On May 17, 2025, the Gather network activated its DePIN 2.0 protocol upgrade, fundamentally restructuring how computing power is allocated across its decentralized physical infrastructure network. The upgrade arrives at a moment when the intersection of decentralized computing and artificial intelligence is drawing unprecedented capital and attention. With Bitcoin trading at $103,191 and the broader crypto market capitalization exceeding $2 trillion for Bitcoin alone, infrastructure projects that can demonstrate genuine utility in the AI economy are commanding premium valuations.
The Agentic Protocol
Gather’s DePIN 2.0 introduces a four-tier computing power allocation system designed to balance accessibility, incentivization, and long-term network stability. The protocol replaces a simpler allocation model with a more nuanced structure that rewards different types of participation:
GMT Basic Computing Power (30%): The foundational tier allocates computing power equally based on the real-time status of online mining machines across the entire network. This provides a baseline reward for all active participants regardless of their investment level, ensuring the network maintains a minimum level of distributed computing capacity.
Super Node Computing Power (5%): Reserved for participants who lock up tokens and meet specific key performance indicators, the super node tier incentivizes high-reliability infrastructure operators. These nodes presumably handle the most critical computational tasks and must demonstrate consistent uptime and performance metrics.
Free DPoS Computing Power (35%): The largest allocation tier combines associated computing power — a referral-based incentive where miners earn bonuses based on the activity of their recruited network — with consensus computing power that rewards long-term coin holding stability. This tier is open to all miners for flexible participation.
GF Locked DPoS Computing Power (30%): An exclusive tier available only to members of the CF (Consensus Family) consensus family who lock 2,000 GAT tokens for a 365-day period. Participants earn a share of 30% of total daily network output proportional to their locked tokens.
Neural Network Integration
The restructuring reflects a broader trend in DePIN networks: the need to provide reliable, high-quality computing resources for AI workloads. As AI model training and inference demands continue to grow exponentially, centralized cloud providers face capacity constraints and pricing pressure. DePIN networks like Gather are positioning themselves as distributed alternatives that can offer competitive pricing by leveraging underutilized computing resources worldwide.
The consensus computing power tier is particularly interesting from an AI integration perspective. The mechanism rewards accounts that maintain continuous coin holdings without on-chain transfers — the computing power allocation increases by 1/365 times daily, up to a maximum of 2x after 365 consecutive days. This creates a strong incentive for long-term stability, which translates to predictable computing resource availability for AI clients who need sustained access to processing power.
The associated computing power mechanism, which rewards miners with 10% of first-level referrals’ total votes and 5% of second-level referrals’ votes, creates organic network growth incentives. For AI compute markets, a larger network of distributed nodes means better geographic distribution, lower latency for edge inference tasks, and greater resilience against single points of failure.
Token Utility
The GAT token serves multiple functions within the Gather ecosystem, and the DePIN 2.0 upgrade deepens its utility:
Staking for Computing Power: Miners must hold and stake GAT tokens to participate in computing power allocation. The more tokens staked, and the longer they are held without transfer, the greater the computing power rewards.
Governance and Access: The CF consensus family requires a 2,000 GAT token lock-up, creating significant demand from infrastructure operators who want access to the premium 30% computing power allocation.
Transaction Medium: As the network matures and potentially integrates with AI agent payment protocols like x402, GAT could serve as a medium of exchange for computing resources traded between autonomous agents.
Potential Bottlenecks
Despite the thoughtful design, several concerns warrant attention:
Lock-Up Risk: The 365-day lock-up period for CF family members creates significant illiquidity risk. If the network underperforms or market conditions change dramatically, participants cannot exit their positions. The protocol explicitly warns that locked tokens cannot be redeemed early.
Transfer Penalty: The consensus computing power reset mechanism — which zeroes out accumulated days and voting limit increases if any on-chain transfer occurs — creates a rigid structure that may discourage active token management and reduce overall market liquidity.
Centralization Tendency: The two-tier DPoS structure (free and locked) could lead to concentration of computing power among wealthy participants who can afford the 2,000 GAT lock-up. This runs counter to the decentralized ethos that DePIN networks claim to embody.
Competitive Pressure: Gather competes with established DePIN projects like Render, Akash Network, and newer entrants that have larger ecosystems and more established AI industry partnerships. The DePIN 2.0 upgrade is a step in the right direction, but execution and adoption will determine whether it meaningfully closes the gap.
Final Verdict
Gather’s DePIN 2.0 protocol upgrade demonstrates genuine technical sophistication in its approach to computing power allocation. The multi-tier structure thoughtfully balances the competing demands of accessibility, incentivization, and network reliability. The mechanisms for rewarding long-term participation and organic network growth are well-designed.
However, the project faces significant headwinds in a crowded DePIN market where established players have deeper developer ecosystems and more visible AI industry integrations. The rigid lock-up structures and transfer penalties may deter some participants. For the upgrade to meaningfully impact Gather’s market position, it must translate into demonstrably better computing resource availability and pricing for real-world AI workloads — not just theoretical tokenomics improvements.
This article is for informational purposes only and does not constitute financial or investment advice. Always conduct your own research before making investment decisions.
four tier allocation system sounds good on paper but the 30% basic tier is basically a participation trophy. real compute demand flows to the top tiers and everyone else farms the baseline
degov_skep_ agreed, and the AI economy narrative only works if they can attract actual inference workloads. Gather competing with Render and Akash for the same GPU supply is going to compress margins hard
35 percent of tokens going to referral based DPoS rewards is the loudest possible signal. name one DePIN project where referral tiers didnt collapse into a recruitment spiral
Education is still the biggest barrier to mainstream adoption
The best projects are the ones quietly shipping during bear markets
30% to basic computing power spread across all online machines is just inflationary distribution dressed up as fairness. the referral tier is where it gets ugly
Bear markets are for building — and builders are delivering
gather comparing itself to akash is laughable. akash has enterprise clients paying real money for A100 time. gather has a four tier whitepaper and referral rewards lol
the four tier system is more nuanced than most DePIN projects bother with. separating basic computing from referral incentives prevents the pyramid concern
Zhen Liu the referral tier is literally how pyramid schemes distribute rewards. calling it nuanced doesnt change the structure
kv_cache_burn 35% of tokens allocated to free DPoS referral rewards is textbook inflationary distribution. the four tier system is just complexity theater
kv_cache_burn referral tier rewarding recruitment is literally multi-level marketing with extra steps. calling it nuanced doesnt change the incentive structure
35% allocation to free DPoS is the largest tier. that referral-based incentive model better not turn into a pyramid structure
30% GMT basic tier keeps the floor but 5% for super nodes feels stingy. why would anyone lock tokens for that slice
Tom W 5% for super nodes feels low but the lockup period is what matters. if its short enough the opportunity cost is minimal
Tom W 5% for super nodes feels low but the lockup period matters. if its short enough the opportunity cost is minimal
35% allocation to free DPoS is the largest tier. that referral-based incentive model better not turn into a pyramid structure
35% to free DPoS is the biggest red flag here. nothing free in crypto, someone is subsidizing and extracting elsewhere
compute_renter 35 percent free DPoS means someone is subsidizing compute costs with token inflation. same pattern as every DePIN project that promises cheap infrastructure
BTC at 103k and Gather is still trying to make DePIN computing happen. Akash actually has paying customers, whats GTHs revenue look like
Hannes V. asking the real question. Akash has paying customers and real revenue. Gather has a four tier allocation system and a prayer
depin_skeptic_99 Akash has actual enterprise customers paying for GPU compute. Gather has a four tier allocation system and referral rewards. not even the same category
depin_skeptic_99 Akash actually ships product and has paying customers. Gather has token allocation tiers and a four tier whitepaper. the gap is embarrassing
35 percent of tokens allocated to referral rewards is textbook MLM distribution. calling it a tier system doesnt change the economics
tier_inflation_ akash has actual enterprise customers paying for GPU compute. gather has a four tier referral structure and zero revenue. the comparison shouldnt exist