Snowflake
S3 as primary storage, compute billed apart.
Pre-seed round, open
We are raising a pre-seed
for NamiDB.
A graph database that lives in your S3 bucket: graph traversal, vector search, and full-text search in one engine, on storage you already pay for. The engine ships today. The round pays to put the Cloud in the market.
- Stage
- Pre-seed
- Target
- $1,000,000
- Instrument
- Post-money SAFE
- Committed
- None yet
- Runway
- 18 months
- Entity
- NamiDB, Inc. (Delaware)
A Rule 506(c) offering, open to accredited investors only.
This offering is available only to accredited investors, as defined in Rule 501(a) of Regulation D. NamiDB, Inc. is conducting it under Rule 506(c) of Regulation D under the Securities Act of 1933. Accredited investor status is verified before any agreement is signed.
$pip install namidb
Why now
Object storage became a database primitive in 2024.
Before 2024, an atomic commit on S3 needed an outside coordinator: DynamoDB, etcd, ZooKeeper. GCS and Azure had preconditions earlier, but S3 is where the volume is, and S3 closed the gap in 2024. Now the write itself is the coordinator.
- August 2024
S3 ships conditional writes on If-None-Match, so a writer can create an object only if nobody else did.
- November 2024
S3 adds If-Match on the ETag. AWS describes it as offloading compare-and-swap to S3 itself.
Every other category already moved
WarpStream
Kafka straight to S3. Confluent acquired it in September 2024.
Neon
Pages in object storage. Databricks acquired it in May 2025.
turbopuffer
Their published infrastructure cost is $70 per TB per month on S3 plus an SSD cache, against $3,600 on RAM plus 3x SSD, at their own stated cache and utilisation assumptions.
Warehouses, streaming, OLTP pages, and vector search all moved to object storage. Graph is the category that has not, and it is still billed by the gigabyte of RAM.
The problem
Priced by RAM, right as the workload became a million idle graphs.
There is nothing cheap to start with.
Neo4j AuraDB Professional lists at $65 per gigabyte of memory per month with a one gigabyte minimum. Business Critical lists at $146. Below Professional there is only a free tier. Amazon Neptune starts around $68 a month on the smallest burstable instance, or roughly $80 on serverless at its one NCU floor, plus $0.10 per GB of storage and $0.20 per million I/Os. Memgraph prices on total memory and does not publish a number.
The workload inverted underneath the pricing.
Gartner projects the average global Fortune 500 enterprise goes from fewer than 15 AI agents in use in 2025 to over 150,000 by 2028. Our own assumption, not Gartner's, is that most of those agents are idle most of the time. A price list built around reserved memory bills all of them at full rate anyway.
Regulated buyers are locked out.
Sovereign and defence workloads cannot ship their graph to somebody else's cloud. The open embedded option lost its maintainer when Kuzu was archived on 10 October 2025 and the team went to Apple. The MIT code lives on in community forks. What left the market was a funded team shipping it.
$66,560
What a terabyte of graph would cost in Neo4j AuraDB Professional memory at the published $65 per GB per month. Neo4j does not sell a Professional instance that large. The ceiling is 128 GB, and a terabyte is a Virtual Dedicated Cloud quote with no published rate at all.
against
$23.55
The same tebibyte in Amazon S3 Standard at $0.023 per GB per month, before request charges.
The solution
We made the bucket the database.
Priced like storage
You pay for bytes at rest, not for reserved memory. A sleeping namespace costs what its bytes cost and nothing more, because compute goes to zero.
Runs anywhere
The same engine in your cloud, in your data centre, or air-gapped with no outbound connection. One Rust binary, no coordinator, no cluster.
Nothing to leave behind
Source-available under BSL 1.1, Parquet-derived data files, your storage account. Walking away costs you a bucket copy.
What ships today
One engine, three deployments, one query across all three.
Embedded
Shippedpip install namidb, or the Rust crate. Point it at storage and run Cypher inside your process. This is the free front door.
Server
ShippedOne binary that speaks HTTP and Bolt 4.4 to 5.4, so existing Neo4j drivers connect without a new client. The Cypher surface is a subset, not Neo4j parity. Runs air-gapped.
Cloud
Closed betaA namespace per tenant or per agent, scaling to zero when nobody queries. This is where the revenue is, and this is what the round pays to launch.
In the engine today
- Cypher
- DiskANN vector index
- BM25 full-text
- Hybrid search with reciprocal rank fusion
- 10 graph algorithms
- Bolt 4.4 to 5.4
- Read-only MCP server
- Python and Rust
- S3, R2, GCS, Azure Blob, MinIO
What we have actually measured
LDBC SNB Interactive at scale 1.0, which is 540,000 nodes from the deterministic datagen, on four IC queries (IC02, IC07, IC08, IC09) at warm p50, where NamiDB runs roughly 3 to 5 times faster than Kuzu on the warm path. Plus a server-side warm p50 comparison against Neo4j Aura, a bulk write throughput run to Cloudflare R2 in the EU, and a raw R2 PUT and GET microbenchmark.
What we have not measured: LDBC SF1, SF10, the other ten IC queries, and the short reads. Neo4j running locally on attached NVMe still beats us, and cold start is the tradeoff object storage hands you. We are not trying to win a single machine benchmark. We are changing what a graph database costs to own.
Competition
Everyone bills you for RAM. We bill you for bytes at rest.
| Neo4j / AuraDB | Amazon Neptune | Memgraph | NamiDB | |
|---|---|---|---|---|
| Where the data lives | Block storage with a RAM page cache | AWS-managed distributed volume | In memory by default, slower on-disk mode since 2.11 | Storage you already own |
| What you are billed for | GB of memory per month, $65 to $146 | Instance hours, NCUs, GB and I/O | Total memory, by their own wording | Bytes at rest and queries served |
| Cheapest paid plan | $65 per month, 1 GB minimum | About $68 per month on db.t4g.medium, about $254 on the smallest memory-optimised instance | Not published, contact sales | Closed beta, published when the Cloud opens |
| Cost of an idle database | Full price of the reserved memory | Full price of the instance you left on | Full price of the reserved memory | Storage only, compute goes to zero |
| Runs air-gapped on your hardware | Enterprise licence, priced per deployment | No | Enterprise licence, priced on memory | Yes, same engine, no connection out |
| Licence | Open core, GPLv3 or commercial | Proprietary, AWS only | BSL 1.1 plus enterprise licence | BSL 1.1, converts to Apache 2.0 |
Competitor figures are published list prices, read in August 2026. The NamiDB column describes the Cloud as designed and in closed beta. The engine columns are shipped and you can run them today.
The market
A $3.6B market compounding in the mid twenties.
TAM
$3.6B
Graph database software in 2026 per Precedence Research, compounding at 24.15 percent to $25.2B by 2035. Mordor Intelligence puts 2026 at $4.21B and 2031 at $14.0B at 27.19 percent.
SAM
$2.0B
The cloud share of that market in North America and Europe. Mordor puts cloud at 71.5 percent of category spend and growing fastest.
SOM
$40M to $100M
The reachable share on a five year horizon, where three bottom-up paths converge: a top-down share of the 2031 cloud segment, 750 enterprise accounts at $50k, and a developer funnel at team-workspace prices.
Sources: Precedence Research and Mordor Intelligence, 2025 to 2026. MarketsandMarkets reports a smaller figure on a narrower definition. Agent projection: Gartner, April 2026. The full bottom-up arithmetic is in the appendix of the deck.
Go to market
Developers adopt it first. Sovereign buyers cannot get it anywhere else.
Developers, before anyone signs anything.
pip install, then embed it in production against your own bucket, then the Cloud or on-premise when it needs to be somebody else's problem, then an enterprise contract. This is the audience we build for: engineers wiring memory into agents, people running automations, anyone whose data is really a set of relationships. Graphs, vectors, and full-text search in one query is the whole reason to switch.
Sovereign and public sector, through Radical.
The engine runs air-gapped on the buyer's own hardware, with no external dependency and no vendor in the path. That is the one thing a managed graph cloud cannot offer. GRD Radical Corporation already sells into those buyers and carries NamiDB in as part of its stack. Government trust takes decades to build. We did not build it, we partnered with the people who did.
Traction
The people who build infrastructure found it first.
114 stars on GitHub.
The profiles behind those stars include engineers who list Cloudflare, Microsoft, Intel, Alibaba, Tencent, Red Hat, Dropbox, AT&T, Fly.io, and Weaviate. These are self-reported affiliations on GitHub, not enterprise accounts, and we would rather say that than let you assume otherwise.
GRD Radical Corporation.
25 years in LATAM cybersecurity, its own 24/7 SOC and a CERT that publishes an RFC 2350 profile, ISO 27001 certified, and a standing public-sector practice. Country and organisation counts are Radical's own figures, and the contracts are available in diligence. This is the channel NamiDB sells sovereign through, and its CEO is a founder here.
Universidad Regional Autonoma de los Andes.
Signed with the university as an institution rather than as a pilot with one lab: academic research, products and startups built on NamiDB, and the people who carry the technology to market with us. A copy of the agreement is in the data room.
The team
Three founders: the engine, the market, the paperwork.
Matias Fonseca
CEO
Founder of Fonles Studios (Ecuador, Peru, El Salvador) and maker of Alheli, a legal AI that answers over case files of up to 60,000 PDF pages, with more than 500 paying clients.
- 500+ paying clients
- 60,000-page case files
Claudio Cortes
Founder, go to market
CEO of GRD Radical Corporation. 25 years in LATAM cybersecurity with its own 24/7 SOC, a CERT that publishes an RFC 2350 profile, ISO 27001, and a standing public-sector practice. Previously 3Com, 2001 to 2009.
- 25 years
- 8 countries
- 3,000+ organisations
Dario Fonseca
Founder, legal
20 years in law and public administration, having led teams of over 100 people inside government. Co-founder of Fonles Studios. Owns NamiDB's corporate structure, licensing, and multi-jurisdiction operations.
- 20 years
- 100+ people led
- Multi-jurisdiction
The ask
$1,000,000 to put the Cloud in the market.
Post-money SAFE on standard terms, 18 months of runway. Nothing is committed yet, so this is first money in. We would rather say that plainly than imply a lead we do not have.
Use of funds
Marketing and distribution
50%The engine is built and almost nobody knows it exists. Brand, developer community, the public-sector channel, and a commercial lead.
Product and engineering
35%Cloud to public launch, usage-based billing, the sovereign on-premise distribution, and the first engineering hires.
Legal and operations
15%Compliance, agreements, public-sector certifications, and intellectual property.
What that buys, in 18 months
The Cloud goes public
Out of closed beta and onto a public price list, with usage-based billing behind it.
The first sovereign deployment
An air-gapped install running inside a public-sector buyer, with the certification pack to repeat it.
Paying tenants and a reference account
Shipped integrations underneath the agent-memory frameworks, so choosing NamiDB is a one line change, and the first Cloud tenants paying for it.
Get in touch
Four fields, then a real
conversation.
Send this and we will come back with the SAFE terms, the deck, and a time to talk. If you would rather just write, the address is matias.fonseca@namidb.com.
Important legal information
No offer where unlawful. This page is a general solicitation made in reliance on Rule 506(c) of Regulation D under the Securities Act of 1933. It is not an offer to sell, or a solicitation of an offer to buy, any security in any jurisdiction in which such an offer or solicitation would be unlawful. Any offer or sale of securities will be made only to verified accredited investors, and only pursuant to definitive offering documents, including a subscription agreement and the applicable Simple Agreement for Future Equity. In the event of any conflict, those documents control, and nothing on this page qualifies, supplements, or amends them.
Accredited investors only. Securities in this offering may be sold only to persons who are accredited investors as defined in Rule 501(a) of Regulation D. NamiDB, Inc. is required to take reasonable steps to verify each investor's accredited investor status. No Simple Agreement for Future Equity will be signed or countersigned, and no investment will be accepted, until that verification is complete. Submitting information through this page does not create any obligation on the part of NamiDB, Inc. to accept an investment, and NamiDB, Inc. may decline any prospective investor for any reason.
Unregistered, restricted, and illiquid securities. The securities offered have not been registered under the Securities Act of 1933 or under any state securities laws, and are being offered in reliance on an exemption from registration. They are restricted securities and may not be resold except pursuant to registration or an available exemption. There is no public market for these securities, none is expected to develop, and investors should be prepared to hold them indefinitely.
No guarantee, and a risk of total loss. An investment in an early-stage company is speculative and involves a high degree of risk, including the risk of losing the entire amount invested. No return, distribution, liquidity event, or future financing is promised, projected, or guaranteed. Past performance and current traction are not indicative of future results.
No government review or endorsement. The Securities and Exchange Commission, any state securities regulator, and any other governmental authority have not reviewed, approved, endorsed, or passed upon the merits or accuracy of this offering or of any information on this page. Any representation to the contrary is a criminal offense.
Forward-looking statements. Statements on this page that are not statements of historical fact are forward-looking statements, including statements about product roadmap, performance, market opportunity, growth, and future financings. They are based on assumptions and expectations as of the date made, are subject to significant risks and uncertainties, and actual results may differ materially. The specific risks include, without limitation: NamiDB Cloud is in closed beta, has no published price list, and generates no revenue; no investor has committed to this round and there is no lead; the published benchmarks cover LDBC SNB Interactive at scale 1.0 on four IC queries only and do not establish performance at any other scale; partner and university agreements are private and unaudited by any third party; third-party market forecasts are estimates by their authors and not commitments; and the engine is licensed under BSL 1.1, which restricts some commercial uses until conversion. NamiDB, Inc. undertakes no obligation to update any forward-looking statement. NamiDB, Inc. is not a reporting company, and the safe harbor for forward-looking statements under the Private Securities Litigation Reform Act of 1995 is not available to this offering.
No advice. Nothing on this page is investment, legal, tax, or accounting advice. Prospective investors should consult their own advisors.
Summary only, not complete. The information on this page is a summary and does not contain all information a prospective investor should consider. It is qualified in its entirety by the definitive offering documents. Third party figures are attributed where used and were read in August 2026.