Just been announced by Dom on X
Now reading it. I have some immediate reactions but will wait until I properly zip my glass of red wine 
and go through it in detail.
9 Likes
8 year gang is now 2 year gang. I wonder what the reactions are going to be like too ^-^
1 Like
My first impressions as I crawl through the document: the āsupply sideā of the paper is well thought and put together, but the ādemand sideā is way more speculative and its thesis is anchored on category creation, not just market share capture. Thatās a much harder problem and has not minor execution risks.
2 Likes
Indeed. The one-time dissolve delay reduction from 8 years to 2 years for all neurons is ambitious but it changes the risk-reward profile long-term stackers accepted when locking in. Will the 10% bonus until 2030 be accepted as āadequate compensationā for losing, what, 6 years of expected lockup premium? But, equally, in any scenario it proposes (or imposes?) an unilateral change in the stacking contract which could lead to some reputational damage.
OTOH isnāt the ādissolution queueā (to cushion any selling pressure coming from neuron dissolution) an implicit admission that many folks will get pi**ed-off by this?
Unless, in haste, Iām reading it wrongly.
Thereās a lot to dissect here.
3 Likes
I see it as fair deal.
Minting rate -44%
2 years 7.0%
2 years with 8-year gang flag 7.7% ( 36.4% less rewards)
Without changes, ICP will be 2-3usd for long time, with changes we actually can have price over 10 usd, might even reach 100+. With sustainable tokenomics, institutions can jump in and pump even higher.
10 Likes
I see it as a fair deal too.
Plus: people often forget the age bonus. With max age bonus, this is ~9.625% APY until 2030, not just 7ā7.7%. for 8yg
7 Likes
Back to this, which I think is not minor considerationā¦
-
As I said, the āsupply sideā is well put together, with specific formulas (the convex reward curve), concrete figures (the 36% estimated rewards reduction), and other etc.
-
In contrast, the ādemand sideā of the paper groups multiple āaspirationalā statements, e.g. āā¦DFINITY believes that forthcoming onchain cloud engines and growth in the self-writing cloud paradigm will push the burn rate well beyond this thresholdā¦ā (from p. 3). There is an acknowledgement of the need for a 15x increase in cycle burn rate, but I canāt see in the document any market sizing, adoption modeling or sensitivity analysis that anchors why this will in fact occur.
IMHO this is a bit problematic. Difficult to claim a 70% reduction when 26 percentage points of that reduction rests on hope rather than mechanism.
How come 26 pp.? Itās in the document for grabs (also on p. 3):
- Voting rewards: from 5.88% to 3.45% for a 41% reduction.
- Node rewards: from 3.84% to 1.97% or a 49% reduction.
- The blend āā¦representing an absolute reduction of 4.30 percentage points (relative reduction of 44%)ā
So 70% - 44% = 26 pp. The 44% is clearly well defined through specific mechanics (simulation results from December 2025 neuron data, specific reward multiplier changes, and quantified node provider exit scenarios). The wedge to get to the 70% target is to a great extent aspirational.
It would have been great if they modeled (and substantiated) the projected adoption curves for cloud services and Caffeine. Thatās IMHO lacking in the document as it stands now, but could be done.
2 Likes
I have not made a judgement, rather made a question which directly follows from reading the document.
That said, on second thoughts I see the point: youāre accepting a 22% yield reduction in exchange for a 75% reduction in maximum lockup duration. In risk-adjusted terms, this is arguably favorable, as illiquidity risk is substantially reduced (and that should command a lower premium anyway).
However, from what I read neurons that start dissolving lose their age bonus. So the 9.625% APY only holds while the neuron remains non-dissolving (otherwise it drops to 7.7% = 2 yr. base at 7% + 10%). This indeed creates an incentive to stay locked, which is presumably intentional, but it does mean the āfair dealā depends entirely on continued commitment rather than being a compensation for past commitment.
Thatās a small nuance.
2 Likes
First thoughts after reading:
Voting rewards
The changes to voting rewards look like a good step in the right direction to reduce inflation and better align longāterm incentives for committed participants. The considerably lower periods may help bolster DeFi, TVL and Liquid Staking, and hopefully leads to increased participation in staking and voting.
Node rewards
Adjusting node rewards toward more sustainable levels and better economics for the network makes sense in principle. However, I was surprised which such large reductions, and as a nonānodeāprovider it is hard to judge how this will affect realāworld operations.
(It was positive in the sense that not all burden was put on long term stakers)
UTOPIA
UTOPIA is presented more as a big-picture narrative than as a concrete, nearāterm plan with realistic timelines, costs, and revenue projections. Right now it feels more like something that has been added to balance the story around reward cuts rather than a wellāspecified pillar that can be evaluated on its own merits.ā
Caffeine AI
Caffeine AI is framed as a major growth and burn driver but without detailed numbers on margins, expenses, profits, or how much sustainable demand and burn it can realistically generate. Without that data, it is hard to separate solid business expectations from optimistic marketing language.ā
Incentives, builders, and waste
The paper says too little about how to actively bring (non-caffeine AI) builders back, fund them effectively, and turn their work into lasting utility and burn for the network. It also does not seriously address how to tackle internal waste or questionable spending.
If there is no good incentive programs to attract builders, does that mean that we expect all builders to pay Dfinity using Caffeine AI 
4 Likes
To achieve the overall Mission 70 target of a 70% inflation reduction (from
9.72% to 2.92%), an additional demand impact of 26% is required beyond the
44% reduction from supply-side measures. At current price levels, this requires
increasing the cycle burn rate from the current 0.05 XDR per second to 0.77
XDR per second.
If token value goes up, Node providers get less minted ICP. Even if ICP goes x2, node rewards go to 1.97 to 0.99. With more burn, we might achieve that goal.
1 Like
We have to approve the changes. Yes ICP still needs adoption but future 8 year gang promo must end.
2 Likes
Raise your hand if you have a 8YG tattoo
1 Like
Where have you seen UTOPIA in the paper. Thereās a new concept of cloud engines. But in my understanding thatās not the same as UTOPIA.
2 Likes
As you say āif token value goes up.ā I hope that indeed happens, donāt get me wrong, but thatās speculative in my book.
The pathway you outlined is indeed possible, but the paper seems to put all the bets on a 15x demand acceleration.
That said, on p. 12 it states that node rewards in XDR are converted to ICP at the 30-day average price. Consequently, node reward inflation (as a percentage of supply) is inversely proportional to ICP price. If ICP price doubles, you are mechanically shaving another percentage point off inflation without any increase in cycle burn. But it cuts both ways: if the measures fail to support ICPās price, this mechanism is limp.
However, now that I think of it, it seems curious they didnāt fully or more explicitly articulated this logic (price appreciation partially endogenous to the supply cuts themselves) in the paper.
- Supply-side cuts reduce sell pressure
- Reduced sell pressure is price-supportive
- Higher price means fewer ICP minted for node rewards
- This compounds the supply reduction
As you say, the supply-side measures may have indirect demand-side effects through the price channel (and that 44% could bootstrap towards the 70% target through such second-order effects).
1 Like
I cant understand the Node provider dynamics. Werenāt the node provider rewards variable and based on the conversion rate and number of nodes? How does he come to the inflation reduction from 3.84% to 1.97%. Maybe its too late and i cant focus enough.
Yeah, you are correct
I never saw UTOPIA in the white paper
The cloud engines somewhat described UTOPIA but also more like a hosting service.
Perhaps the difference is this:
- If you want UTOPIA you need high performance web servers, and if you dont have that you can source this from hosting providers?
Anyone else knows this?
The paper calculates both the 3.84% baseline and the 1.97% target at the same price level, with the 3.84% seemingly taking January 2026 prices, as in p. 3 it says ātotal ICP minting decreases from 9.72% (January 2026) to 5.42% (January 2027)ā¦ā
Link this up to what is stated on p. 15: āPlease note that while the actual node reward inflation levels depend on the ICP/XDR conversion rate, the relative reduction percentage does not depend on it.ā
If you cut XDR-denominated rewards by 49%, you cut ICP minting by 49% regardless of what conversion rate you use, because both numerator and denominator scale identically with price.
But the document does not say what price or conversion rate was used for the 3.84% baseline: spot price (and when), 30-day average, some reference value?
1 Like
Iām super excited about the cloud engine part plus the new subnet organization has a lot of potential in so many applications. We need more information on it tbh. The new tokenomics plan is has good tradeoff, it can really push the price up which is more important than short term rewards and overall network health. Looking forward for more information and as always execution is key, but this is very bullish.
6 Likes
I hope this is something we can pass ASAP, by the end of the month or a really soon timeline. We desperately need the change.
1 Like
Can we find a list of how much XDR nodes are individually getting paid somewhere?