The market is weak, and subnets are declining. Are we still heading down?
Amidst all this, a solution is emerging to support prices: buybacks. We will analyze this practice through subnet 89, InfiniteHash.
Let's dive in.

Sum of Alpha Prices: soon below 1?

After a brief recovery to 1.09 in early September, the Sum of Alpha Prices has started falling again since September 14. Its price fell from 1.09 down to the 1.02 support (-6.42%).
The level is holding for the moment, proving that some investors are taking advantage of low prices to buy.
We're at a crucial moment. Here's what you need to watch out for in the coming days.
Clear rejection of the 1.02 level. This must happen with a significant buying volume and ideally a bottom pattern (higher lows, inverted HS, double bottom, etc.). For now, the upward movement is too light to consider it a rebound. The zone is still being worked on.
On the other hand, if our 1.02–1.00 zone breaks down with strong volume, it would be very bearish for the long-term trend of subnets.
There’s strong symbolism around these two levels:
1.02 is the ATL support we’ve held since the start of DTAO
1.00 is a psychological level that everyone sees as a floor.
Breaking below both would clearly signal a loss of confidence from investors in dTAO.
Keep in mind: being at 1.00 or 1.01 doesn’t mean the support is broken, just like being at 1.03 doesn’t mean it’s rejected. We need to see significant volume to confirm where the market wants to go.
Fortunately, safeguards exist to help support the market if the Sum of Alpha Prices drops under this threshold.
What happens if the Sum of Subnets fall below 1?
Despite common belief, Sum of Subnets prices can fall below 1. If it happens, a network mechanism to maintain economic balance is triggered. TAO emissions are proportionally reduced across subnets, helping protect the value of the network. This natural adjustment supports a healthier ecosystem over time until demand lifts the sum back above 1.

Market Down, Leaders Down

We use Mentat Minds indexes. Details on calculation are on this page. Mentat 5 and Mentat 15 cover the leaders, while Mentat 70 gives a wider picture of the subnet market.
Highest marketcap subnets were the most resilient since the start of dTAO, but yesterday's dump hurt Mentat 5 (-10.14%) more than Mentat 15 (-9.43%) and Mentat 70 (-5.56%).
Does this mark a turning point where the overall subnet market shows more stability than its leading subnets? Or was it just a move that hit the biggest players hardest?
If we look closely at the top subnets, we see that only Ridges has experienced a sharp decline, which led the index at the start of the day before losing 19.04%.
There doesn’t seem to be any fundamental news behind this Ridges drop. This is the risk of subnets with a very high ADR. We explained this in our September 9th’s edition.

We will monitor the evolution of these indexes in future reports
Largest teams on Bittensor
Macrocosmos seems to be bucking the market trend. Their price has even been rising (+8.33% over the last two weeks).

Conversely, Rayon Labs’ subnets are going through a difficult stretch: -20.02% over the same period.


InfiniteHash, exemplary buybacks
If you want to skip the subnet analysis and directly dive into buyback mechanisms, click here.

Broken support?
InfiniteHash is in a key zone, around the purple support level. We are below the line, but not enough to say that it has been broken. We will see in the coming days whether the bulls hold out. If the move extends, the next support to watch is 0.003 (in green).
However, InfiniteHash has performed very well in recent months. It is up 217% between July 7 and today, while the market (Mentat 70) has fallen 32% over the same period. (Mentat 70 graph just below)

Mining Alpha tokens with Bitcoin hashrate
On InfiniteHash, Miners join by linking their Bitcoin ASICs to the subnet pool and their hashrate is tracked on-chain. The subnet earns income in bitcoin based on the miners hashrate. Instead of classic Bitcoin payouts, participants earn Alpha tokens directly on Bittensor, reflecting the work they contribute.
The subnet currently ranks 53rd, with a market cap of $2.73 million, a volume of $147.33K in the last 24 hours, making it the 42nd most traded token. Volume is relatively healthy for its size, but still low in absolute terms. Only 108 people have traded the token during this timeframe.
Beware of centralization

The three largest wallets hold a large portion of the supply: 26.05%.
Usually, the three largest wallets hold a maximum of 18% of the supply. There is therefore significant centralization, which represents a risk in the event of massive sales.
But the behavior of these wallets is actually reassuring.
The first seems to belong to the subnet Owner. There have been no sales or transfers from it. This suggests the subnet owner has very strong convictions in its subnet.
The second wallet is used for buybacks. No sales to report and continuous purchases. A very interesting mechanism that we will study shortly.
The third largest portfolio has never sold or made any transactions. It remains completely passive.
It will be necessary to closely monitor these wallets, which could have a major impact in the future. Despite this centralization, a very interesting mechanism has been put in place to transfer the value of the subnet into its token.
InfiniteHash: the best way to support their token?
InfiniteHash posts updates on their Twitter account: they purchase several thousand dollars worth of their Alpha every day.

If we look onchain, we can see that these purchases are indeed taking place every day. This puts constant buying pressure on the subnet token.

We can easily see that InfiniteHash's buyback is transparent and very rigorous.
A buyback is when a subnet uses its resources to buy its own token on the market, aiming to support or increase the token’s price and incentivize network participants.
All revenues from the subnets (mined bitcoins, approximately $5k per day) are used to buyback the subnet token. This mechanism continuously buys up Alpha tokens, supporting the token’s value.
This setup creates a simple but powerful feedback loop:
Miners contribute their bitcoin hashrate to the subnet and are paid with new Alpha token emission.
Every Bitcoin mined by the subnet is immediately sold, and used to buy Alpha tokens on the market—creating steady buying pressure on the subnet.
As more miners join, more Bitcoin is mined and more Alpha is bought up, which makes Alpha more valuable.
As Alpha’s value rises, the subnet becomes even more attractive to new miners and holders, strengthening both the ecosystem and rewards for participants.
A key improvement would be automation, since the team currently handles the entire buyback process manually. They are very consistent and have every interest in continuing to be so for the future of their subnet.
As we saw in the technical analysis, this practice seems to have supported the share price rather well. This is due to the purchases made, but also thanks to the positive sentiment that this type of practice generates among investors.

The buyback model is certainly one of the most important narratives for Bittensor in the coming months. When done rigorously and transparently, as InfiniteHash has shown, it allows the subnet token to be placed at the center of the subnet tokenomics and gives it value and interest.
Since buybacks are made with the revenue generated by subnets, the subnet must earn money from its activities in order to make real buybacks.
It will be very interesting to see how other subnets on the network manage to integrate such models to add value to their tokens.
Subnets need stability, investors need reassurance. This practice seems to be able to rally everyone in order to bring Bittensor to a healthier market.

Thank you for reading this newsletter. If you are interested in the subnet market, please feel free to click on the following link.
Disclaimer: This newsletter does not constitute investment advice. It is a tool designed to help you understand the Bittensor subnet market. Always do your own research before investing.


