At some point today, you will trust a computer you have never seen, in a building you could not locate, run by a company whose contract you have not read. It might approve a payment, schedule a shift or hold the small square of data that lets your own software wake up. The arrangement feels weightless. Then the service stops, and the invisible acquires mass.
Jonathan Hatzor built his second company around that awkward moment. Parametrix does not keep the cloud running. It measures when cloud, software and data-center services do not run, calculates what the interruption can cost, and structures insurance that pays when an agreed trigger is met. In a business famous for paper, the proposition has the neatness of a circuit: event, signal, payment.
The simplicity was expensive to earn. Before an insurer could cover digital downtime, somebody had to make downtime observable. Before a reinsurer could support the policy, somebody had to show how one provider's failure might accumulate across a portfolio. Before a broker could sell it, somebody had to explain why the client's cyber policy might not answer as expected. Parametrix had to build the instrument and the market's confidence in the instrument at once.
01 / The useful obsessionStart with the blind spot
Hatzor came to insurance through software. From 2016 to 2019, he co-founded and served as chief technology officer of Matter, which used 3D imagery to create virtual property tours. The company was acquired by Treedis. The jump from buildings rendered in three dimensions to insurance policies looks severe until the common habit appears: take something difficult to see, give it coordinates, and make it useful.
He has said the Parametrix idea grew from an interest in cloud technology and a wish to innovate inside insurance. He and co-founders Ori Cohen and Tamir Carmel looked beyond established lines toward risks that ordinary policies did not address cleanly. Cloud downtime stood out. Companies were moving essential work onto third-party infrastructure, yet the financial interruption could sit in a coverage gap. The cloud had become ordinary. Its failure had not.
Their answer was parametric. Traditional business-interruption claims can require a long excavation of what was lost and why. A parametric policy agrees in advance on an external, measurable event and the payment attached to it. Parametrix watches the services, defines the trigger and supplies the data. The philosophical shift is small but consequential: settle the argument before the outage.
The outage-to-payout circuit
02 / The hard saleBorrowed trust becomes earned trust
Ask Hatzor for a favorite Parametrix memory and he does not reach first for a fundraising announcement. He points to securing reinsurance capacity from new partners. An unknown startup was proposing a new product for a risk the market had barely recognized. Established reinsurers had to believe both the peril and the machine that measured it.
“Getting their buy-in wasn't just about financial backing; it was about validating our vision.”Jonathan Hatzor, on winning reinsurance support
That distinction explains much about building in insurance. Software can ship on the founder's conviction. A policy needs capital, licenses, distribution and a shared faith in definitions. Parametrix could not merely be right about cloud dependency. It needed conservative institutions to put their balance sheets beside the claim.
Broker education became another stubborn piece. Hatzor has described brokers as trusted advisers with little spare attention beyond renewals and familiar lines. A novel policy may solve a real problem and still fail to travel. Parametrix invested in repetition: explaining the gap, explaining the trigger, explaining why the new cover belonged beside redundancy and existing cyber protection. Innovation, in this corner of finance, sometimes wears the costume of a patient seminar.
03 / From policy to marketWhen an outage becomes an asset class
In April 2024, the idea reached an unusual destination. Parametrix served as modeling and calculation agent for Cumulus Re, Hannover Re's catastrophe bond covering sustained cloud-outage risk. Cat bonds are instruments for transferring rare, severe insurance exposures to capital-market investors. A cloud interruption had become defined enough to sit in a structure more commonly associated with wind, earthquakes and other physical catastrophes.
The bond mattered because digital risk tends to bunch. Thousands of companies may depend on the same provider, region or application. An insurer writing each one individually can discover, too late, that many supposedly separate policies share the same switch. Parametrix's Infrasight and analytics work map those concentrations. The product is not a dramatic warning that everything will fail everywhere. Hatzor's own formulation is more measured: the data suggests a universal global collapse is unrealistic, while a large event could still exceed recent incidents.
That nuance is the business. If every cloud risk were perfectly correlated, it would be difficult to insure. If none were correlated, there would be little need for specialized accumulation models. The map between those poles tells carriers how much limit they can offer, where they may need reinsurance and which parts of a portfolio genuinely diversify one another.
“Continuity is definitely better than interruption, even if the interruption is followed by an insurance claim that's paid fast.”Jonathan Hatzor, on practical resilience
04 / Keep the paper scheduleA technologist's analog streak
Hatzor is not sentimental about an insurance payout. He has argued that companies should understand and avoid losses before they transfer them. His best illustration is wonderfully mundane: businesses have ground to a halt when a cloud outage disabled employee scheduling. If the schedule is mission-critical, he said, keep a contingency. The old paper process probably worked a few years ago. Preserve the knowledge needed to use it.
It is a useful quirk in the pitch. The founder selling a sophisticated technology policy also recommends remembering where the clipboard lives. Insurance and redundancy are not rival religions. A company can distribute services across providers, write a fallback procedure and insure the loss it cannot economically engineer away. The objective is not to win a claim. It is to remain a business.
His public voice follows that practical line. In 2024, after the CrowdStrike disruption, Hatzor focused on the difficulty of tracing losses through the digital supply chain and the potential scale of a true cloud-provider event. In 2025, after an AWS us-east-1 outage, Parametrix said it identified affected customers, estimated exposure and began its claims process within 24 hours. Each incident served as both evidence and curriculum.
05 / The infrastructure decadeUptime moves into the finance model
By late 2025, Parametrix had widened its frame. Its data-center SLA insurance protects operators from contractual service-credit exposure when performance slips. CyberPMX embeds parametric digital business-interruption protection inside a conventional cyber and technology-errors policy. The monitoring network, the company said, covered more than 7,000 SaaS, PaaS and IaaS services and over 750 data centers, producing more than a billion performance observations each month.
The timing is pointed. AI infrastructure demands vast, expensive facilities whose economics depend on sustained performance. Downtime can become a rent credit, a covenant problem or a mark against an asset's value. Hatzor increasingly speaks to lenders, data-center operators and investors because uptime is no longer only an engineer's service metric. It is entering the spreadsheet where capital decides whether to arrive.
Parametrix announced a $27 million Series B in December 2025, led by Mundi Ventures, FirstMark Capital and Hannover Digital Investments, after reporting threefold annual top-line growth. Hatzor said the money would accelerate expansion of the category, product work and teams in the United States. In his own announcement, he joined the corporate milestone to a personal one: he and his wife, Maya, had welcomed their second daughter, Lily, joining older sister Emmy. Growing a company while growing a family, he wrote, had sharpened his purpose.
That note lands because his career is otherwise built around abstractions: latency, aggregation, triggers, reserves. Family restores the scale. The cloud may be a planetary network, but an outage becomes real one stopped shift, one stalled shop and one worried operator at a time.
06 / The quiet productLegibility
The obvious Parametrix product is insurance. The more durable product may be legibility. Hatzor and his colleagues have spent years giving an invisible dependency a history, a frequency, a severity and a price. Once those exist, more people can act: a risk manager can buy cover, an insurer can adjust a portfolio, a reinsurer can supply capacity, and an investor can decide whether a data center's promise is credible.
There is a founder lesson tucked inside the actuarial machinery. New categories do not begin when somebody coins a phrase. They begin when the tools arrive that let other people make decisions. Parametrix had to watch the cloud before it could insure the cloud. It had to build the signal before it could sell certainty.
The internet will blink again. It is too useful, too complicated and too physical not to. Hatzor's wager is that the pause need not remain a mystery. Measure it carefully, agree what it means, and the invisible can finally appear on the balance sheet.