The economic case
Once at least 50% of bakers (115 of 229, per the TzKT API) activate BLS (tz4) consensus keys, the Tallinn protocol’s all-bakers-attest model activates, replacing today’s sampled-subset attestation with universal attestation aggregated into a single signature per block. This drives the network toward the Seoul protocol’s full theoretical efficiency improvement of ~886 MB/day eliminated (a 63× reduction, from ~900 MB/day to ~14 MB/day).
The problem: tz4 adoption is plateauing well short of threshold. As of today, 77 bakers (33.6%) have switched — but the rate has collapsed from a peak of 19 new adopters in January 2026 to just 4 over the last 30 days. A logistic curve fit to the 8 months of post-Seoul data projects a natural asymptotic ceiling around ~81 bakers (~35.5% of the network) — see the trajectory data below. Without intervention, all-bakers-attest never activates and the network captures only partial efficiency gains in perpetuity. The hardware-signer compatibility issue (Ledger doesn’t support BLS) and cloud KMS gaps are real barriers that organic adoption hasn’t overcome and likely won’t.
A detailed cost analysis of this bloat across the baker + core-infrastructure fleet (~250 nodes) — using a realistic 50/50 split between cloud-hosted and bare-metal/home operators, and accounting for both bandwidth egress and storage — yields a conservative floor of ~$47,000 per year in collective operational cost that BLS aggregation would eliminate, in perpetuity.
That figure is a floor, not a ceiling. The full Tezos network — per TzKT’s node statistics — is closer to ~600+ nodes once you include indexers, block explorers, RPC providers, wallet backends, and dApp infrastructure. Most of those non-baker nodes run Archive or Full nodes on premium cloud infrastructure, where the higher all-cloud cost rates ($0.08/GB egress, $0.08/GB-month storage) actually apply. Scaling the analysis to the full network puts real annual savings plausibly in the $100,000–$140,000 per year range.
A one-time bonus pool of approximately $95,000 to accelerate reaching the 50% threshold is therefore recouped by collective network savings within 8–12 months, with every subsequent year delivering $100k+ in pure efficiency gain distributed across the operator community.
Adoption trajectory
Tz4 adoption since the Seoul upgrade activated on September 19, 2025 (data: TzKT update_consensus_key operations, filtered for publicKey starting with BLpk, deduped by sender):
| Month | New adopters | Cumulative | % of network |
|---|---|---|---|
| 2025-09 | 7 | 7 | 3.1% |
| 2025-10 | 12 | 19 | 8.3% |
| 2025-11 | 3 | 22 | 9.6% |
| 2025-12 | 14 | 36 | 15.7% |
| 2026-01 | 19 | 55 | 24.0% |
| 2026-02 | 13 | 68 | 29.7% |
| 2026-03 | 4 | 72 | 31.4% |
| 2026-04 | 1 | 73 | 31.9% |
| 2026-05* | 4 | 77 | 33.6% |
* through May 17, 2026
Two adoption bursts (Seoul activation; Tallinn activation on January 24, 2026) followed by sharp deceleration. Recent rate:
-
Last 30 days: 4 adoptions
-
Last 60 days: 8 adoptions
-
Last 90 days: 12 adoptions
-
April 2026: 1 adoption the entire month
Fitting a logistic curve to the cumulative adoption series yields:
-
Asymptotic ceiling K ≈ 81 bakers (~35.5% of the network)
-
Inflection point: ~January 4, 2026 — already months in the past
-
The curve is in its laggard tail
This is the classic S-curve pattern: easy adopters first, harder ones never. The remaining 152 bakers face barriers organic momentum doesn’t overcome — Ledger hardware incompatibility, cloud KMS gaps, change-aversion at corporate operators, plain inertia. Without external incentive, the projected trajectory leaves the network roughly 34 bakers shy of the threshold, indefinitely. All-bakers-attest stays dormant. The full bandwidth and storage savings stay locked.
Anyone can reproduce this analysis: pull the TzKT operations endpoint, filter on the BLS public-key prefix, dedupe by sender address, fit a logistic.
By the numbers — annual data saved
Translating those dollar figures back into raw data shows what’s actually being eliminated:
-
~800 TB/year of network bandwidth (gossip egress across the 250-node baker + core-infrastructure fleet)
-
~28 TB/year of new permanent storage across the ~88 Full and Archive nodes in that fleet
-
Scaling to the full ~600+ node network per TzKT: ~1.9 PB/year of bandwidth and ~70 TB/year of new permanent storage — with the non-baker tail (indexers, RPC providers, block explorers) skewing even more heavily toward Archive than the baker fleet does
Storage compounds. Every year without BLS aggregation adds another ~70 TB of permanent bloat that Archive nodes must keep forever: ~350 TB over five years, ~700 TB over ten. Indexers and explorer operators — who already shoulder the highest disk costs in the ecosystem — feel that compounding directly.
Why 50% is the threshold
The 50% threshold is a hard protocol-level switch, not a soft target. It’s encoded in the Tallinn upgrade (Tezos’s 20th protocol, activated on mainnet January 24, 2026) as a protocol parameter — all_bakers_attest_activation_threshold, set to 50% (Octez protocol docs).
The BLS efficiency story unfolds across two consecutive upgrades:
-
Seoul (19th, September 2025) introduced BLS attestation aggregation, replacing the ~200 individual attestation signatures per block with a single aggregated signature for any baker using a tz4 consensus key. Today, blocks carry both individual attestations from tz1/tz2/tz3 bakers and one aggregated attestation from tz4 bakers — so efficiency gains scale gradually as more bakers switch.
-
Tallinn (20th, January 2026) added the all-bakers-attest model — every baker attests every block, replacing today’s sampled subset — which strengthens consensus security, simplifies rewards calculation, and unlocks the full bandwidth and storage savings when paired with widespread tz4 adoption.
The Tallinn proposal gates the all-bakers-attest feature on the 50% threshold deliberately: bakers set the pace, the protocol waits, and once the threshold is crossed the feature stays on permanently even if participation later dips (Nomadic Labs announcement).
Until 50% is crossed, the all-bakers-attest model stays dormant and the bulk of the available efficiency gains remain locked.
Proposed bonus structure
I propose that the Tezos Foundation (and/or the Tezos Commons) fund a one-time bonus distributed to the first 115 bakers to activate BLS consensus keys, ranked by activation block height. Bakers ranked 116+ receive nothing from this bonus.
The reward uses three tiers matched to the actual risk profile of switching:
| Tier | Positions | Per-baker reward | Tier total |
|---|---|---|---|
| Pioneers | #1–#5 | $3,000 | $15,000 |
| Early adopters | #6–#30 | $1,500 | $37,500 |
| Threshold fillers | #31–#115 | $500 | $42,500 |
| Total pool | $95,000 |
The tiering reflects the asymmetric difficulty of going early. The first five bakers operate BLS in production with no established playbook — they’re the ones who debug BLS signer setups (RPi BLS Signer, Signatory, TezSign, or Russignol) and write the consensus-key migration playbooks for everyone else. By position #30 the operational patterns are documented in Discord and Agora threads. By #115 the migration is a recipe. Each tier roughly matches the real switching cost at that position: pioneer hardware + many hours of careful testing for the first five, routine migration for the tail.
Implementation details
-
Payment. Bonus denominated in USD and paid in XTZ to the baker’s delegate (manager) address at the activation-block exchange rate. This keeps payouts predictable for recipients while letting the Foundation budget cleanly, and sends to the persistent baker identity rather than the rotatable tz4 consensus key.
-
Tie-breaking. For bakers activating in the same block, tie-break deterministically by baker address (alphabetical), so the ranking is fully reproducible.
-
Verification. Fully verifiable on-chain — anyone can sort delegates by consensus-key activation height to reproduce the payout list.
Why the Tezos Foundation is the right funder
The Foundation’s stated mandate is to sustainably deploy resources to support the long-term success of the Tezos blockchain and community, with cost-effectiveness and return on investment as explicit metrics. This proposal sits squarely inside that mandate:
-
Baker support is an explicit Foundation funding category. The Foundation grants program lists Baking as a core area of interest, with the stated goal of empowering bakers through standardized tools and infrastructure. Driving tz4 migration is the same kind of operator-level transition the ecosystem already supports through tooling — RPi BLS Signer (Nomadic Labs), Signatory (ECAD Labs), TezSign (Tez Capital, community-funded via the Tezos Ecosystem DAO), and Russignol (independent — see Disclosure below). A one-time bonus to close the adoption gap is a natural extension of work the broader ecosystem already underwrites.
-
The ROI is exceptional even by Foundation standards. A $95,000 one-time outlay that unlocks $100,000+/year in perpetual network savings recoups within year one, and every subsequent year is pure return — effectively a 100%+ first-year ROI investment that never depreciates. By the Foundation’s own stated metric of cost-effectiveness, it’s hard to identify a higher-leverage use of treasury funds.
-
It protects existing Foundation R&D investment. The Foundation has funded core protocol development at Nomadic Labs, Trilitech, and Functori for years through its Core Engineering and Infrastructure vertical. That work delivered both the BLS aggregation in Seoul (September 2025) and the all-bakers-attest model in Tallinn (January 2026). Seoul’s aggregation is partially captured today through the 77 bakers who’ve switched, but Tallinn’s all-bakers-attest model — the larger of the two efficiency wins — remains locked behind the 50% threshold, and the adoption-trajectory data above shows it won’t activate without intervention. A modest one-time bonus to unlock that work is essentially insurance on a substantially larger prior R&D investment.
-
It strengthens decentralization. Lower operational costs reduce the barrier to running a baker — disproportionately benefiting independent and home operators, which is the population the Foundation has consistently worked to support through operator tooling and baking-infrastructure grants. A more sustainable cost structure for participation makes the validator set more diverse, more resilient, and more decentralized over time.
A note on safe adoption
This proposal rewards speed, but bakers shouldn’t trade speed for security. Test BLS key migration on Ushuaianet first, verify your signer and key-custody setup, and don’t activate on mainnet until you’re confident. A missed attestation cycle — or worse, a key compromise — will cost far more than the bonus pays.
Disclosure
I operate baker Baktz (tz1R4PuhxUxBBZhfLJDx2nNjbr7WorAPX1oC) and am the author of Russignol, an open-source BLS signing tool. Russignol was built and is maintained entirely on my own time, without funding from the Foundation, the Ecosystem DAO, or anyone else — I earn no income from it, and never have. It exists because closing the tz4 adoption gap matters to the network.
Baktz already migrated to tz4 on December 24, 2025, ranked #32 in the adoption timeline (data confirmable via the TzKT operations query in the Adoption trajectory section). Under the proposed tier structure, that places me in the threshold-filler tier — I would receive $500. Flagging this so my self-interest is visible up front.
There’s no direct financial interest at stake on the Russignol side, but ecosystem visibility and reputation are real, so I’m flagging it. Two commitments to keep the proposal clean:
-
Tool choice. Bakers have four BLS signing options — RPi BLS Signer, Signatory, TezSign, and Russignol. Pick what suits your setup, not the one I made.
-
Reward. Any payout I receive will be allocated entirely to Russignol development costs (hardware, public test infrastructure, ongoing maintenance), with a public accounting of the spend — consistent with how I’ve operated Russignol from the start. None of it becomes personal income.
The ask
Tezos Foundation / Tezos Commons Foundation: would you consider funding this bonus? Given that it fits squarely within the Foundation’s Baking funding category and its stated cost-effectiveness criteria, what’s the right next step — a formal application through the standard grants process, a community vote on Agora, or a direct proposal to your grants team?
Fellow bakers and community members: feedback welcome on the tier breakpoints, the pool size, and alternative structures. I’m especially interested in two questions:
-
Pool size. I’ve calibrated $95k against the conservative end of the savings estimate (recouped within year one). Should the pool be larger — say, matching a full year of network-wide savings at the upper bound ($140k) — to make the pioneer-tier reward materially larger? Or kept smaller for political acceptability?
-
Tier structure. The 5 / 30 / 115 breakpoints and the $3,000 / $1,500 / $500 amounts are my best estimate of where the real risk-and-effort lines fall. Are the breakpoints right, or should more weight shift to the pioneers (e.g. 3 / 20 / 115 with a $5,000 pioneer reward)?
Looking forward to the discussion.
