Author: Teng Yan, Chain of Thought
Translation: Golden Finance xiaozou
In 2021, I was an Axie Infinity player and operated a small scholarship guild. If you didn't experience that era, let me tell you — it was absolutely wild.
Axie Infinity made people realize that cryptocurrency and gaming could be combined. Essentially, it is a simple Pokémon-style strategy game where players build a team of 3 Axies (very fierce warriors), each with unique abilities. You can lead your team against others, earning SLP token rewards by participating in the game and winning.
But what really excites non-gamers is the potential to earn money through gaming. The rapid rise of Axie was due to two major mechanisms:
The first is Breeding Axies. Get two Axies, use SLP tokens to breed them, and voilà — a new Axie is born that combines the unique abilities of the two original Axies. Thus, rare and powerful Axies (what players call OP Axies) became hot commodities, giving rise to a bustling breeding market.
The second mechanism is the scholarship program. Players from around the world start lending Axies to 'scholars.' These players often come from developing countries like the Philippines or Argentina, who cannot afford the upfront cost of over $1,000 to purchase 3 Axie NFTs. Scholars earn tokens by playing daily and share profits with the scholarship guilds, which typically take a 30-50% cut.
At its peak, especially during the pandemic in 2019, Axie had a significant impact on the local economies of developing countries. In the Philippines (where about 40% of Axie Infinity users are located), many players earned far above the minimum wage. Guilds profited handsomely.
This addresses a key issue for game developers: player liquidity. By incentivizing players to actively play for several hours each day, Axie ensures that there is always an opponent waiting, providing a more engaging experience.
But this comes at a cost.
To solve the player liquidity issue, Axie gave away large amounts of tokens to incentivize player participation. The story starts here. With no cap on SLP, the tokens inflated wildly, causing prices to plummet and the ecosystem to collapse. When tokens depreciated, players left. Axie quickly turned from a 'play-to-earn' darling into a cautionary tale.
But what if there was a way to solve the player liquidity issue without relying on unsustainable token economics?
This is exactly what ARC / AI Arena has been quietly working on for the past three years. Now, it's starting to bear fruit.
1. Player liquidity is the lifeblood.
Player liquidity is the lifeblood of multiplayer games and key to long-term success.
Many Web3 and indie games face the 'cold start' problem — too few players to quickly match or form a thriving community. They do not have the marketing budgets or natural IP awareness that big game companies possess. This leads to long wait times, inability to match, and high churn rates.
These games typically suffer a slow, painful demise.
Therefore, game developers must prioritize player liquidity from the very beginning. Games need activities of various kinds to maintain fun — chess requires two players, while large-scale battles need thousands of players. Skill matching mechanisms further raise the bar, requiring more players to maintain fairness and appeal.
For Web3 games, the stakes are higher. According to Delphi Digital's annual gaming report, the user acquisition cost for Web3 games is 77% higher than that of traditional mobile games, making player retention critical.
A strong player base ensures fair matching, a vibrant game economy (i.e., more item trading), and more active social interactions, making games more enjoyable.
2. ARC — AI Game Pioneer
Developed by ArenaX Labs, ARC is leading the future of AI online gaming experiences. In short, they use AI to solve the player liquidity issues that plague new games.
Today, most AI bots in games are simply too poor. Once you spend a few hours mastering the tricks, these bots become very easy to defeat. They are designed to help new players but fail to offer much challenge or engagement for experienced players.
Imagine AI players' skills being comparable to top human players. Imagine being able to compete against them anytime, anywhere, without waiting for matchmaking. Imagine training your AI player to mimic your play style, owning it, and earning rewards based on its performance.
This is a win-win for both players and game companies.
Game companies use human-like AI bots to make games popular, increase player liquidity, enhance user experience, and improve retention rates — key factors for new game entrants to survive in a competitive market.
Players gain a new way to engage with the game, building a stronger sense of belonging as they train AI and compete against it.
Let’s take a look at how they are doing it.
3. Products and Architecture
Parent company ArenaX Labs is developing a range of products to address player liquidity issues.
Existing product: AI Arena, an AI fighting game.
New product: ARC B2B, an AI-driven game SDK that can be easily integrated into any game.
New product: ARC Reinforcement Learning (RL).
(1) AI Arena: The game.
AI Arena is a fighting game reminiscent of Nintendo's Super Smash Bros, featuring a variety of quirky cartoon characters battling in an arena.
But in AI Arena, every character is controlled by AI — you are not playing as the warrior but as their coach. Your task is to use your strategy and expertise to train your AI warrior.
Training your warrior is like preparing a student for battle. In training mode, you turn on data collection and create combat scenarios to fine-tune their actions. For example, if your warrior is close to the opponent, you can teach them to block with your shield and then follow up with a combo. How about fighting from a distance? Train them to launch ranged attacks.
You can control what kind of data you collect, ensuring that only the best actions are recorded for training. With practice, you can refine hyperparameters for more technical advantages or simply use user-friendly default settings. Once training is complete, your AI warrior can enter combat.
Starting anything is hard — training an effective model takes time and experimentation. My first warrior fell off the platform several times, not due to being hit by an opponent but just failing. But after several iterations, I successfully created a well-performing model. It's very satisfying to see your training pay off.
AI Arena introduces additional depth through NFT warriors. Each NFT character has unique appearance traits and combat attributes that affect gameplay. This adds another layer of strategy.
Currently, AI Arena operates on the Arbitrum mainnet, and only those with AI Arena NFTs can access it, maintaining exclusivity within the community while refining gameplay. Players can join guilds, gather champion NFTs and NRN for on-chain battle rankings, and earn rewards. This is to attract loyal players and drive competition.
Ultimately, AI Arena serves as a showcase for ARC's AI training technology. While this is their entry point into the ecosystem, the true vision extends far beyond the game itself.
(2) ARC: Infrastructure.
ARC is an AI infrastructure solution designed for games.
The ArenaX team built from scratch, even developing their own gaming infrastructure, because existing solutions like Unity and Unreal did not meet their vision.
Over the past three years, they have meticulously designed a powerful tech stack capable of handling data aggregation, model training, and model verification for imitation and reinforcement learning. This infrastructure is the backbone of AI Arena, but its potential is much greater.
As teams continually refine their technology, third-party studios are beginning to find ARC, seeking licensing or white-labeling for the platform. Recognizing this demand, they formalized ARC's infrastructure into a B2B product.
Today, ARC directly collaborates with game companies to provide AI gaming experiences. Its value proposition is:
Permanent player liquidity as a service.
Integrating AI gameplay as a simple integration.
Permanent player liquidity as a service.
ARC focuses on human behavior cloning — training specialized AI models to mimic human behavior. This differs from the primary use cases of AI in today's games, which employ generative models to create game assets and use LLMs to drive dialogue.
With the ARC SDK, developers can create human-like AI agents and expand them according to game needs. The SDK simplifies the heavy lifting. Game companies can introduce AI without dealing with complex machine learning.
After integration, deploying AI models requires just one line of code, with ARC handling infrastructure, data processing, training, and backend deployment.
ARC collaborates with game companies to help them:
Capture original gameplay data and transform it into meaningful datasets for AI training.
Identify key gameplay variables and decision points related to game mechanics.
Mapping AI model outputs to in-game activities to ensure smooth functionality — for example, linking AI's 'right-click' output to specific game controls.
How does AI work?
ARC employs four types of models for game interactions:
Feedforward neural networks: Suitable for continuous environments with numerical features like speed or position.
Tabular agents: Especially ideal for games with a limited number of discrete scenarios.
Hierarchical and convolutional neural networks are under development.
There are two interactive spaces related to ARC's AI models:
The state space defines the agent's understanding of the game at any given moment. For feedforward networks, this is a combination of input features such as player speed or position. For tabular agents, these are discrete scenarios the agent may encounter in the game.
The action space describes what agents can do in the game, ranging from discrete inputs (like pressing buttons) to continuous controls (like joystick movements). This maps to game inputs.
The state space provides input for ARC's AI models, which process the input and generate output. These outputs are then transformed into game actions through the action space.
ARC closely collaborates with game developers to identify the most critical features and design state spaces accordingly. They also test various model configurations and sizes to balance intelligence and speed, ensuring smooth and engaging gameplay.
According to the team, there is particularly high demand from Web3 companies for their player liquidity services. These companies are paying for better player liquidity, and ARC will allocate a significant portion of this revenue for NRN token buybacks.
Bringing AI gameplay to players: Trainer platform.
The ARC SDK also allows web3 companies access to their game's trainer platform, enabling players to train and submit agents.
Similar to AI Arena, players can set up simulations, gather gameplay data, and train blank AI models. These models evolve over time, integrating new gameplay data while retaining prior knowledge, eliminating the need to start from scratch with each update.
This opens up exciting possibilities: players can sell their custom-trained AI agents in the market, creating a new layer of in-game economy. In AI Arena, skilled trainers can form guilds and provide training expertise to other companies.
For companies that fully integrate agent capabilities, the concept of Parallel Play becomes vivid. AI agents are available around the clock and can participate in multiple matches or game instances simultaneously. This addresses player liquidity issues and creates new opportunities for user engagement and revenue.
But that's not all...
(3) ARC RL: From one-to-one to many-to-one.
If AI Arena and the ARC trainer platform feel like single-player modes (where you can train your own AI models), then ARC RL resembles multiplayer mode.
Imagine this: an entire game DAO pooling its gameplay data to train a shared AI model, with everyone co-owning the model and benefiting from it. These 'master agents' represent the collective wisdom of all players and can change esports through competition driven by collective effort and strategic collaboration.
ARC RL uses reinforcement learning (i.e., 'RL') and crowdsourced human gameplay data to train these 'super-intelligent' agents.
The way reinforcement learning works is by rewarding agents for optimal behavior. It is particularly effective in games because the reward function is clear and objective, such as damage dealt, gold earned, or victories.
This is unprecedented:
DeepMind's AlphaGo defeated professional human players in Go matches, training through millions of self-generated games, refining its strategy with each iteration.
I hadn't realized this before, but even before chatGPT was created, OpenAI was already well-known in the gaming circle.
OpenAI Five crushed top human players in Dota 2 using reinforcement learning, defeating the world champions in 2019. It mastered advanced strategies like teamwork through accelerated simulations and massive computational resources.
OpenAI Five runs millions of games daily, equivalent to 250 years of simulated games each day, powered by 256 GPUs and 128,000 CPUs. By skipping graphic rendering, it significantly accelerated its learning speed.
Initially, the AI exhibited unstable behavior, such as wandering aimlessly, but quickly improved. It mastered basic strategies like crawling on paths and stealing resources, eventually evolving into complex maneuvers like ambushing.
The key concept of reinforcement learning is that AI agents learn how to succeed through experience, rather than being directly told what to do.
ARC RL differentiates itself by using offline reinforcement learning. AI agents learn not from their own trial and error, but from the experiences of others. It's like a student watching videos of others riding a bike, learning from their successes and failures, utilizing that knowledge to avoid falling and improve faster.
This approach offers an additional benefit: collaborative training and shared ownership of the models. This not only makes powerful AI agents more accessible but also aligns the motivations of players, guilds, and developers.
In creating 'super-intelligent' game agents, there are two key roles:
Sponsors: Leader-like guilds that stake large amounts of NRN tokens to start and manage RL agents. Sponsors can be any entity, but are likely to be gaming guilds, DAOs, web3 communities, or even popular on-chain personalized agents like Luna.
Players: Individuals who stake a small amount of NRN tokens to contribute their gameplay data for training agents.
Sponsors coordinate and guide their player teams to ensure high-quality training data, giving their AI agents a competitive edge in agent competitions.
Rewards are allocated based on the performance of super agents in competitions. 70% of the rewards go to players, 10% to sponsors, and the remaining 20% to the NRN treasury. This structure provides a consistent incentive mechanism for all participants.
Data Contribution
How do you make players willing to contribute their gameplay data? Not easy.
ARC makes providing gameplay data simple and rewarding. Players do not need expertise; they just need to play the game. At the end of a session, they are prompted to submit data to train a specific agent. A dashboard tracks their contributions and the agents they support.
ARC's attribution algorithm ensures quality by assessing contributions and rewarding high-quality, impactful data.
Interestingly, even if you're a poor player (like me), your data is still useful. Bad gameplay can help agents learn what not to do, while skilled gameplay can teach the best strategies. Redundant data is filtered out to maintain quality.
In short, ARC RL is designed as a low-friction, mass-market product centered around co-owning agents that exceed human capabilities.
4. Market Size
ARC's technology platform is versatile, supporting various types of games such as shooting games, fighting games, social casinos, racing, card trading games, and RPGs. It is tailored for games that need to maintain player engagement.
ARC's products primarily target two markets:
ARC primarily focuses on indie developers and companies rather than established giants. Due to limited brand influence and distribution resources, these smaller companies often struggle to attract players early on.
ARC's AI agents tackle this issue by creating a vibrant game environment from the start, ensuring dynamic gameplay even in the early stages of the game.
This may surprise many, but the independent gaming sector is indeed a major force in the gaming market:
99% of games on Steam are indie games.
In 2024, indie games accounted for 48% of total revenue on Steam.
Another target market is Web3 games. Most Web3 games are developed by emerging companies that face their own unique challenges, such as wallet logins, crypto skepticism, and high user acquisition costs. These games often have player liquidity issues, and AI agents can fill the gaps to keep the games engaging.
Although Web3 games have recently struggled due to a lack of engaging experiences, signs of recovery are emerging.
For instance, one of the earliest AAA Web3 games, Off the Grid, recently achieved early mainstream success, with 9 million wallets conducting 100 million transactions in the first month. This paved the way for broad success in the industry and created opportunities for ARC to support this revival.
5. ARC Team
The founding team behind ArenaX Labs has rich expertise in machine learning and investment management.
CEO and CTO Brandon Da Silva previously led machine learning research at a Canadian investment firm, focusing on reinforcement learning, Bayesian deep learning, and model adaptability. He pioneered a $1 billion quantitative trading strategy centered on risk parity and multi-asset portfolio management.
COO Wei Xie manages a $7 billion liquidity strategy portfolio at the same company and oversees its innovative investment projects focused on emerging fields such as AI, machine learning, and Web3 technology.
ArenaX Labs secured $5 million in seed funding in 2021, led by Paradigm and participated by Framework ventures. The company received $6 million in funding in January 2024, led by SevenX Ventures, FunPlus / Xterio, and Moore Strategic Ventures.
6. NRN Token Economics — A Healthy Reform
ARC/AI Arena has a token — NRN. Let’s first take stock of the current situation.
Examining the supply and demand sides will give us clearer insights into the trends.
(1) Supply Side
The total supply of NRN is 1 billion, with approximately 409 million (40.9%) in circulation.
At the time of writing, the token price is $0.72, which means a market cap of $29 million, with a fully diluted valuation of $71 million.
NRN was launched on June 24, 2024, with 40.9% of the circulating supply coming from:
Community airdrop (8% of the total).
Foundation treasury (10.9%, with 2.9% unlocked, linear unlock over 36 months).
Community ecosystem rewards (30%).
Most of the circulating supply (30% of the 40.9%) consists of community ecosystem rewards, with the project managing these tokens and strategically allocating them to staking rewards, game rewards, ecosystem growth initiatives, and community-driven programs.
The unlock schedule is reassuring, with no major events in the short term:
The next unlock is the foundation's OTC sales (1.1%), starting in December 2024, with a 12-month linear unlock. This will only increase the monthly inflation rate by 0.09%, unlikely to raise significant concerns.
The distribution for investors and contributors (50% of the total supply) will not begin to unlock until June 2025, and even then, it will be a linear unlock over 24 months.
Currently, selling pressure is expected to remain quite manageable, mainly stemming from ecosystem rewards. The key is trusting the team to strategically deploy these funds to drive protocol growth.
(2) Demand Side
NRN v1 — Player Economics.
Initially, NRN was designed as a strategic resource linked to the AI Arena game economy.
Players stake NRN on AI players, earning rewards if they win and losing part of their stake if they lose. This creates a dynamic of direct stakes, transforming it into a competitive sport and providing economic incentives for skilled players.
Rewards are distributed using the ELO system to ensure skill-based balanced payouts. Other sources of income include game item purchases, cosmetic upgrades, and entry fees for competitions.
The initial token model relied entirely on the success of the game and the continuous influx of new players willing to purchase NRN and NFTs to participate in the game.
Let's talk about why we're so excited...
NRN v2 — Player & Platform Economics.
NRN's improved v2 token economics introduces powerful new demand drivers by extending the utility of the token from AI Arena to the broader ARC platform. This evolution transforms NRN from a game-specific token into a platform token. In my view, this is a very positive shift.
The three new demand drivers for NRN include:
Revenue from ARC integration. Game companies that integrate ARC will generate income for the treasury through integration fees and ongoing royalties tied to game performance. The treasury funds can drive NRN buybacks, grow the ecosystem, and incentivize players on the trainer platform.
Trainer market fees. NRN derives value from fees charged in the trainer field, allowing players to trade AI models and gameplay data.
Participating in ARC RL staking: Both sponsors and players must stake NRN to join ARC RL. As more players enter ARC RL, the demand for NRN increases accordingly.
Especially exciting are the revenues for game companies. This marks a shift from a purely B2C model to a hybrid B2C and B2B model, creating a sustained influx of external capital into the NRN economy. As ARC has a broader target market, this revenue stream will exceed what AI Arena can generate on its own.
While the trainer market fees have potential, they depend on whether the ecosystem can reach critical scale — enough games, trainers, and players to sustain active trading activity. This is a long-term endeavor.
In the short term, staking in ARC RL may be the most direct and reflective demand driver. A well-funded initial reward pool and the excitement of new product launches could spark early adoption, driving up token prices and attracting participants. This creates a feedback loop of rising demand and economic growth. However, conversely, if ARC RL struggles to maintain user engagement, demand may quickly disappear.
The potential for network effects is enormous: more games → more players → more games joining → more players. This virtuous cycle can position NRN as the core token in the Crypto AI gaming ecosystem.
7. The Mother of Game AI Models
What is the conclusion? The advantage of ARC is its ability to promote various game types. Over time, it enables them to collect a unique database of specific gameplay. As ARC integrates with more games, it can continuously feed this data back into its ecosystem, creating a virtuous cycle of growth and improvement.
Once this cross-sectional game dataset reaches critical quality, it will become a highly valuable resource. Imagine using it to train general AI models for game development — opening up new possibilities for large-scale design, testing, and optimization of games.
It's still early, but in the era where data is the new oil, the potential in this area is limitless.
8. Our Thoughts
NRN evolves into platform gaming — token repricing.
With the launch of ARC and ARC RL, the project is no longer just a single product game company, it now positions itself as a platform and AI gaming solution. This shift should lead to a repricing of the NRN token, which was previously constrained to the success of AI Arena. The introduction of new token sources through ARC RL, coupled with revenue-sharing agreements with game companies and external demand from trainer transaction fees, creates a broader and more diversified foundation for NRN's utility and value.
Success is closely tied to game partnerships.
ARC's business model ties its success to the companies it partners with, as revenue streams are based on token distribution (in Web3 games) and game royalty payments. Games closely tied to this are worth watching.
If ARC games achieve great success, the resulting value will flow back to NRN holders. Conversely, if the partnered games struggle, the value flow will be constrained.
Looking forward to more integrations with Web3 games.
The ARC platform is ideal for Web3 games, where competitive gameplay with incentive mechanisms perfectly complements the existing token economy.
By integrating ARC, Web3 games can immediately enter the 'AI agent' narrative. ARC RL gathers the community, motivating them towards a shared goal. This also opens up new opportunities for innovative mechanisms, such as making 'games to airdrop' activities more appealing to players. By combining AI and token incentives, ARC adds depth and excitement that traditional games cannot replicate.
AI gameplay has a learning curve.
AI gameplay has a steep learning curve, which can create friction for new players. It took me an hour to figure out how to properly train my players in AI Arena.
However, the player experience friction in ARC RL is lower because AI training is handled in the backend while players play the game and submit data. Another pending question is how players will feel when they know their opponent is AI. Does this affect them? Will it enhance or diminish the gaming experience? Only time will tell.
9. A Bright Future
AI will open up entirely new breakthrough experiences in the gaming world.
Teams like Parallel Colony and Virtuals are advancing the development of autonomous AI agents, while ARC forges its niche by focusing on human behavior cloning — offering an innovative way to tackle player liquidity challenges without relying on unsustainable token economics.
The transition from a game to a mature platform is a huge leap for ARC. This not only opens up larger opportunities through collaboration with game companies but also redefines how AI integrates with gaming.
With its improved token economics and the potential for strong network effects, ARC’s bright path seems to have just begun.
