Epic’s New Launcher Admits It Was AI Built: 30x Faster Library Loads, Product Proof and Player Trust#

Hello everyone, Zhanlan Yingren here. The most talked-about story in games this week was not a AAA blockbuster but a launcher. Epic has released a preview of a brand-new Epic Games Launcher, and the first sentence of the official blog is already a candidate for the year’s highlight reel. Verbatim:

“Programming is hard. So, instead of doing that, we vibe coded a new Epic Games Launcher that’s way faster and more responsive than the current one, and has a new look.”

In plain terms: programming is hard, so we skipped it and vibe coded a launcher that is faster, snappier and re-skinned. A major publisher putting AI-built right into the opening line of an official blog, then immediately tossing out a set of benchmark numbers aimed at Steam — that is a first in the PC platform scene. When the news landed, our newsroom argued about it all day. Some called it the signature moment when AI-written software first got a loud official debut in a major publisher’s consumer product. Others went straight to the uncomfortable question: would you hand your credit card details to a program an AI cobbled together?

As a player who spent twenty years in the mud of the Steam, Origin and Epic exclusives mess, my feelings about the word launcher are complicated: it is the first door you walk through every time you sit down to play, and also the layer of shell you curse the most and can never quit. In this piece, we unpack the whole announcement, the fine print on those performance numbers, the speed-versus-stability fight, where this sits in the launcher wars, and how players are actually reacting — all in one pass.

1. The event replay: one blog post, one tweet, and one huge pivot#

House rule: timeline first, judgment after. One caveat — outlets disagree by one day on two key dates, so we present both without forcing a merge:

By Sweeney’s account, the goal of the huge pivot is to reorganize developers for maximum AI acceleration — in his own words, not doing the same work with fewer people, but letting everyone build more and better products, to the benefit of every Unreal Engine team. When that statement landed, this site had already covered the organizational and labor angle (see our October 14 piece on the AI first reorganization); this time the story moves from the meeting room to the product itself — the new launcher is the first consumer-facing deliverable of that route.

The other thread not to miss is the official marketing posture. Epic’s official tweet ran: We vibe coded the Epic Games Launcher to be faster than the other guy. At the same time, the account self-deprecatingly pre-announced bugs: Is this bad boy faster? Yes. Will it have some bugs? Yes. Do we want to hear about them? Hell yes. — adding that user feedback will shape the final product. A major publisher writing we-admit-there-will-be-bugs into launch messaging is rare in launcher-war history: it is marketing ease, and it doubles as a pre-emptive shot at quality expectations.

The first half of that opening sentence — programming is hard — deserves its own entry. The opening line of a major publisher’s blog is traditionally the most carefully worded real estate: either vision or user value. This time Epic put a near-guileless confession there, then immediately explained the remedy: stop writing code, vibe code one instead. The product’s selling point, the development method, and the brand attitude — three things compressed into one joke. Whether or not this route ultimately works, the move of making the development method itself the selling point in an opening line will be cited repeatedly by anyone studying how big publishers convert to AI: it is the first time AI-written software has appeared this loudly in a top publisher’s player-facing official narrative.

2. How to read the performance numbers: two comparison frames and one undisclosed test environment#

The numbers are the hook, so let us lay the official claims out in a table:

Three things to grab from this table.

First, the official line actually runs on two comparison frames: against Steam, it is cold boot into the store and tray resume; against Epic’s own old launcher, it is library load and overall startup. The scariest number in the headlines, 30 times faster, is measured against Epic’s own old version — and anyone who used the old library knows what that baseline was worth. The genuinely Steam-facing pair is 5.47 seconds against 10.07 seconds on cold boot.

Second, every one of these numbers is an official figure: Epic commissioned an external vendor to run the tests, but the system environment used was not disclosed — a point at least one outlet flagged explicitly. In other words, the comparison table you are looking at is a report card with no test report attached. Reference it, yes; treat it as settled, not quite.

Third, the tray-resume pair deserves a mention: 0.36 seconds against 0.42 seconds, an absolute gap of 0.06 seconds. One outlet joked the advantage was beside the point — human perception cannot distinguish 0.06 seconds, so putting it in a headline battle report against Steam is marketing more than meaning.

One vertical comparison is also worth recording: when Epic teased the new launcher earlier this year, the promise was a 5x boot speedup; the installed official figures now are library loads 30x faster and startup 2x faster. Teased figures and shipped figures were never the same currency, but landing far beyond the original promise did add volume to the announcement. Whether that is marketing leading engineering or engineering genuinely breaking through awaits independent third-party retesting — none is public yet.

Straighten that thread out on its own:

3. What the AI actually built: a from-scratch new stack#

Store VP and GM Martin Keely confirmed in interviews that this was not a patch job on the old launcher but a rebuild from zero: a brand-new tech stack, with the app shell — download manager, library, and the app itself — plus the new store architecture rewritten entirely, and the discovery page and product detail pages refreshed as well. On AI’s role, his wording was measured: AI is one of the tools for writing and reviewing code in the new stack; which models were used, Epic has not disclosed. He also stressed that design and art remain human-led.

The preview’s feature list shipped alongside it, arranged here:

• Store, library and social panels
• Unreal Engine developer page
• Recently played list
• A centralized download hub with live download speeds
• Activity feed showing playtime and install dates
• Detail view and grid view toggle
• Visual search
• Collections
• Voice chat
• Customizable top bar

For the official account of what was cut, the blog has a vivid analogy: this first version is a speedboat, not a yacht — fewer features, light and quick, but the shopping cart and other core commerce functions are in place. As product strategy, the trade is smart: launcher legacy bloat mostly comes from feature pile-up, and shipping light first and iterating on feedback lines up neatly with the official line that user feedback will shape the final product.

Platform support is the other heavyweight item: the new launcher covers Windows and macOS and supports Linux for the first time — though multiple reports say Linux is not actually usable during the preview. We unpack the Linux thread and what it means for players in section seven.

4. The speed-versus-stability tension: payment flows do not forgive move-fast#

Vibe coding has been a buzzword for a while, and software engineering’s critique of it was rehearsed at length this week by the press: better-hidden bugs, security holes, and poor maintainability — code generated without understanding is code that cannot be extended without understanding. Those critiques usually sound like faculty-lounge chatter, but dropped into the specific case of a launcher, each one lands somewhere vital.

The sharpest came from Polygon: handing credit card information to a vibe coded program is unsettling. The sentence has weight because a launcher is not a game. A game crashing? Quit and wait for the patch. A launcher is the layer that runs resident in your system, touches payments, and manages your entire digital library — the waypoint between the player and the wallet. The same bug is an experience problem in a game and a funds-security problem in a payment flow. That is why the official 46% faster battle report and the official admission of bugs can coexist while the player trust ledger has to be kept separately: speed is a bonus, payment safety is the pass line, and there is no accelerating past a pass line.

In fairness, Epic did not dodge. Pre-announcing bugs, soliciting feedback, and labeling the build a preview all say the company knows the distance between preview and final. The real issue is that vibe coding’s structural quality risks — above all, lagging security audits — do not dissolve because the posture is candid. The preview stage is exactly the window when those risks surface; whether players ride along, watch, or wait for final is each person’s own risk call, and section eight offers concrete advice.

5. Twenty years of launcher wars: where this announcement sits#

Only a longer timeline shows where this event stands. The last twenty years for PC players trace roughly this arc:

• Steam: two decades of sedimentation into the industry default, dominant throughout
• Publisher launchers like Origin and Uplay: arriving in sequence, then falling or fading in sequence
• Epic: chasing on exclusives and free games, carving out its own market against Steam’s gravity

A launcher has never been just a tool; it is the front line of the platform wars — and this round’s gunpowder is hidden in the development method.

Epic’s own read of the war is clear-eyed. The store chief has said the goal on PC is not to topple Steam but to win enough market share that publishers have no choice but to ship on the Epic Games Store. That framing is so pragmatic it borders on conservative, and it explains why the new launcher needed Steam as its reference object — not to declare victory, but to prove it deserves space on players’ drives.

Two more background details belong in the coordinates. One: months ago Epic brought in Blizzard Battle.net’s head to help improve the store — Battle.net is one of the oldest launchers in the business, and hiring its leader shows the store line has been mapped for a while; the new launcher is no impulse. Two: Keely also told media Epic will improve game discovery and social recommendations. Connect the threads: the new launcher is a product-side proof — a from-scratch stack and a set of benchmark numbers proving Epic’s engineering to players and industry, paving the way for discovery and social features to come.

And its true historical coordinate is this: the first time a major publisher has written vibe coding into the official narrative of a consumer product, with benchmark numbers against a competitor attached. Twenty years of launcher wars fought over exclusives, revenue splits and free games; now there is a new front — whose development method is more AI.

6. The player community’s trust ledger: official self-deprecation and historical debt#

On player reaction, a scope note first: our search surfaced no quantitative community reviews or sentiment clustering, so what follows rests on qualitative accounts relayed by media. Readers should know that.

Even so, the outline of the trust problem is clear. Supporters see hard numbers and the good faith of first-time Linux support; skeptics’ ammunition comes from Epic’s own history. Line up the old debts the media has recapped and you have a ready-made trust liability sheet:

• Layoffs of more than 1,000 people earlier this year, officially attributed to declining Fortnite engagement
• An earlier admission that Fortnite skins were made with AI, which drew player criticism
• A controversial AI-generated Darth Vader NPC that players goaded into swearing
• An Unreal Engine 6 teaser showcasing a large batch of generative AI tools

Put that liability sheet next to an AI-built launcher and player wariness explains itself: a company repeatedly singed by AI on the content side is now handing its core distribution tool to an AI workflow too, stacked on top of payment sensitivity. Alarm was close to inevitable.

Epic saw it coming. The will-it-have-bugs-yes self-deprecation and the hell-yes on feedback amount to pulling players into the test process — using the community’s collective trial and error to hedge vibe coding’s quality risks. That strategy may suffice for the preview stage, but trust is ultimately priced by the final release: if bugs get fixed fast and security audits hold, AI-built turns from marketing copy into engineering proof; if the final version breaks payments or security, this loud debut becomes ground zero of the backlash.

Set the two camps side by side and the essence of the dispute is not whether AI-written code works but what players are willing to hand to AI-written code. A launcher is not ordinary software: it holds the library players paid real money for, holds the payment channel, and runs resident in the system. The same error tolerance players might grant an indie game does not transfer to software holding credit card data. The more candidly Epic admits there will be bugs, the less discount players can be asked to take on payment safety — the two coexist fine.

7. What Linux players got: anti-cheat is the hard bone#

For Linux players, the bigger news hiding in this release is not the launcher: Epic confirmed Fortnite will come to Linux alongside the new launcher, on one explicit condition — Epic must first rebuild its own Easy Anti-Cheat (EAC) for the platform. And the blog’s wording is that Fortnite arrives once this reaches critical mass with enough distro maintainers.

Lay the current state and the conditions into a table for Linux players to find their row:

Why is anti-cheat the precondition for Fortnite on Linux? The official coupling of the two gives the logic away: competitive games’ anti-cheat digs deep into the system floor, and a platform switch means rewriting core components — heavier engineering than the launcher’s stack swap. The official path is pragmatic too: reach critical mass with enough distro maintainers first, bring the game after. Linux players are not waiting for a launcher; they are waiting for a whole layer of low-level platform cooperation to land. For the new launcher, Linux support is good faith shown for the first time; for Fortnite, this is the first time the long-running flagship truly opens its door to Linux players — provided that hard bone really gets cracked.

8. Practical advice: should you install this preview now?#

Finally, the question players actually care about: install the preview or not? Here is the list by situation:

Three general notes: first, a preview is a preview — the official copy itself says there will be bugs, so do not migrate your whole game library in a hurry; second, Epic has given no timetable for the final version, so there is no solid just-two-more-weeks-and-it-is-stable story; third, the feedback channel is an officially welcomed entrance, and filing a report there beats raging in forums — they did ask for it. One note for collectors: the new launcher’s activity feed records playtime and install dates, a small bonus for anyone who likes tracing their own history; dual detail/grid views and collections are all live in the preview too — ten minutes of hands-on will tell you whether it is worth switching.

Zooming out: from the industry’s observation deck, this release and Epic’s earlier organizational reorg complete a two-piece puzzle. The first piece was how the organization changes — developers reorganized for maximum AI acceleration. This one is what the product looks like — a from-scratch stack, 30x library loads, and a candid admission of bugs. Fit the two together and programming is hard, so we did it differently stops being a joke and becomes a published product route. Whether that route holds up depends on the answers security and stability deliver in the coming months; and for the industry at large, a major publisher’s loud consumer-product demonstration has landed — to follow or not, and how, is the multiple-choice question every platform company will face in the next year or two.

Frequently asked questions#

Q: Is the new launcher 30 times faster than Steam?

A: No. In the official framing, 30 times faster refers to library load speed against Epic’s own old launcher; the Steam-facing numbers are cold boot into the store, 5.47 seconds against Steam’s 10.07 seconds (46% faster), and tray resume, 0.36 seconds against 0.42 seconds. Do not blend the two frames.

Q: Are these performance numbers trustworthy?

A: They are all official figures, from tests Epic commissioned an external vendor to run, but the test environment was not disclosed and no independent third-party tests exist yet. Treat them as reference, not verdict.

Q: Can I install it on Linux now?

A: The new launcher officially supports Linux for the first time, but multiple reports say it is not actually usable during the preview; Fortnite will arrive only after Easy Anti-Cheat is rebuilt on Linux and reaches critical mass with enough distro maintainers, and no timetable has been given.

Q: When does the final version ship?

A: Epic has not published a timetable; the only line on record is that user feedback will shape the final product.

Q: Is Epic trying to topple Steam?

A: Per the store chief’s earlier public comments, the goal on PC is not to topple Steam but to win enough market share that publishers have to list games on the Epic Games Store. The new launcher’s benchmark comparisons are more about proving it deserves to be installed than a declaration of war.

Scope and caveats#

Closing: same door, new way of building it#

Time for something personal.

I have installed nearly every launcher that ever had a name on the PCs I have owned. Some stopped updating long ago; some changed names and faces entirely; only the row of desktop icons kept rotating. So when Epic tore its launcher down to the studs, my first reaction was not surprise but familiarity — behind that door stands one generation after another of publishers who all wanted to be the way you launch your games.

As an art editor who earns a living drawing, the sentence I cared most about in this whole announcement was Keely’s design-and-art-remains-human-led line. Code can be handed to tools; what a launcher ultimately looks like, and whether it feels good to open, still takes people turning it over again and again. That is the detail that settles me most in this story: Epic handed speed to the new tool and kept the storefront’s face for humans.

Whether to install the preview — match yourself to the checklist in section eight. As for where the AI-built road leads: speed is its calling card; whether it holds steady, only time gets to stamp that.

— Zhanlan Yingren, NTGame Art Editor