# Thinking out loud: Sia's second decade - demand, the agentic economy, and evolving how we fund the future By [Notes from the Second Curve](https://paragraph.com/@lastbubble2035) · 2026-07-07 --- **The short version, up front.** Storage demand is effectively infinite, and it's now compounding on two curves - everything humans create, and everything our machines create on our behalf. Sia has the cheapest storage per terabyte in the space, private by construction, and it should be the natural home for the coming wave of agentic-AI data. To capture that and stay healthy for the next decade, I think three things are worth doing - _in this order_: 1. **Grow demand.** Orient the ecosystem - SDKs, grants, narrative - around one identity: _Sia is the storage and memory layer of the agentic economy._ 2. **Add a second, usage-aligned funding source** by bringing the dormant Siafund block - a revenue share literally created to fund Sia's development - back into the Foundation's service. This _increases_ Foundation funding. Nothing is given up. 3. **Only then**, on a slow, checkpoint-gated, revenue-triggered glide path, retire the printed development subsidy that today dilutes every holder - with miner security left fully intact. The Foundation's funding should go _up_ before any part of it is ever wound down. None of step 1, and the first move in step 2, requires a hardfork, a budget cut, or a dollar of risk - the do-it-this-week starting points are at the very bottom. This isn't a price argument. It's about demand, alignment, and resilience. A word on where I'm coming from, because it shapes how to read this. First, I've built on Sia - a sovereign, agent-native personal data archive that I now use every day. It's a working v0: my own reading, captured with one click from my browser, content-addressed, encrypted onto Sia, indexed for meaning-based search entirely on my own machine, and queryable by AI assistants - several gigabytes of my real life on this network, growing daily (client-side encrypted under my own keys; the hosted indexer I use for convenience can't read a byte, and indexers are swappable by design). I built it because I think this protocol is perfectly positioned for what's coming, and I'd rather demonstrate that with something running than argue it in the abstract. Second, I'm about as aligned with this network as a person can be: I've held Siacoin and Siafund for nearly a decade and backed Nebulous itself, and I've held all of it through 99% drawdowns, with every intention of holding for many years more - not because I'm waiting for an exit, but because I believe in the thesis and in this team. So read what follows not as a holder talking his book (a decade of conviction through that kind of drawdown is the opposite of a trade), and not as a complaint about price, but as one long-term builder-owner thinking out loud about how the most quietly important protocol in this industry has its best decade yet. And the timing is right. The Storage app shipping isn't a one-off - it's the culmination of a team that has quietly, relentlessly shipped through the depths of a brutal market: the V2 hardfork and Utreexo, renterd, hostd, walletd, indexd, and now a real consumer app anyone can use. That was the hard part, and it's done. What's left is enormous but different in kind - turning a production-grade protocol into the obvious home for a wave of demand that's only just beginning to form. And that isn't something the Foundation can or should carry alone; it's exactly where the rest of us have to show up. The temptation in a down market is to go quiet. I'd argue the opposite: this is the moment for the ambitious conversation, while the cost of the status quo is most visible and the will to change is highest. No down-cycle should go to waste. One framing up front, because it shapes everything: **the prize is demand.** Sia's market value isn't low because the tokenomics are broken; it's low because the network's economic throughput is still small. The subsidy's dilution is real but modest - the case for evolving it is alignment, not emergency. No incentive redesign substitutes for real usage. So the bigger half of this post is about demand - and the tokenomics half is offered only in service of it. The opportunity: Sia as the storage layer of the agentic economy ---------------------------------------------------------------- Start with the demand side, because it's the whole game. The world's appetite for storage is effectively infinite, and it's now compounding on two curves at once: everything humans create, and everything our machines increasingly create and consume on our behalf. The question for the next decade isn't whether that demand exists - it's who captures it. Sia should be the natural, default choice for a meaningful share of it. We have the cheapest storage per terabyte in the space, private by construction - encrypted and sharded so no host can see a byte; what's left is to make us the obvious place that demand lands. The next decade of computing will be defined by autonomous software - agents that remember, retrieve, reason, and act on our behalf. Every one of those agents needs somewhere to keep its memory, its working state, its retrieved corpora, and an auditable record of what it did and why. Today that data sits in hyperscaler buckets owned by three companies. That is precisely the centralization Sia was built to undo - except now the stakes aren't just our photos, but the memory and provenance of the intelligence we're increasingly delegating our lives to. This is Sia's opening, and it's a big one: * **Agent memory and state.** Cheap, encrypted, content-addressed, user-owned storage for the persistent memory and context of autonomous agents. * **Retrieval corpora.** The document and vector stores behind RAG pipelines - exactly the large, dynamic datasets the rest of the world is now trying to move off expensive, egress-taxed object storage. * **Training-data provenance.** Content addressing gives a cryptographic, tamper-evident record of which data fed which model. Compliance and audit teams care deeply about this now. Arweave makes this pitch for permanence; Sia can make it for economical, mutable, sovereign storage. * **Sovereign personal archives.** Our own data - searchable, private, ours. (This is what I've built, and use daily.) And the timing is fortunate: S3 has become the lingua franca of modern data and AI tooling - the interface PyTorch, the inference frameworks, and every serious data platform already speak - and Sia now speaks it too. The Foundation's developer-first direction (a free SDK, frictionless onboarding, in-browser retrieval) is the right one, because the bottleneck to demand has never been the protocol - it's been the integration distance between Sia and the tools developers already use. Closing that distance is the single highest-leverage thing we can do. I'd push us to think bigger about where Sia plugs into the rest of the sovereign stack: * **Decentralized compute.** The agentic economy needs decentralized compute and decentralized storage to meet. Networks like Bittensor are building the former; Sia is the most mature, battle-tested answer for the latter. There's a natural complementarity here worth actively pursuing rather than waiting for. * **Private payment and identity rails.** Sia already encrypts and shards so no host can see your data. Pair that with shielded payments and modern account abstraction, and you can imagine a fully private storage-plus-payment stack - sovereignty end to end, not just at the storage layer. Privacy as infrastructure, composable across protocols. * **Complementary, not combative.** We don't need to beat Filecoin, Storj, or Arweave at their own games. Arweave is for permanence; Sia is for economical, deletable, private, sovereign storage at the lowest cost per terabyte. That's a different and arguably larger market, and it's ours to take. None of these are things any one person decrees - they're directions. But if we orient the ecosystem (grants, SDKs, partnerships, narrative) around "Sia is the storage and memory layer of the agentic economy," I think the demand follows. That is the flywheel that matters most. The structural friction: today's funding loop works against us -------------------------------------------------------------- Now the harder part, said as plainly and un-dramatically as I can. The Foundation is funded by a perpetual SC subsidy - 30,000 SC per block, matching the 30,000 that goes to miners, so half of all new issuance exists to fund development. The subsidy itself is fixed in SC; the protocol can't mint more when price falls. The pressure shows up one step downstream: salaries and grants are paid in dollars, so the Foundation converts treasury SC to fiat - and when price falls, meeting the same dollar budget consumes more coins, moving more supply to market in exactly the seasons when natural buyers are scarcest. To be fair and precise: the Foundation is not a distressed seller - it holds reserves, publishes its financials, and can time conversions sensibly, and at today's scale these flows are small against market volume. The deeper issue isn't a death spiral; it's structure and alignment: development is funded by printing and selling new supply regardless of whether the network grows, rather than by the network's own economic success. That's the loop worth fixing - not because it's an emergency, but because a better one exists. I want to be careful here: this was a reasonable design in 2021. It bootstrapped the Foundation through a genuinely hard transition after Nebulous, and it worked. The question isn't whether it was right then - it's whether it still serves us now, and whether there's a structurally better way to fund the future that turns this loop from vicious to virtuous. A better incentive: the Siafund flywheel ---------------------------------------- I think there is, and it's already built into Sia: Siafunds. Siafunds capture 3.9% of every completed storage contract — a revenue share that is _aligned with actual usage_ and _non-dilutive_, because it redistributes SC already in circulation from real economic activity rather than printing new supply. Imagine the Foundation funded primarily from that stream. Every integration they ship, every enterprise they onboard, every terabyte of agentic data they help land on the network flows back to them as development funding. The better they do their job, the more resources they have to do it. That's a virtuous flywheel - the exact inverse of today's loop, where weakness begets selling begets weakness. The honest caveat, and it matters - so let me put the numbers on the table before someone else does: **Siafund revenue cannot fund the Foundation today. It is not close.** Public trackers put network storage revenue in the low hundreds of thousands of dollars a year; 3.9% of that is on the order of $10-15K a year across all 10,000 Siafunds. The development subsidy, meanwhile, mints roughly 1.6B SC a year - about $1M even at today's depressed price — and the Foundation's actual budget draws on treasury besides. For Siafund revenue to replace the subsidy, network revenue would need to grow by roughly two orders of magnitude. I state that gap plainly because it is the entire reason for the sequencing above: this is not a proposal to swap funding sources. It's a proposal to grow demand ~100x - which is what the whole first half of this post is about - to add the Siafund stream as it becomes material, and only then, on triggers that don't fire unless the growth is real, to retire the printed subsidy. If the demand thesis fails, the taper never triggers and nothing changes. The plan is only as good as the growth - which is exactly how it should be. Picture a multi-year glide path, published and transparent: * The Foundation's SC subsidy tapers on a programmatic schedule - not to zero overnight, but stepping down predictably over, say, two-plus years. * As demand grows (driven by the very work above), Siafund revenue rises to fill the gap. * Make the trigger computable, not discretionary: Sia has no on-chain governance and shouldn't invent one for this. The honest design is a single, pre-committed schedule whose steps are conditioned on an on-chain-verifiable measure - trailing Siafund fee volume is protocol-level data - with conservative thresholds and a runway buffer. If revenue doesn't materialize, steps simply never trigger. Changing the schedule itself would mean another hardfork, and that friction is a feature: it keeps the commitment credible in both directions. * The miner reward (the other 30,000 SC/block) stays untouched, preserving the network's long-term security budget. Only the _printed development subsidy_ sunsets, replaced by _earned_ revenue. The exact schedule is for the Foundation and community to design together - I'm not attached to specific numbers. What I'm attached to is the _direction_: from printing-and-selling toward earning, on a path that aligns the Foundation's incentives perfectly with growing the economic value of the network. And to be clear about sequence, because it's the whole point: the taper is the _last_ step, not the first. Nothing comes down until a second, usage-aligned funding source is real and growing - which is exactly where the next section comes in. The Foundation should be _better_ funded through this, not asked to gamble its runway. And notice the second-order benefit: sunsetting the development subsidy roughly **halves** SC's new issuance. The funding fix and the supply-narrative fix turn out to be the same move - which brings me to two more things worth putting on the table: one a real opportunity, the other an unforced error we can fix today. The Siafunds were built for exactly this ---------------------------------------- Here's a fact worth sitting with. Siafunds exist for one reason: they were created to fund the development of Sia by capturing a share of the network's revenue. That is their entire design purpose - a purpose-built development-funding instrument. And yet the great majority of them - the last publicly cited figure is roughly 8,600 of 10,000, though that number traces to older Foundation documentation and, as far as I can tell, has never been publicly verified on-chain - appear to sit dormant in the wound-down Skynet Labs (formerly Nebulous) entity, throwing off a revenue stream that funds no development at all. Bringing that block back into the service of the network isn't really an "ask" - it reads to me as the logical next step, squarely in the interest of both the community and the Foundation. It is the single cleanest source of the sustainable, usage-aligned funding described above: a revenue share that was _designed_ to fund Sia, returned to the body that actually funds Sia, compounding exactly as intended as the network grows. This is the kind of foundational work that, done once and done right, strengthens Sia for the coming decades. I'll be direct about my own position, because it matters: I was an investor in that entity, so I have a foot in both camps - and I think that's precisely why this is win-win rather than zero-sum. But let me be equally direct about what the block is worth, because the arithmetic above applies here too: as an income stream today it throws off on the order of ten thousand dollars a year - as a cash-flow asset it is nearly worthless. Its real value is optionality on the demand thesis, and that optionality is worth more to the network (as permanent alignment) than to a wound-down estate (as a stranded line-item). Any path here should be structured accordingly: a donation, nominal consideration, or an earn-out tied to future network revenue. If a meaningful up-front price were demanded, the right answer is simply no - this must never cost the Foundation real treasury. For the estate, converting a stranded, illiquid position into purpose - with perhaps modest upside tied to Sia actually succeeding - is the best realistic outcome anyway; for the protocol, it's a permanent, mission-aligned funding engine that grows with adoption. The mechanics aren't the obstacle: Siafunds transfer like any wallet asset. The real questions are who controls the block today and whether there's the will to act - and the honest first step is simply to establish the facts. Those Siafunds sit on a public ledger, so confirming the holding address, its current balance, and its transaction history would settle - transparently - what is actually there and whether any of it has moved or been sold over the years. That feels like the reasonable place to start, well ahead of any talk of a transfer. And the natural party to open that conversation with Skynet Labs is the Foundation itself - it is the legitimate counterparty here, and the one without my conflict of interest. This should be Foundation-led, with any eventual valuation and structure left to independent parties so that no one is marking their own homework. It's an opportunity hiding in plain sight, and I'd love for us to explore it together. We're telling our own supply story badly ---------------------------------------- Separately, and regardless of any of the above: our supply narrative is costing us, and it's an unforced error. Newcomers see "unlimited supply" and walk away. But that label badly misrepresents reality. Sia issues a _fixed_ number of new coins per block - it has for years - against a supply base that keeps growing. The arithmetic consequence is that Sia's inflation _rate falls every single year._ It's in the mid-single digits today (counting both the mining reward and the development subsidy) and mechanically declining toward lower single digits over time; sunset the subsidy and it roughly halves again. "Unlimited" sounds like a faucet left running. The truth is a predictable, declining, disinflationary curve - the opposite of the runaway emission the word implies. This is a communications fix, not a protocol change, and we can do it today: a clear, simple, visual explainer of Sia's real issuance curve. I've taken a first cut at exactly that - see the chart below. It costs us nothing and removes a needless objection that's been quietly costing us holders and builders for years. ![](https://storage.googleapis.com/papyrus_images/cb801326a4255320cdff030d953e2376ead68002eb23552ba9504350db2af51e.png) Anticipating the hard questions ------------------------------- I'd rather raise the strongest objections myself than have them raised for me. Here are the ones I'd press if I were reading this, with my honest answers. **"Siafund revenue is still paid in SC - so the Foundation still sells SC. How is this any different?"** The difference is dilution - and let me size it honestly rather than dramatize it: the full subsidy is roughly 1.6B SC a year, low-single-digit percent of supply, about $1M at today's price. Not an emergency; a structural misalignment. Subsidy SC is _newly minted_\- it expands the supply every block, a quiet tax on every holder. Siafund SC is _already-circulating_ coin, earned from renters who bought it to pay for real storage. One prints new supply; the other recycles supply that demand already pulled into the network. The Foundation selling earned revenue is just an economy functioning. The Foundation selling freshly-printed coins is dilution. **"Won't tapering the subsidy starve development before Siafund revenue can replace it?"** Only if we do it carelessly - which is why it's a checkpoint-gated transition, not a switch. The taper should be tied to trailing Siafund revenue and a runway buffer: if demand (and therefore revenue) doesn't grow, the taper slows or pauses. The goal is to _de-risk_development funding by making it grow with the network, never to gamble it against a deadline. **"This needs a hardfork. That's a major lift and a real risk."** It is - and that's a feature, not a bug. A change this consequential _should_ require broad, deliberate consensus and careful engineering; it can't and shouldn't happen because one forum post asked for it. I'm not proposing we fork tomorrow. I'm proposing we begin the multi-year conversation now, while the post-V2 window is open and the Foundation has just shown - with V2 itself - that it can ship complex forks cleanly. **"We've been through painful treasury changes before. Why reopen this?"** Because the lesson from last time was about _process_, and this is built around that lesson: transparent, gradual, community-designed, and reversible at every checkpoint - the opposite of a unilateral move. If the community doesn't want it, it doesn't happen. I'm offering a direction to debate, not a decision to ratify. **"Does this hurt miners?"** No. The 30,000 SC/block miner reward is untouched in every scenario. Mining economics don't change, and the network's long-term security budget stays exactly as designed. Only the Foundation's _separate_subsidy portion is on the table. **"If supply is the worry, why not just propose a hard cap like other projects?"** Because a cap would trade away Sia's long-term security to win an optics argument. Perpetual issuance is what funds perpetual mining security - it's the deliberate answer to the problem Bitcoin is slowly walking into as its block reward vanishes. The honest fix isn't to cap the supply; it's to stop _printing the development subsidy_ and to tell the disinflation story properly. You'll notice I'm explicitly _not_ asking for a cap. **"Aren't you just trying to shrink supply to pump the price?"** Lower issuance is a _side effect_ of fixing the funding model, not the objective - and I've tried hard to say so plainly. I'm not proposing buybacks, burns, or anything touching the miner reward. If this were a price play, removing pro-cyclical sell pressure over a multi-year glide path would be a strange and slow way to run it. It's about resilience, which helps the network at any price. **"On the Skynet Labs Siafunds - you were an investor. Isn't this self-serving?"** I disclosed exactly that so you can weigh it - and let me complete the disclosure: I also hold a small number of Siafunds directly, and have for nearly a decade. A post that elevates the Siafund story, and any Foundation transaction involving the Skynet block, would create the first public reference point for Siafund value in years - which would touch my own units too. I can't remove that overlap; I can only surface it before someone else does, and point at the structural answer: it's exactly why I've argued the terms should be nominal or earn-out-based, why valuation and structure should sit with independent parties, and why the Foundation should lead. Beyond that: I'm not claiming ownership of anything or suggesting expropriation - those Siafunds are someone's legitimate property, and any resolution would have to be voluntary and legal. What I'm asking for _first_ is simply clarity: a revenue stream this central to Sia's design currently flows to a wound-down entity, and the community deserves to understand its status. If there turns out to be a legitimate, willing path to align it with the network's future, that's upside for everyone - down to the smallest holder. **"Can Sia actually serve AI workloads? It's slower than centralized storage."** Honestly, latency is a real limitation, and I won't pretend otherwise. But a large share of agentic data isn't latency-critical: archival agent memory, training and retrieval corpora, provenance records, warm context, backups. Sia competes where cost, durability, privacy, and sovereignty matter more than single-digit-millisecond reads - and v2, parallel retrieval, and in-browser access keep narrowing the gap. We should target the workloads we win, not the ones we don't. **"Why would anyone pick Sia over Filecoin, Storj, or Arweave?"** We don't need to win the whole market. Sia is the cheapest per terabyte, genuinely private (encrypted and sharded, so no host sees your data), deletable (GDPR-workable, unlike permanent-only storage), and it needs no token lockup to use - you just pay. Against Arweave we're complementary, not competitive: Arweave is for permanence, Sia for economical, mutable, sovereign storage. The agentic-memory niche plays directly to those strengths, and nobody owns it yet. **"'Agentic economy' sounds like a buzzword. What concretely gets built?"** Fair. Concretely: a durable, content-addressed memory-and-state SDK for autonomous agents; native read/write integrations so agents can use Sia directly; S3-drop-in for RAG corpora; and provenance tooling for training data. I've built a working implementation myself - it runs today, capturing and retrieving my own data on Sia daily, with AI assistants answering questions from it - precisely so this is a working example others can fork, not a slide. **"Why now, in the middle of a downturn?"** Because that's when it's honest and possible. A drawdown makes the structural cost of the status quo visible and the appetite for change genuine; a bull market breeds complacency and would make this look like greed. With V2 shipped and real products launching, this is the natural moment to design the growth phase - not the moment to go quiet. What I'm committing to ---------------------- I don't want to be someone who shows up with opinions and leaves. So: I've built on Sia - a sovereign, agent-native personal data archive, running and in daily use - and I'm developing it into a reference implementation for _durable memory for AI agents on Sia_, the kind of thing other builders can fork. I'll be writing more about that build publicly soon. I'm glad to contribute the issuance explainer. And I'm happy to help think through the funding-transition framework in whatever venue is most useful. I'd rather earn a hearing through contribution than ask for one. To be unambiguous about intent ------------------------------ None of this is about price. It's about demand, alignment, and resilience. A network whose development funding grows as it succeeds, whose incentives point everyone the same direction, and whose supply story is told honestly is simply a stronger network - at any price. What I'm actually asking ------------------------ Starting with things that cost nothing and require no hardfork, no budget cut, and no legal risk: **Now, at zero cost:** 1. Could the Foundation confirm the on-chain status of the dormant Siafund block - the holding address, its current balance, and whether it has moved or been sold over the years? Pure transparency, and it settles what we're even working with. 2. Could we co-produce and promote a clear, honest explainer of Sia's disinflationary issuance? I've drafted a first version (the chart above) and will build it out properly. **Within the roadmap you already have:** 3. Can we adopt "Sia is the storage and memory layer of the agentic economy" as an explicit organizing thesis for SDK and grants priorities? My own build is already behind it - running, storing my real data on Sia every day - as an early reference implementation. **For deliberate discussion - ideally taken up on a community call or governance thread in the next month or two:**4\. Is there appetite to (a) explore bringing the Siafund block into the Foundation's service as _additive_ funding, and - separately, and only if that proves out - (b) a transparent, revenue-gated path to retire the printed subsidy that currently dilutes every holder? I'm not asking anyone to ratify anything. I'm asking us to take the two free steps now, and to put the bigger questions on a real calendar - so this becomes a process, not a thread that scrolls away. Ten years ago this network was an idea between two college friends. We survived the hard years - the funding collapse, the long drawdown, the rebuilds - and came out the other side with V2 shipped and real products in users' hands. The foundation is laid. And the largest wave of data demand in history is forming right now, as humans and machines alike reach for storage that's private, sovereign, and theirs. We are built to catch it. Let's align our incentives, tell our story honestly, and go meet it. The next decade starts now. \- _lastbubble2035_ --- *Originally published on [Notes from the Second Curve](https://paragraph.com/@lastbubble2035/sias-second-decade)*