papers/morpheus/loi/lux-morpheus-loi.tex

\documentclass[11pt,a4paper]{article}
\usepackage{../shared/luxcover}
\usepackage{../shared/proposal}
\usepackage[utf8]{inputenc}\usepackage[T1]{fontenc}
\usepackage{amsmath,amssymb}
\usepackage{listings}\usepackage{xcolor}\input{../shared/lstlang}

\newcommand{\code}[1]{\texttt{\small #1}}
\newcommand{\lab}[1]{{\color{muted}\footnotesize\,[#1]}}

\title{Letter of Intent}
\author{Zach Kelling\\\textit{Lux Industries, Inc.}\\\texttt{[email protected]}}
\date{2026}

\begin{document}
\luxcoverpage

\noindent
\begin{tabular}{@{}p{0.12\linewidth}p{0.84\linewidth}@{}}
\textbf{To} & The Morpheus community and its core contributors, through the MRC
process and the public governance forum.\\[0.4ex]
\textbf{From} & Zach Kelling, Lux Industries, Inc., with Hanzo AI, Inc. and Zoo
Labs Foundation.\\[0.4ex]
\textbf{Date} & 28 August 2026. Lapses 27 October 2026 unless extended in writing.\\[0.4ex]
\textbf{Status} & Non-binding in its entirety. Not an offer to sell a security.
Every item is a proposal your governance may accept, amend, or reject.
\end{tabular}

\vspace{1.0ex}

\noindent MOR trades at \$1.58. The protocol earns about \$476{,}000 a year and
the number is falling. We read all sixty-two repositories, at full history, from
the deployed addresses and the source. This letter is a menu. Take as much or as
little as you want. The parts do not depend on each other, and the schedule at
the end states the unflattering findings first, because you will find them in our
commit history anyway.

\section*{What is true today}

We work from your numbers.

\begin{itemize}
\item \textbf{Market.} MOR \$1.58, market capitalisation about \$14M. DEX depth
about \$1.3M, most of it protocol-owned in one Safe. Trailing 24-hour
volume about \$57{,}000.\lab{verified}
\item \textbf{Revenue.} Real revenue is Lido and Aave yield, about \$476{,}000 a
year, down from \$3.24M in the launch quarter of 2024. Seventy-five percent funds
a buyback of about \$511{,}000 a year. Emission is about 12{,}000 MOR per day, tapering
from 8 February 2024 to zero in 2040, 42M terminal, no premine. Emission outruns
real revenue about ten to one.\lab{verified}
\item \textbf{Capital.} The pool holds about \$20M of third-party stETH.
Principal is withdrawable one to one. This is not a ponzi. Two problems are real:
dilution outruns revenue, and settlement is fakeable.\lab{verified}
\item \textbf{Settlement.} Compensation is \code{pricePerSecond} times elapsed
seconds on a provider-signed receipt. The receipt signs \code{inputTokens} and
\code{outputTokens} (\code{abi.go:18--19}). The contract decodes five of the
seven fields and drops both token counts (\code{SessionRouter.sol:237}). Model
identity is never committed. A provider can serve an 8B model, sign a 70B price,
and settle clean.\lab{verified}
\item \textbf{Controls.} No slashing (the grep is empty). No timelock. Unbounded
mint (\code{MOROFT.\_mint}, no cap). Upgrade authority over the capital contracts
is an anonymous 5-of-9 Safe.\lab{verified}
\item \textbf{Demand.} Eighteen providers on \code{active.mor.org}, largely
reselling Venice and OpenAI subscriptions. No usage, session, or revenue figure
is published anywhere in the corpus.\lab{verified}
\end{itemize}

Credit where it is earned. Treasury bifurcation capped the July 2026 stipend leak
near one percent of the pool per day. Session accounting decodes the on-chain
\code{Transfer} log rather than trusting a spend estimate. Emission splits pro
rata by yield actually contributed, not by TVL. Roughly five engineers carry all
of this. The honest problem is not fraud. It is that emission buys nothing and
settlement proves nothing.

\section*{The menu}

Four tiers. Each is complete on its own. Adopt one and stop, or take them in
order. None requires the next.

\subsection*{Tier 0: hygiene. No vote, no us, no cost.}

Three live credentials sit in source or history. Two repositories have no CI.
One wallet function mints funds to a throwaway address. Rotate the keys, make a
failing test fail the build, and fix \code{deriveAddressFromPrivateKey}. The
fund-losing path closes today. Cost is zero and none of it depends on this
letter.\lab{verified}

\subsection*{Tier 1: verifiable settlement and real demand.}

The problem is fakeable settlement and published demand near zero. The fix is
built. \code{mord} makes settlement un-fakeable: faked work earns nothing, model
substitution is slashed. It compiles and its forty tests pass as of 29 August
2026.\lab{verified} \code{lux/ai} supplies attestation, a verification quorum, and
\code{SlashProvider}, verifying locally against nvtrust, SPDM, and RIM with no
cloud dependency.\lab{verified} \code{api.hanzo.ai} joins the marketplace as an
attested provider under the same terms as any other, and Enso, a learned router,
scores model selection against live provider bids rather than benchmarks.\lab{verified}
Behind it stand consumer surfaces with 100{,}000{,}000+ launch reach:
\code{hanzo.chat}, \code{hanzo.app}, \code{hanzo/dev}, \code{hanzo/desktop},
\code{zoo/app}, \code{lux.exchange}, \code{lux/wallet}.\lab{client} The fee is
charged on verified volume and feeds your existing buyback. The quantified outcome
is below.

\subsection*{Tier 2: governance and a use for the token.}

The problem is no timelock, nine anonymous keys, and no productive use for MOR
beyond holding it. \code{lux/dao} attaches a freeze guard and a timelock to the
existing Safe first: the keyholders keep their keys and lose only the ability to
act without notice. A governor follows, executing through the Safe rather than
around it.\lab{verified} Then the token earns a use: \code{lux/dex} runs the LX
order book, \code{lux/liquid} offers self-repaying lending against MOR (VaultV2,
a dozen adapters), and \code{lux/bank} adds accounts and cards.\lab{verified} MOR
gains a timelock and a reason to move, and the next stipend bug is capped by
policy rather than by luck.

\subsection*{Tier 3: a sovereign chain, and the raise.}

The problem is that MOR secures nothing and emission buys nothing.
\code{l2-template} stands up an L2 in minutes; Hanzo and Zoo were both made this
way, with a post-quantum root anchored in Quasar Horizon finality.\lab{verified}
A sovereign Lux L1 follows, on which MOR is gas and stake. The staking argument
is the next section. The indicative raise sits here, and it is \lab{pending}.

\section*{L1 versus L2, and what MOR secures}

Morpheus settles on Base, an L2. MOR secures nothing there. Two paths follow.

An L2 ships now. \code{l2-template} produces a chain in minutes and Morpheus
keeps Base-grade settlement. But MOR stays non-productive: security comes from
the stack, not from the token, so emission continues only to dilute.

A sovereign L1 makes MOR the gas and the staking asset. Validators post MOR;
emission pays for validator security and finality instead of pure dilution. The
same roughly 12{,}000 MOR per day that today buys nothing then buys a security
budget.\lab{derived}

The cost is honest. An L1 needs a validator set that exists and is decentralised,
and that does not exist on day one. So the path is the L2 first, the sovereign L1
when the validator set is ready. Bid discovery migrates before settlement, or not
at all. One more thing stated plainly: \code{mlkem} and \code{fhe} are activated;
\code{mldsa} and \code{slhdsa} are present but not. Post-quantum transport is
live. Post-quantum signatures are staged, not live.

\section*{The outcome, quantified}

The funnel comes from our joint go-to-market model. Every post-reach number is a
modelled assumption, not a Morpheus measurement.\lab{illustrative}

100M reach, about 2M monthly active, about 40{,}000 opting into a private or
decentralised inference tier, about 20{,}000 routed, and 1{,}000 to 10{,}000
verified sessions per day near-term, bounded by the eighteen providers on hand.
Two assumptions carry it: the funnel ratios hold within an order of magnitude,
and provider count grows past eighteen to lift the near-term ceiling.

Take the low case. Ten thousand verified sessions per day is about 3.65M verified
sessions per year. It already exceeds all currently published Morpheus demand,
which is about zero.\lab{verified for the ${\sim}0$; illustrative for the 10k}
The fee rides on verified volume and feeds the buyback:

\begin{itemize}
\item At \$0.10 per verified session, the buyback gains about \$365{,}000 a year
from usage.\lab{illustrative}
\item At \$0.50 per verified session, about \$1{,}800{,}000 a year.\lab{illustrative}
\end{itemize}

Today's buyback is about \$511{,}000 a year, funded by deposit yield that is
falling.\lab{verified} Both figures above come from usage, not deposits, and both
grow with sessions and providers rather than shrinking with yield. That is the
whole point: revenue that scales with use instead of with TVL. What we do not
know: no session or revenue figure is published in the corpus, so we cannot
calibrate the funnel from your data. The ratios are our assumption and we will
report them against measured traffic once Hanzo takes live volume.

\section*{Why us}

Lux, Hanzo, and Zoo are the one counterparty that already ships both halves this
fix needs: the attested backend and un-fakeable settlement (\code{mord},
\code{lux/ai}), and the 100M-reach consumer surfaces with a KYC base onboarded
through Hanzo IAM and \code{lux/wallet}. Verifiability without demand is a lab.
Demand without verifiability is what Morpheus has today.

\section*{The indicative raise}

Indicative, not an offer.\lab{pending} A raise clears only into demonstrated
on-chain revenue and a KYC base already onboarded through Hanzo IAM and
\code{lux/wallet}. Amount, instrument, jurisdictions, and participant eligibility
are for securities counsel in each relevant jurisdiction; none is set here.
\code{lux/exchange} can run the offering, its securities-exchange and ICO function
itself pending SEC approval of a proposed regulation.\lab{pending} Proceeds go to
protocol-owned liquidity and a DAO-held treasury under \code{lux/treasury} with
published balances, not to a multisig or an operating company. The sequence is
fixed: demand and verifiability first, token rails second, the L2 and the raise
third. We would not want a mandate obtained any other way, because a treasury the
community did not vote for is just another multisig.

\section*{Schedule A: what we found}

Every item has a location so it can be checked or disputed. We withdraw any item
shown to be wrong. None is an argument for a lower price; no price is attached to
this letter.

\noindent\textbf{Credentials.} An Infura key in \code{dashboard-v2
config/index.tsx:26}, shipped to browsers. A live \code{SENTRY\_AUTH\_TOKEN} in
\code{dashboard-v2 .env.sentry-build-plugin}. A Helius mainnet key in
\code{MORSOL Anchor.toml} at \code{ca01d42}. Rotate before answering this letter.

\noindent\textbf{Tests that cannot fail a deploy.} \code{Morpheus-Marketplace-API
build.yml:212} runs pytest, warns, and continues, so 236 tests cannot block a
release. \code{Morpheus-Lumerin-Node} invokes \code{go test} zero times and
\code{go test ./...} fails on main. \code{dashboard-v2} and
\code{Morpheus-Marketplace-APP} have no \code{.github} directory.

\noindent\textbf{Fund loss.} \code{morpheus-deploy wallet.ts:165}
\code{deriveAddressFromPrivateKey} ignores its argument and returns
\code{randomBytes(20)}; \code{morpheus init} prints \code{Smart Wallet created}
and the README says to fund it. Fix today.

\noindent\textbf{Settlement and trust.} Settlement commits nothing about the
weights, quantization, tokenizer, or backend that served a request, so
substitution is undetectable from settlement data. \code{MySuperAgent} trusts a
client-supplied \code{x-wallet-address} header across 27 files, and the hosted
gateway multiplexes all Web2 callers onto one shared consumer wallet.

\noindent\textbf{Build state.} The subgraph HEAD cannot compile.
\code{SmartContracts} main has been frozen since 29 September 2025.
\code{BuildersV4.sol} lives only on a feature branch while the docs scope its
deployed address for a 100k bounty. Six completed efforts sit unmerged; finishing
them is a faster route to visible progress than anything we propose.

\section*{Legal}

\noindent\textbf{Non-binding.} This letter expresses intent only. It creates no
obligation to negotiate, to continue, or to transact. No binding obligation
arises unless and until definitive written agreements are executed by all
parties. Only the confidentiality paragraph is intended to be binding.

\noindent\textbf{No offer of securities.} Nothing here is an offer to sell, or a
solicitation to buy, any security, token, or instrument in any jurisdiction. Any
raise is indicative and conditional on the opinion of securities counsel as to
structure, instrument, exemption, eligibility, and transfer restrictions. No
terms are set and no participant has been solicited.

\noindent\textbf{No exclusivity, no control.} We request no standstill, no right
of first refusal, and no matching right. Nothing here transfers ownership of the
protocol, its marks, its contracts, or its treasury, and no signer is added to or
removed from any multisig by this letter.

\noindent\textbf{Confidentiality, costs, law.} Each party keeps the non-public
terms confidential and uses the information only to evaluate these proposals,
excepting what is public, independently developed, or required by law. Morpheus
governance may publish this letter in full, including Schedule A, and we would
prefer that it does. Each party bears its own costs. Any definitive agreement
specifies governing law; this letter does not. This letter lapses 27 October 2026
unless extended in writing.

\vspace{1.6ex}
\noindent\rule{0.42\linewidth}{0.5pt}\par
\vspace{0.3ex}
\noindent Zach Kelling\\
Lux Industries, Inc.\\
For itself, with Hanzo AI, Inc. and Zoo Labs Foundation\\
\texttt{[email protected]} \quad 28 August 2026

\vspace{1.0ex}
\noindent{\color{muted}\footnotesize Referenced repositories are public or
available on request. Every claim in Schedule A is sourced to a file, a line, or
a commit. Provenance labels: \emph{verified} on-chain, in-repository, or from
DefiLlama; \emph{illustrative} modelled with a stated assumption; \emph{pending}
external approval; \emph{client} our own reach.}

\end{document}