is harsh; it’s that the penalty is harshest precisely where the game is least readable. Endgame maps stack modifiers that can produce unfair damage spikes, and pairing an unpredictable threat model with a total-loss penalty converts experimentation into risk management. Players don’t respond to that by getting better. They respond by netdecking the builds that can’t die and refusing to touch anything else — which is the opposite of what a game with this much build space should want.

Friction Elsewhere

Campaign backtracking is slow, and travel between objectives eats real time in a game already asking for hundreds of hours. Currency and loot scarcity hits some builds disproportionately, creating a self-reinforcing gap where the builds that need gear most can least afford it. Balance patches have swung hard — nerfs land between leagues with a severity that has repeatedly invalidated builds mid-season. And league launches have historically been a reliable source of server strain; this is a long-standing pattern at GGG, not a new one.

The art direction is the least controversial thing in the package. This is one of the best-looking ARPGs on the market — an oppressive dark-fantasy palette of wet stone, decaying flesh, and guttering light, rendered with a physicality that makes Diablo IV’s grimdark look like a mood board. Rain pools. Armor is filthy. Corpses persist and get in the way. Environmental storytelling does real work across the campaign’s zones, and the monster design in particular shows an animator’s understanding that a threat you can read is scarier than a threat that’s merely loud.

The UI is a split verdict. In-combat readability is strong by design — telegraphs land, status effects are legible, and the game is built so you rarely die because you couldn’t see something. Out of combat, the inventory, crafting, and tree interfaces are dense, sparse on tooltips, and clearly designed by people who already know the systems. Form has not lost a war with function so much as function has been assumed rather than taught.

Performance is the real asterisk. A 2D-to-3D ARPG of this scope and effect density is demanding hardware-wise: the game leans on heavy shader compilation and stacked on-screen effects, which is exactly the profile that produces stutter on first encounters and frame pressure on mid-range GPUs, and league launches add server-side latency on top. None of this is fatal; all of it is chronic to the design.

On Steam, PoE 2 is a mouse-and-keyboard game that happens to have strong controller support — and that distinction matters more than usual here. Mouse-and-keyboard gives you precise skill targeting and instant access to a wide hotbar, which is what the endgame’s build complexity actually demands. Controller support is well considered for the genre, with a radial menu and a dodge-roll mapped for a stick, but it constrains your usable skill count and makes ground-targeted abilities meaningfully harder to place. Choose based on your build, not your couch.

Steam Deck is playable rather than pleasant on paper. The dense text of the passive tree and inventory is a poor fit for the Deck’s 800p screen, and sustained endgame mapping is the kind of workload that tests both thermals and battery. Treat it as a campaign-and-casual-mapping device.

Two Steam-specific realities worth planning around: the game is online-only, so there’s no offline play and no meaningful use of Steam Cloud — your character lives on GGG’s servers, with all the launch-week queue implications that carries. And the trade economy runs primarily through an out-of-client web interface. You will alt-tab constantly. For a game this polished elsewhere, the trade experience is stubbornly, deliberately archaic — GGG considers the friction a feature, and reasonable people disagree.

The Steam review score is also being read wrong by a lot of people. That aggregate isn’t a quality signal; it’s a compatibility signal. The complaints inside the negative reviews are remarkably consistent, and they’re the ones listed above.

Path of Exile 2 is a better game than its Steam score and a harder recommendation than its best moments deserve. Grinding Gear Games has built something with a higher ceiling than most of the genre — the combat has real weight, the boss design is excellent, the build system is close to inexhaustible, and the monetization is restrained in a way that shames its competitors.

What’s holding it back isn’t ambition or execution. It’s a set of specific, fixable design choices — the map-loss death penalty chief among them — that convert the game’s depth into anxiety and push players toward the narrow band of builds that can survive their own mistakes. The trajectory is encouraging: recent sentiment is up, patches have landed, and GGG has a long track record of iterating past exactly this kind of problem. But the game currently in front of you asks for a tolerance that most people don’t have and shouldn’t be expected to develop.

Buy it if the pitch above sounds like your kind of game, ideally with a build guide open in a second tab. If it clicks, it will hold you for years. If it doesn’t, you’ll know within a weekend, and you’ll know why.