Israel correspondent, Tel Aviv — a Western-aligned lens, stated openly. The innovation economy, Eastern Mediterranean gas, and defense technology. Signed columns, each an argument; the views are the correspondent’s own.
Opinion — the views in these columns are the correspondent’s own.
The innovation economy, Eastern Mediterranean gas, and defense technology.
Career
Tamar Levi is 31, writes like a database and argues like a prosecutor. A statistics graduate of the Hebrew University of Jerusalem, she came to journalism through data — scraping startup funding rounds in Tel Aviv for a student project, then realizing the dataset was telling a story about power that no opinion column could. Four years covering Israel's startup economy as a reporter, then a year in Berlin on Europe's very different relationship with innovation: cautious where Tel Aviv is reckless, regulated where Tel Aviv improvises.
She states her lens openly, as the Bureau's correspondents do: she is pro-West, and she makes no apology for it. Open societies out-innovate closed ones, she argues, and the data — patents, startups, capital deployed — backs her up. Her readers know exactly where she stands — which is the point: declared bias, she argues, beats everyone else's performed neutrality.
What shaped her: the week in 2023 when a defense-tech story she had been building for months — a dataset mapping dual-use exports — collided with a breaking conflict, and she had to decide in one night what could be published and what could get people killed. She published the data, held the names, and won the 4 a.m. argument. "Numbers don't have opinions," she says. "But the person choosing which numbers to count does. I tell you which person I am."
Based in Tel Aviv, she covers the innovation economy, Eastern Mediterranean gas, and defense technology.
Personal
Tamar Levi grew up in Haifa, the daughter of a software engineer and a nurse, on a street where the hill met the sea. She lives in Florentin, Tel Aviv, with her long-term partner, a chef — they have no children, and the apartment's real resident is the dog, a street rescue with opinions. She is a rock climber, spending weekends on limestone crags where her phone, by design, gets no signal. Hebrew and English are spoken at home, sometimes mid-sentence; her partner says this is what happens when a journalist falls in love with a man who reads recipes the way she reads leaks.
Timeline
2018 — Statistics degree, Hebrew University of Jerusalem
2019–2022 — Technology and startup reporter, Tel Aviv
2023 — Innovation correspondent, Berlin
2024–2025 — Data journalist covering defense tech and energy, Tel Aviv
2026 — Correspondent for innovation and technology, Magna Bureau
OpinionThe innovation economy
The Miracle Had a Price Tag.
By Tamar Levi · Israel correspondent · Tel Aviv, September 29, 2026
In January 2020, no coronavirus vaccine had ever been licensed anywhere on earth. By December, two of them were in human arms. Eleven months. That is the part they engrave on the trophy. This column is about the rest of the story — the part with the money.
Start with the anomaly, because the anomaly is the whole story. SARS appeared in 2003. Roughly twenty vaccine candidates entered development; most never reached a clinic, because the outbreak was contained and the money walked away. MERS appeared in 2012; its candidates did little better — a Phase 1 here, a stalled trial there. Researchers were admirably blunt about why: with so few cases, an efficacy trial would have needed a hundred thousand participants, which was, in their word, not feasible. To this day, no vaccine has ever been licensed for SARS or for MERS. The industry’s position was never that a coronavirus vaccine was impossible. It was that nobody would pay for one.
Then came the paying customer. Operation Warp Speed, launched May 15, 2020, was a roughly $10 billion public-private program with a stated target of 300 million doses by January 2021. Read its contracts the way an investor would. Moderna received about $955 million for development and a $1.5 billion advance order for 100 million doses — and its vaccine, mRNA-1273, was co-developed with scientists at the NIH’s own Vaccine Research Center, who had been working on a MERS spike-protein vaccine that gave the whole effort its head start. Pfizer took no federal research money — give credit where it is due — but it took a $1.95 billion advance purchase agreement for 100 million doses, while its partner BioNTech took a €365 million grant from the German government and a €100 million development loan from the European Investment Bank. Governments did not merely fund the research. They bought the product before it existed.
Now the liability — or rather, the absence of it. Effective February 4, 2020, the U.S. Health Secretary issued a PREP Act declaration covering COVID-19 countermeasures: manufacturers, distributors, and administrators were immunized from legal liability for claims arising from the design, development, testing, manufacture, labeling, marketing, sale, and use of the products. The only remedy left standing was a federal compensation program. Consider what that means as a business proposition. The research was subsidized. The purchase was guaranteed in advance. And the lawsuits were, by declaration, illegal.
“The research was subsidized. The purchase was guaranteed in advance. And the lawsuits were, by declaration, illegal.”
The numbers that followed belong in the commercial record books. Pfizer reported $36.781 billion in Comirnaty revenue in 2021 — the largest single-year sales figure for any pharmaceutical product in history, nearly double the previous record-holder, Humira, at $20.7 billion. In 2022 it did $37.806 billion. Across 2021–2024, Pfizer’s own filings show roughly $91.2 billion in recognized Comirnaty revenue — with, by the company’s guidance, an income-before-tax margin in the high twenties, after a fifty-percent gross profit split with BioNTech. Moderna’s arc is steeper: $803 million in total revenue in 2020, $18.5 billion in 2021, net income of $12.2 billion — a company that had never sold a product in its history, transformed in twelve months into one of the most profitable enterprises on earth.
And then the price went up. With government stockpiles exhausted, Pfizer floated a commercial price of $110 to $130 per dose for 2023. Two American legislators wrote, in fury, of “yet another massive corporate payday from the ongoing pandemic,” noting the price bump would land on top of tens of billions already banked. The fury changed nothing. The price held.
Europe’s chapter deserves its own telling, because it is still under criminal investigation. In 2021, the president of the European Commission negotiated the EU’s largest vaccine contract — up to 1.8 billion doses, worth up to €35 billion — in part through text messages with Pfizer’s chief executive. When journalists asked to see the messages, the Commission first said they had not kept them, then said they were too ephemeral to count as documents. The European Ombudsman found maladministration. In May 2025, the EU’s General Court ruled the Commission had been wrong to refuse access. The European Public Prosecutor’s Office has taken over a Belgian criminal probe into the procurement. No charges have been filed — say that plainly — but consider the sentence you just read: a €35 billion public contract, negotiated by SMS, and the messages are gone.
None of this erases the science, and a serious column says so. The mRNA platform was decades in the making; the spike-protein work on SARS and MERS, though never commercialized, handed the COVID effort its running start; running trials in parallel and manufacturing at risk were genuine logistical feats. The eleven months from published sequence to Phase 3 readout were earned — and they were also bought, shielded, and pre-sold. Both things are true. That is precisely the point.
Every founder dreams of product-market fit. This was something rarer: the market was mandatory, the customer was everyone, the research was publicly funded, the purchase was guaranteed before approval, liability was abolished by declaration, and the biggest contract in European history was negotiated by text message. You can admire the science with your whole heart and still ask, with a straight face, what business the business was in.
It was never just a vaccine. It was the greatest product launch in the history of capitalism — and like all great launches, the product was only half of it.
Loading the discussion…
OpinionThe future of work
The Job Is Dead. Long Live the Ledger.
By Tamar Levi · Israel correspondent · Tel Aviv, September 27, 2026
AI is not taking jobs — it is unbundling them. The successor unit of economic identity is not the gig; it is the verified ledger of what you can do.
The job was a package deal — and the deal is expiring.
For about a century, the word "job" did an astonishing amount of work in a human life. It was not merely a way of earning money. It was an identity ("what do you do?"), a pension plan, a health policy, a social circle, a schedule, a wardrobe, and — in the unspoken sociology of it — a verdict on your worth. Lose the job and you did not simply lose income; you lost your answer to the first question strangers ask.
Artificial intelligence is not taking jobs, in the simple sense the headlines imply. It is doing something more profound: it is unbundling the bundle. The functions a job performed are being separated, automated, or relocated — and the single unit called "employment" is dissolving into its constituent parts. What comes next is not mass unemployment. It is the end of the job as the atom of economic identity.
The bundle, itemized
Consider what your grandparents' job actually contained, once you unpack the box:
Income — the obvious part. A predictable flow, priced by the month. Identity — a title that answered the social question before it was asked. Insurance — pensions, sick leave, the feeling that the future was hedged. Community — colleagues, rivals, gossip, the daily theater of other people. Structure — a rhythm: mornings, deadlines, the small mercies of routine.
Five functions, one container. The genius of the twentieth-century firm was bundling them so tightly that people confused the container for the contents. The tragedy, now unfolding, is discovering the difference.
AI attacks each function separately. Income-generating tasks — drafting, summarizing, coding, analyzing, designing — are the first to be automated, because they were always the most legible part of work. Structure is being replaced by the algorithmic gig: tasks arrive, you accept, the software schedules you. Community has already migrated to chat windows. Identity is thinning fastest of all — try explaining to your children that you are "an analyst" when the analysis is done by a model at 3 a.m. for fractions of a cent.
What survives the unbundling is insurance — and that is precisely the part the market provides worst on its own. The pension was the job's quiet miracle: deferred, paternal, unglamorous. It turns out you cannot automate the promise that someone will take care of you when you are old. You can only fund it. And the funding models assumed the bundle would last forever.
The ledger era
Here is the thesis, stated plainly: the successor unit to the job is not the gig. It is the ledger — a verified, portable record of what you can actually do, proven by output rather than by title.
Think about how work was priced before the job. The guild master did not hire a journeyman for his attendance; the ledger recorded his pieces. The nineteenth-century pieceworker, the Hollywood contractor, the open-source contributor — all operated on ledgers of proven output. The job was the interlude: a century in which firms found it cheaper to rent people by the month than to measure them by the task, because measurement was expensive. AI has made measurement cheap. The interlude is over.
The emerging economy already shows its contours. Reputation markets — ratings, portfolios, verified delivery histories — are becoming the resume that matters; nobody asks for your job title on a platform, they ask for your stars. Micro-credentials outrank diplomas in hiring pipelines that never see a human. The most valuable workers are accumulating what economists should call attested capacity: proof, cryptographically or socially verifiable, that they have done the thing, on time, well.
We are returning to a piecework world — except the pieces are cognitive, and the overseer is an algorithm that never sleeps.
What the ledger breaks
This transition is not kind, and it is dishonest to pretend otherwise. The job's bundle contained protections that the ledger does not automatically provide.
The volatility tax. A salary smooths luck; a ledger exposes you to it. The top decile of ledger workers — the verified, the specialized, the fluent in self-marketing — will earn multiples of the old salary. The median will ride a roller coaster of feast and famine. Income inequality stops being a statistic about firms and becomes a biography of individuals.
The identity vacuum. "What do you do?" currently resolves to a noun — engineer, teacher, manager. In the ledger economy it resolves to a paragraph: a portfolio of verified outputs, perpetually under revision. That is honest, and it is exhausting. A generation raised on nouns will have to learn to narrate themselves continuously — the psychological equivalent of being perpetually on a job interview.
The insurance gap. This is the crisis that matters. Pensions, healthcare, disability — the entire welfare architecture of the developed world was built on the assumption of stable employers deducting stable contributions from stable payrolls. Unbundle the job and you unbundle the funding mechanism. States that cannot rebuild social insurance around the ledger — portable benefits, contribution systems tied to individuals rather than employers — will watch their safety nets fray into symbolism.
The credential crisis. The ledger's dirty secret is that verification is unevenly distributed. A coder's output is legible to machines; a caregiver's is not. Professions whose value is relational, embodied, or tacit — nursing, teaching, craft — score poorly on ledgers designed by software, and their workers risk being priced by proxies that miss the point. The ledger measures what it can see. What it cannot see, it discounts. That is not a technical bug; it is a civilizational one, and any ledger economy that cannot price care will reproduce the oldest injustice in economics: the work that sustains life being valued least.
Some countries are already sketching the answers: portable benefit accounts that follow the worker across platforms, public ledgers of verified skill that function as a national resume, contribution rules for algorithmic employers. These are not luxuries. They are the plumbing of whatever comes after the job, and the countries that build it first will have an enormous advantage in the only resource that still matters: people willing to take risks.
The dark joke
There is a cruel symmetry in all of this. The twentieth century's great labor achievement was the job — the deal that traded obedience for security. The twenty-first century's great labor technology is undoing the deal, task by task, while keeping the obedience: the algorithmic manager is more demanding than any foreman, and it never buys you a retirement watch.
The optimists say the ledger liberates: no more bosses, no more commutes, no more pretending. The pessimists say it atomizes: no more colleagues, no more pensions, no more cover. Both are right, which is why the transition will feel like both a liberation and a layoff — often in the same week, often to the same person.
The honest question is not whether the job survives. It does not. The question is what we build in the space where the bundle used to be: a ledger economy with portable insurance and real freedom, or a ledger economy with neither. The technology has already voted. The politics have not.
The job was the twentieth century's answer to the question of how a person converts time into a life. It was a good answer — stable, dignified, insurable — and it is being retired. What replaces it will be decided not by the models, which only measure, but by the institutions we build around the measurement. The ledger is coming. Whether it is a cage or a canvas is still, mercifully, up to us.
Premium Essay
Continue reading
This essay continues beyond the opening. Sign in with your free Bureau account to keep reading — Premium unlocks Investigations in full.
Free with a Bureau account · No payment details required to sign in
Loading the discussion…
OpinionInnovation economy & defense tech
We Built the Tech. Now Watch the Giants Fight Over It.
By Tamar Levi · Israel correspondent · Tel Aviv, September 27, 2026
In Tel Aviv, a startup founder will pitch you an idea, take your money, and have a working prototype before your flight home. This is not a metaphor. I have watched it happen between a landing and a takeoff.
The innovation economy here runs on a simple fuel: a small country with no natural resources except the people, and a military that trains nineteen-year-olds to run operations most armies entrust to generals. The pipeline is famous — unit to startup, startup to exit — and the results are everywhere: the chips in your phone, the security in your cloud, the pricing in your insurance. We are the laboratory the world pretends it could take or leave — while quietly building its future inside it.
The numbers embarrass the metaphors, so: six percent of GDP into research and development — the highest share in the OECD — in a country with no oil, no minerals, and half its land desert. Nine thousand startups, more per capita than anywhere on earth. The pipeline is not a metaphor; it is a personnel file: Unit 8200 discharges teenagers who ran operations most armies entrust to generals, and within five years they are founding companies. Mobileye sold to Intel for fifteen billion. Wiz — cloud security, four friends from the same unit — became the largest acquisition in Google’s history. The chips in your phone, the firewall in your bank, the algorithm pricing your insurance: check the labels sometime.
Which brings us to the giants. Washington wants the tech, and offers aid with strings long enough to reach the Knesset. Brussels wants the tech, and offers lectures about everything else. The Gulf — the new friend — wants the tech, and offers actual money with admirable directness. And the East wants the tech too, quietly, through every channel that does not require a press release. Each of these giants arrives believing it is doing us the favor — the market access, the legitimacy, the partnership. Read it the other way around: four bidders, one laboratory, and the laboratory is not the one worried about being ghosted.
The bidders, in order of volume. Washington: nearly four billion a year in military aid, geopolitics’ most generous allowance — every dollar with a review process, a technology-transfer committee, and guidance on which markets we may sell to. Brussels: Horizon research money, partnership frameworks, and running commentary on everything else, delivered with the confidence of a continent that has not won a technology race in thirty years. The Gulf, the new friend: actual money, admirable directness, normalization deals opening markets from Dubai to Riyadh — no lectures, just term sheets. And the East: quiet delegations, unsigned memoranda, interest through channels that need no press releases. Four bidders, one laboratory. We take all four meetings, sell what we can to each, and keep the crown jewels — the source code, the algorithms, the next generation — at home. Not cynicism. Inventory management.
“Washington offers aid with strings. Brussels offers lectures. The Gulf offers cash. The East offers silence. Four bidders, one lab — and the lab is not the nervous one.”
The gas fields taught what the startups taught. Fifteen years ago we imported nearly every drop of energy; then Tamar, then Leviathan — billions of cubic meters beneath the Mediterranean — and suddenly the country that invented drip irrigation was exporting molecules to Egypt and Jordan, with a regional gas forum headquartered in Cairo of all places. Energy independence, drilled and engineered by our own stubbornness. Europe, short of Russian gas, found our phone number. What you develop yourself, nobody can take away — the startups taught it first, and the sea confirmed it.
And defense — the original startup. Iron Dome, David’s Sling, Arrow: systems with combat records no brochure can fake, intercepting rockets by the thousand on live television. Nothing sells like that. Arms exports break records year after year; every air force with a budget wants equipment that has worked, under fire, last Tuesday. The engineers who miniaturized the interceptors miniaturize the medical devices; the algorithms that track a rocket track a tumor. War made us good at technology. Technology may, eventually, make war less necessary. Eventually.
“What you develop yourself, nobody can take away — the startups taught us first, and the sea confirmed it.”
So we do what small, clever countries do: we sell to everyone we can, partner where we must, and keep the crown jewels at home. The gas fields off our coast taught us the same lesson the startups did: what you develop yourself, nobody can take away. The giants will keep circling, each convinced the laboratory exists for their benefit. It does not. It exists for its own — built in a desert, under pressure, while the experts called it impossible. They can lecture, condition, and flatter all they like. The dependency runs one way, and everyone in the room knows which. The chutzpah is not the marketing. It is the business model.
Small, clever countries have one advantage over large ones: they cannot afford illusions. No resources, no hinterland, no margin for error — only people, and the pressure that turns people into engineers. The giants will keep circling, each convinced the laboratory exists for their benefit, offering strings, lectures, cash, or silence. We will keep taking the meetings, signing the deals, and building the next thing in a desert the experts called impossible. The chutzpah is not the marketing. It never was. It is the business model — and business, thankfully, is excellent.