{"slug": "space-and-time-enterprise-1-proof-0-prices-2026", "title": "Space and Time Enterprise: 1 Proof, 0 Prices [2026]", "summary": "Space and Time launched Space and Time Enterprise on November 9, 2026, a data-privacy platform built on its patented Proof of SQL technology that lets businesses prove calculations on private data are correct without exposing the underlying records, according to the company's GlobeNewswire release. The Microsoft-backed company, whose investors include Microsoft's M12 venture fund, says AI applications and auditors can verify a query result without touching the private database. Space and Time did not clearly establish pricing, performance figures or availability details in the records reviewed.", "body_md": "Space and Time introduced **Space and Time Enterprise** on November 9, 2026, a data-privacy platform that lets a business [prove that calculations on private data are correct without exposing the underlying records](https://www.globenewswire.com/news-release/2026/11/09/3398785/0/en/space-and-time-enterprise-brings-privacy-and-verifiable-data-to-ai-applications.html). The Microsoft-backed company says AI applications, auditors, and other stakeholders can check a result without ever touching the private database.\n\nThe engine under the product is Proof of SQL, which Space and Time describes as patented technology. It attaches a verifiable proof to a database query so a third party can confirm the computation ran correctly. Space and Time is backed by M12, Microsoft’s venture fund, which gives the launch more weight than a typical crypto-adjacent infrastructure announcement.\n\nThis analysis covers what the launch confirms, what it leaves open, and why the distinction matters. A proof of correct computation is a narrow guarantee, and AI teams already wrestling with [verifying which model answered](https://shattered.io/anthropic-flashprove-model-substitution-zksnark-v0-16-0-2026/) will recognize the pattern. The rest of this piece sorts the confirmed facts from the marketing language and offers five predictions for what comes next.\n\nDon't miss new tech stories on Google\n\nAdd Shattered once in the Google app and our stories show up in your news suggestions.\n\n[Add now](https://www.google.com/preferences/source?hl=en&q=https%3A%2F%2Fshattered.io%2F)\n\n## What Space and Time Enterprise Actually Is\n\nAccording to the company’s release, distributed through GlobeNewswire, Space and Time Enterprise is a platform built for organizations that hold sensitive data and want AI systems to use it without handing over the records. The core promise is simple to state. A business runs a calculation on private data, and a stakeholder can later confirm the answer is correct without seeing the data that produced it.\n\nThe stakeholders named in the announcement include AI applications and auditors. That pairing is revealing. Auditors have always needed access to underlying records to sign off on a number, and that access carries its own privacy and liability cost. An AI application, meanwhile, often needs to consume a figure from a database it has no business reading in full. One verification mechanism serves both audiences.\n\nThe announcement date, November 9, 2026, matters for a second reason. Coverage from Cryptonews, syndicated on Yahoo Tech, framed the launch as a platform that lets AI agents verify private data. Other outlets, including [Crypto Briefing](https://cryptobriefing.com/space-and-time-enterprise-ai-private-data/), wrote it up as a platform for AI verification of private data more broadly. Both framings point at the same product, but they do not make identical claims, a gap the agents section below examines.\n\nPricing, performance figures, and availability details were not clearly established in the records reviewed for this report. Readers should treat any number floating around social media about speed or cost as unverified until Space and Time publishes documentation. For now, the launch is a product category announcement with a named technology behind it, not a spec sheet.\n\n## How Proof of SQL Works in Plain Terms\n\nSQL is the standard language businesses use to ask questions of a database. Someone writes a query, the database crunches the matching records, and a number or a table comes back. The weak point is trust. The person reading the answer has to believe the database ran the query faithfully, on the right records, without a bug or a tampered row along the way.\n\n### The receipt idea\n\nProof of SQL attacks that trust gap by generating a proof alongside the result. Think of it as a receipt. The receipt ties the answer to the query and to the specified records, and anyone holding it can check the math without rerunning the query on the raw data. Per the company’s description, the technology proves that a computation was correctly executed on the records in question.\n\n### Where the cryptography comes in\n\nReports on the launch describe the approach as drawing on [zero-knowledge](https://shattered.io/zero-knowledge-proof-age-verification-vs-id-upload-2026/) techniques, a branch of cryptography that lets one party demonstrate a statement is true without revealing the information behind it. The fact sheet for this story does not detail the proof system, so this article does not describe its internals. What matters for buyers is the interface: a query goes in, a result and a verifiable proof come out, and the private rows stay where they are.\n\nThat design separates two jobs that normally live together. One party stores and processes the data. Another party, which may be an auditor, a counterparty, or an AI application, only needs to be convinced the processing was honest. Splitting those roles is the entire product thesis.\n\n## What the Proof Does Not Prove\n\nThe most important sentence in the verified coverage is a limitation. Proof of SQL can verify that a computation was executed correctly on specified records. It does not establish that those records are themselves accurate. That caveat appeared in the reporting on the launch, and any honest reading of the product has to start there.\n\nConsider a lender that feeds an AI underwriting tool a table of income figures. If someone entered a wrong salary into that table, Proof of SQL would still certify that the average was calculated correctly from the wrong salary. The receipt confirms the arithmetic and the query logic. It cannot reach back to the person who typed the number.\n\n| Question a stakeholder asks | Does Proof of SQL answer it? | Why | \n|---|---|---|\n| Was the SQL query executed correctly? | Yes | This is the stated purpose of the technology | \n| Was it run on the specified records? | Yes | The proof links the result to those records | \n| Are the underlying records accurate? | No | Coverage states the proof does not establish data accuracy | \n| Is the dataset complete? | No | Completeness is a data-quality property, not a computation property | \n| Did the data come from a trustworthy source? | No | Provenance sits outside the scope of a query proof | \n| Can the verifier see the private rows? | No | The platform is built to prove results without exposing records | \n\nThe last three rows of that table reflect reasoning from the stated limitation rather than separate claims from the company. They follow logically from a proof that covers computation only. Buyers should still ask Space and Time directly how it recommends pairing the product with data-quality controls.\n\n## Why Private Data Is the Sticking Point for Enterprise AI\n\nEnterprise AI projects tend to stall on one question: can the model use our sensitive data without leaking it? Banks, insurers, hospitals, and manufacturers hold records that regulators, contracts, and common sense keep behind locked doors. Yet the most valuable AI use cases need exactly those records.\n\nThe usual workarounds each carry a cost. Copying data into a vendor’s environment widens the attack surface. Redacting fields strips out the signal. Letting an agent roam a database with broad credentials invites mistakes, a risk that [the Claude Code secrets exposure on GitHub](https://shattered.io/claude-code-settings-local-json-1379-secrets-github-2026/) illustrated in a very different setting. When tools hold more access than they need, something eventually leaks.\n\nSpace and Time Enterprise targets a narrower slice of that problem. It does not claim to make a model safe or a dataset clean. It offers a way for an application to receive a computed answer and a proof, so the application never needs raw rows. If the numbers an agent acts on arrive with a receipt, a human or an auditor can later confirm those numbers were not fabricated by the pipeline.\n\nThat framing also fits how enterprises buy security tooling. Procurement teams rarely adopt a product because it is clever. They adopt it because it lets them say yes to a project that legal and compliance would otherwise block. A verifiable-computation layer gives risk officers a concrete artifact to review.\n\n## AI Agents Versus AI Applications: Reading the Headline Carefully\n\nThe headline carried by Cryptonews and syndicated on [Yahoo Tech](https://tech.yahoo.com/ai/copilot/articles/microsoft-backed-space-time-launches-130102976.html) says the platform lets AI agents verify private data. The company release uses broader language about AI applications. The exact wording that the platform supports AI agents as a formal product capability was not established in the release itself.\n\nThat is not a trivial distinction. An AI application can be a dashboard that summarizes a verified revenue figure. An AI agent takes actions, calls tools, and chains steps together with some autonomy. A verification layer that serves the first category does not automatically deliver the guarantees an autonomous agent needs, such as proofs for each step of a multi-tool workflow.\n\nAgent builders are already hunting for guardrails. Cost control is one example, covered in our look at [the three agent limits Forrester wants in place](https://shattered.io/ai-bill-shock-forrester-three-agent-limits-openai-invoice-2026/). Verifiable inputs are another guardrail, and a natural extension of the same discipline. An agent that books a payment based on a proven query result is easier to defend than one that trusted a free-form database read.\n\nThe takeaway for readers: use the agents framing as a description of where the market is heading, not as a confirmed feature list. Until Space and Time publishes agent-specific documentation, the supportable claim is that AI applications can verify query results.\n\n## The M12 Backing and What It Signals\n\nSpace and Time is backed by M12, Microsoft’s venture fund, per the announcement coverage. Corporate venture backing is not an endorsement of any specific product, and nothing in the verified record says Microsoft will integrate or resell Space and Time Enterprise. The relationship is an investment, and readers should not read it as a distribution deal.\n\nStill, the backer shapes how enterprise buyers perceive risk. A startup selling cryptographic verification to a bank faces a credibility hurdle. A name like M12 on the cap table does not clear that hurdle, but it signals that at least one large technology company’s investment team examined the technology and chose to back it. That tends to shorten the first meeting.\n\nThe company has long talked about verifiable queries over both on-chain and off-chain data, as White’s comments later in this piece show. Enterprise Edition aims that mindset at private corporate databases, which is a more conservative market. The move is plausible, but it will test whether the technology can satisfy procurement checklists built for traditional software.\n\nCoverage also identified Nate Holiday, CEO of MakeInfinite Labs, as a speaker around the launch. A verbatim official quote from him was not clearly established in the records reviewed, so this article does not put words in his mouth.\n\n## Confirmed Versus Unconfirmed: The Fact Check\n\nLaunch coverage tends to blur what a company said, what a reporter inferred, and what a headline writer compressed. The table below separates those layers for Space and Time Enterprise, using only what the verified record supports.\n\n| Item | Status | Source basis | \n|---|---|---|\n| Launch date of November 9, 2026 | Confirmed | Company release and launch coverage | \n| Data-privacy platform for proving calculations on private data | Confirmed | Company release | \n| Powered by patented Proof of SQL | Confirmed | Company release and coverage | \n| Backed by M12, Microsoft’s venture fund | Confirmed | Company release | \n| Proof does not establish record accuracy | Confirmed | Launch coverage | \n| Pricing | Unconfirmed | Not clearly established | \n| Technical performance figures | Unconfirmed | Not clearly established | \n| Availability details | Unconfirmed | Not clearly established | \n| Formal AI agent capability | Unconfirmed | Release speaks of AI applications | \n| Verbatim quote from Nate Holiday | Unconfirmed | Named as speaker, wording not established | \n\nFive items are solid and five are open. That ratio is normal for a launch day, but it argues for patience. Teams evaluating the product should request a technical brief, a pricing sheet, and a clear statement on agent support before committing engineering time.\n\n## Expert View: Space and Time on Trust-Minimized Queries\n\nThe quotes below come from Jay White, co-founder and head of research at Space and Time, in a [company blog explainer on Proof of SQL](https://www.spaceandtime.io/blog/exploring-space-and-time-proof-of-sql). They describe the technology generally and do not come from the Enterprise launch itself, so read them as the company’s philosophy rather than a product specification.\n\nWhite frames the problem as a database question. In his words, “We need some database where you can query the database using SQL, and you can get a result back and you can guarantee, you can know, that you get the right result without having to trust that database.”\n\nHe then explains what that guarantee changes for operators and customers. “What it means is that you don’t have to trust Space and Time. You don’t have to trust your data warehouse,” White says. He compresses the pitch further elsewhere with a single line: “You don’t need to trust anyone.”\n\nWhite also describes how failures would surface. “The only time it’s going to execute is if the warehouse actually does something malicious, and then you’ll get a notification that, hey, they did it wrong.” That sentence is the clearest statement of the model. Verification stays quiet when everything works and raises an alarm when a result does not match its proof.\n\nTwo points stand out. First, the language is about trusting the database operator, not about trusting the data. That lines up with the limitation covered earlier. Second, “malicious” is a strong word, and most real-world data errors are honest mistakes, which a computation proof will not catch.\n\n## Market Impact: Who Gains and Who Feels Pressure\n\nThe launch lands in a market where AI vendors keep promising enterprise readiness while customers keep asking for proof. Several groups have reason to pay attention.\n\n- **Regulated industries:** Finance, insurance, and healthcare teams gain a possible way to let AI consume computed results without moving raw records.\n- **Audit and assurance firms:** Verifiable computation could shift part of an audit from sampling records to checking proofs, though that is a forecast and not a stated product feature.\n- **Data warehouse vendors:** Customers may begin asking whether query results can ship with proofs, which turns verification into a competitive checkbox.\n- **AI platform vendors:** Model providers face growing pressure to show customers how inputs were produced, not only how outputs were generated.\n\nWhether any of this becomes a large revenue line is unknown. Space and Time did not disclose pricing in the material reviewed, and no customer names appeared in the verified record. Early adopters will likely be firms that already feel regulatory heat and have engineers comfortable with cryptographic tooling.\n\nBroader AI trust is a live political topic as well. Our reporting on [how 76% of voters want AI rules](https://shattered.io/silicon-valley-trump-playbook-cracks-76-percent-want-ai-rules-2026/) shows public appetite for accountability. Products that produce an auditable trail give companies a concrete answer when regulators or customers ask how an AI-driven number was derived.\n\n## Competitive Comparison: Ways to Trust a Query Result\n\nSpace and Time Enterprise does not compete in a vacuum. Teams already have several ways to gain confidence in a database answer. The table compares the main approaches in general terms. It describes categories, not specific vendors, and it contains no benchmark data.\n\n| Approach | What you trust | Exposes raw data to verifier? | Main weakness | \n|---|---|---|---|\n| Vendor attestation and contracts | The database operator | Often yes during audits | Depends on honesty and process | \n| Audit logs and access reviews | The logging system | Usually yes | Logs can be incomplete or altered | \n| Hardware-based confidential computing | Chip and firmware makers | No, in principle | Trust shifts to hardware vendors | \n| Proof-based verification such as Proof of SQL | The math and the proof system | No | Proves computation, not data accuracy | \n| Redacted or synthetic data sharing | The redaction process | Partly | Loses detail and signal | \n\nProof-based verification sits in an interesting spot. It asks the verifier to trust mathematics rather than a company or a chip. In exchange, it covers a narrower claim than some alternatives. A team could reasonably combine approaches, for instance pairing proofs with strong data-entry controls and access reviews.\n\nThe comparison also shows why this category will not replace existing controls. No single technique answers every question a regulator asks. Proof of SQL covers the computation step and leaves provenance, completeness, and accuracy to other tools.\n\n## Historical Context: From Audit Logs to Cryptographic Receipts\n\nBusinesses have tried to prove their numbers for as long as they have kept books. Paper ledgers gave way to database audit trails, and audit trails gave way to third-party attestations. Each step reduced one kind of risk while leaving the underlying trust question in place.\n\nZero-knowledge proofs began as an academic idea in the 1980s, the product of cryptographers asking whether you could convince someone of a fact without revealing why it was true. For decades the techniques stayed mostly theoretical because the computing cost was too high. More recently, falling computing costs have pulled them toward practical products.\n\nSpace and Time has pitched verifiable queries across both on-chain and off-chain data. Enterprise Edition points that idea at private corporate data, where the trust problem looks different but the need for proof is similar.\n\nThe AI boom adds urgency. When a model summarizes a data table, nobody can easily check the summary against the table. Verified query results offer a way to anchor model inputs to numbers that can be proven, even if the model’s own reasoning remains opaque. The same worry about checking what AI systems actually did runs through [our guide to evaluating privacy tools before trusting them](https://shattered.io/how-to-evaluate-privacy-tools-before-trusting-them-with-your-data/).\n\n## Risks and Open Questions for Buyers\n\nNo product launch resolves every question, and this one leaves several open. A careful evaluation should probe the following areas before any pilot.\n\n### Scope of the guarantee\n\nAsk which SQL features the proof system supports. Real enterprise queries use joins, aggregates, window functions, and vendor-specific extensions. A proof system that covers only a subset limits where the product fits. The verified record does not list supported query types.\n\n### Operational fit\n\nPerformance and cost were not established in the reviewed material. Proof generation adds work on top of normal query execution, so latency, hardware needs, and pricing will determine whether the product suits interactive workloads or only batch reporting. Request measured numbers on your own data shape instead of relying on marketing claims.\n\nTeams should also ask how verification keys, proof formats, and versioning are managed, who audits the proof system itself, and what happens when a proof fails. A failed proof is the whole point of the design, yet an [incident response](https://shattered.io/category/security/) plan for it does not appear in the public material. Treat that as a pilot requirement.\n\n## Predictions: What to Watch Next\n\nThese are analyst judgments based on the confirmed facts, not reported plans from Space and Time.\n\n- **Documentation will clarify agent support.** The gap between the agents headline and the applications wording is easy to close. Expect the company to publish technical material that spells out which workflows its proofs cover.\n- **Pricing will follow pilots.** Enterprise tooling rarely posts prices at launch. Look for pilot announcements with named design partners before any public rate card.\n- **Rivals will adopt the same language.** Once verified AI inputs become a buying criterion, other data and AI vendors will add their own claims about provable results, some stronger than others.\n- **Buyers will pair proofs with data-quality tooling.** The accuracy limitation is too obvious to ignore. Vendors who bundle provenance and validation checks with computation proofs will tell the more complete story.\n- **Regulators will notice.** Any technique that lets a firm show how an AI-relevant number was computed gives supervisors a new thing to ask about, whether or not rules ever mandate it.\n\nThe riskiest prediction is the second one, since a surprise price announcement is always possible. The safest is the fourth, because it follows directly from the limitation the coverage itself highlighted.\n\n## What This Means for Developers and Data Teams\n\nFor engineers, the practical lesson is to design AI pipelines so that every number a model or agent acts on has a traceable origin. Even without adopting Space and Time Enterprise, the principle holds. Log the query, the data version, and the result together. Treat a free-form database read inside an agent loop as a risk to be justified, not a default.\n\nData teams should map where private data meets AI in their stack and rank those points by sensitivity. Places where a computed figure, rather than raw rows, would satisfy the application are the best candidates for a verification layer. Places where the model truly needs raw text or records will need different controls.\n\nSecurity and compliance leads can use the launch as a prompt to ask vendors a pointed question: how can I confirm that the number your AI shows me came from the data you say it did? Answers will vary widely, and the quality of the answer is informative. More of our model coverage lives in the [AI and machine learning section](https://shattered.io/ai-machine-learning/), including the [Meta Muse Ember agent model](https://shattered.io/meta-muse-ember-agent-execution-model-0-10-per-million-tokens-2026/) that shows how fast agent tooling is moving.\n\nThe takeaway: Space and Time Enterprise is a credible, narrowly scoped answer to a real problem. It lets organizations show that a calculation on private data was done correctly, without opening the database to the party checking. Backing from M12, Microsoft’s venture fund, lends it standing, and Proof of SQL gives it a named, patented technology to point to.\n\nThe limits are just as clear. The proof does not vouch for the accuracy of the records, the agents wording goes further than the release does, and pricing, performance, and availability remain unconfirmed. Watch for documentation and named pilots over the coming weeks. Those will show whether the launch is a durable enterprise product or a well-timed positioning move in the race to make AI answers provable.\n\n## Frequently Asked Questions\n\n### What is Space and Time Enterprise?\n\nIt is a data-privacy platform introduced by Space and Time on November 9, 2026. It lets businesses prove that calculations on private data are correct without exposing the underlying records.\n\n### What is Proof of SQL?\n\nProof of SQL is Space and Time’s patented technology for proving that SQL queries were computed correctly. It produces a verifiable proof that ties a result to the query and the specified records.\n\n### Does Proof of SQL prove my data is accurate?\n\nNo. It verifies that a computation was executed correctly on specified records. It does not establish that those records are themselves accurate.\n\n### Who backs Space and Time?\n\nSpace and Time is backed by M12, Microsoft’s venture fund. The reviewed record describes an investment relationship and does not mention a product integration or reseller deal with Microsoft.\n\n### Does the platform officially support AI agents?\n\nThe Cryptonews headline refers to AI agents, but the company release uses broader language about AI applications. A formal agent capability was not established in the release.\n\n### How much does Space and Time Enterprise cost?\n\nPricing was not clearly established in the available records. Contact the vendor for current terms.\n\n### Who spoke about the launch?\n\nCoverage identified Nate Holiday, CEO of MakeInfinite Labs, as a speaker. A verbatim official quote from him was not clearly established, so none is reproduced here.\n\n### Can auditors check results without seeing the database?\n\nAccording to the company release, yes. AI applications, auditors, and other stakeholders can verify results without accessing the private database.", "url": "https://wpnews.pro/news/space-and-time-enterprise-1-proof-0-prices-2026", "canonical_source": "https://shattered.io/space-and-time-enterprise-proof-of-sql-ai-private-data-nov-9-2026/", "published_at": "2026-11-09 20:43:28+00:00", "updated_at": "2026-11-09 20:51:47.612090+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-infrastructure", "ai-agents"], "entities": ["Space and Time", "Proof of SQL", "Microsoft", "M12", "GlobeNewswire", "Cryptonews", "Crypto Briefing"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/space-and-time-enterprise-1-proof-0-prices-2026", "markdown": "https://wpnews.pro/news/space-and-time-enterprise-1-proof-0-prices-2026.md", "text": "https://wpnews.pro/news/space-and-time-enterprise-1-proof-0-prices-2026.txt", "jsonld": "https://wpnews.pro/news/space-and-time-enterprise-1-proof-0-prices-2026.jsonld"}}