A colossal trove of raw intelligence—11,500 terabytes of intercepted Palestinian phone calls and communications—has been quietly stored and processed on Microsoft Azure cloud servers under a secretive 2021 partnership between the tech giant and Israel’s elite Unit 8200. A joint investigation by The Guardian, +972 Magazine, and Local Call, based on internal documents and interviews with serving and former intelligence officers, reveals a military surveillance apparatus that has blurred the lines between civilian and military technology, with profound consequences for privacy, human rights, and corporate accountability.
The system, which went live in 2022, gave Unit 8200—often described as Israel’s equivalent of the NSA—a customized, segregated area on Microsoft’s cloud platform to store, analyze, and replay “a million calls an hour,” according to intelligence sources. This near-limitless capacity allowed the unit to harvest communications from Palestinians not only in the occupied West Bank and Gaza but also abroad, effectively turning ordinary civilians into subjects of mass surveillance.
Microsoft engineers worked alongside the agency to embed advanced security features, transferring up to 70% of the unit’s sensitive data to Azure. The servers hosting this data are located in Microsoft facilities in the Netherlands and Ireland, the investigation found. While Microsoft maintains that CEO Satya Nadella was unaware that the cloud services would be used for intercepting civilian communications, multiple sources familiar with the project told investigators that key Microsoft staff understood the system’s sweeping capabilities.
The partnership was sealed in a 2021 meeting between Nadella and Yossi Sariel, then the commander of Unit 8200. Internal recordings reviewed by The Guardian show that the discussion avoided explicit mention of mass surveillance, instead referring cryptically to “sensitive workloads.” Yet, documents later surfaced detailing how Azure would house raw intelligence and audio files. “This is not a simple storage solution,” one source from the unit said. “It’s an analytical engine that feeds into operational targeting.”
Three sources from Unit 8200 confirmed that data stored on Azure directly facilitated the preparation of deadly airstrikes and “shaped” military operations in Gaza, where more than 60,000 people have been killed since October 2023, including over 18,000 children. In the West Bank, sources said, Azure-hosted intelligence was used to blackmail individuals, place them in administrative detention, or retroactively justify killings. The system’s playback feature meant that if a person became a target later, their past conversations—often years old—could be mined for incriminating fragments.
This revelation has ignited a firestorm of criticism from human rights groups, technology ethicists, and even Microsoft’s own employees and shareholders. In April 2025, during Microsoft’s 50th Anniversary Copilot Event, an employee named Ibtihal Aboussad stood up and accused the company of having “blood on your hands,” demanding an immediate end to all contracts with the Israeli Ministry of Defense. The protest underscored the deepening moral fissures inside one of the world’s most powerful technology companies.
Meanwhile, a coalition of more than 60 shareholders, led by the Religious of the Sacred Heart of Mary, filed a resolution calling for an independent report on the effectiveness of Microsoft’s human rights due diligence processes. “We are deeply disturbed by Microsoft’s apparent ineffective due diligence with regard to human rights violations by its customers against the Palestinian people,” the group stated.
Microsoft’s public response has been a study in careful parsing. The company acknowledged providing Israel’s military with cloud computing services and disclosed limited emergency support after the October 2023 Hamas attacks. However, following an internal review, it stated that it found “no evidence to date that Microsoft’s Azure and AI technologies have been used to target or harm people in the conflict in Gaza.” A spokesperson added that the company was never aware of “the surveillance of civilians or collection of their cellphone conversations using Microsoft’s services.”
But critics point to the deep integration of Microsoft engineers into the project. According to the investigation, these engineers not only built enhanced security layers but also helped migrate and manage the sensitive data, which eventually grew to over 11,500 terabytes. Microsoft staff were reportedly instructed never to mention Unit 8200 by name, using code words instead. “If you’re building a custom, fortified section of your cloud specifically for a military intelligence unit, at what point does plausible deniability become complicity?” asked one technology ethics researcher not involved in the investigation.
The IDF, for its part, issued a statement that appears to contradict the documentary evidence: “We appreciate Microsoft’s support to protect our cybersecurity. We confirm that Microsoft is not and has not been working with the IDF on the storage or processing of data.” This denial, however, is flatly contradicted by the internal files, which include detailed descriptions of the cloud infrastructure and its use.
The ethical and legal implications are manifold. By providing the pipeline for mass surveillance, Microsoft risks being seen as facilitating potential violations of the right to privacy and other fundamental freedoms under international law. Indiscriminate collection and storage of an occupied population’s communications, without individualized suspicion, raises potential war crime concerns, especially when the data is used to guide lethal strikes. The fact that the data sits on commercial cloud infrastructure, rather than sovereign military servers, also obfuscates lines of accountability: which legal and regulatory frameworks apply when a U.S.-based corporation processes the phone calls of Palestinians through European data centers for an Israeli military client?
This case is not an isolated incident but part of a growing trend where cloud giants become entangled in global conflicts. Amazon, Google, and Microsoft have all vied for lucrative defense and intelligence contracts, often touting the speed and scalability of their AI tools. Project Nimbus, a $1.2 billion contract with Google and Amazon, similarly provides cloud services to the Israeli government, raising parallel ethical alarms. As artificial intelligence becomes more embedded in military decision-making, the risk of “black-box” targeting, where algorithms flag potential threats with little human oversight, grows exponentially.
Internal dissent at Microsoft is a bellwether of broader tech worker unease. The protest at the Copilot event, though quickly quashed by security, resonated online and in activist circles. It follows years of employee activism at Google, Amazon, and other firms over projects like Project Maven and law enforcement facial recognition. “We didn’t sign up to build tools of occupation,” said one Microsoft contractor, speaking anonymously. “But when you look at the layers of secrecy, it’s hard to know what you’re really contributing to.”
Human rights organizations, including Amnesty International and Human Rights Watch, have long warned about the weaponization of big data in asymmetrical conflicts. The Azure-hosted surveillance system, they argue, is a textbook example of how commercial technology can amplify state power to an unprecedented degree, enabling the cataloguing of an entire population’s social fabric. Once collected, such data rarely disappears; it can be repurposed for profiling, propaganda, or political manipulation for decades.
Legal experts also raise the prospect of corporate liability under laws like the EU’s new Artificial Intelligence Act, which places strict requirements on AI systems used for biometric categorization or emotion recognition in law enforcement and military contexts. If Azure’s AI services were used to analyze voice patterns or predict behavior, Microsoft could face substantial regulatory fines and reputational damage in Europe.
Yet, the company’s posture remains one of cautious neutrality, emphasizing its “standard commercial relationship” with the Israeli Ministry of Defense and its commitment to “ethical principles.” The dissonance between these statements and the on-the-ground reality, as documented by the investigation, has only deepened the controversy. Nadella personally has not addressed the specifics of the Azure-Unit 8200 collaboration, and the company’s fiscal disclosures treat defense revenue as part of a broader commercial cloud segment, making it difficult for investors to assess risk.
The U.S. government, too, is in a delicate position. Microsoft is a major federal contractor and a linchpin of national cybersecurity strategy. Any formal investigation into the Azure-Unit 8200 partnership could strain U.S.-Israeli intelligence cooperation, which has long been a cornerstone of regional policy. Congress members from progressive factions have called for hearings, but movement has been slow.
As the war in Gaza continues and tensions in the West Bank escalate, the role of technology in the conflict will only come under sharper focus. The case of Microsoft and Unit 8200 is a pivotal test of whether the tech industry’s voluntary human rights assessments and ethical AI pledges amount to anything more than public relations. “Transparency is the absolute minimum,” said one shareholder activist. “Without it, we’re just enabling atrocities in the dark.”
The investigation’s findings leave little ambiguity: Microsoft’s cloud was an essential piece of a surveillance apparatus that many legal scholars argue violates international humanitarian law. Whether the company chooses to confront these findings head-on or retreat further behind contractual opacity will shape not only its legacy but also the future rules of engagement for technology in warfare.
For now, the 11,500 terabytes of recorded Palestinian lives sit on servers in the Netherlands and Ireland, a digital archive of occupation that grows by the hour. The question remains: who will take responsibility for what happens next?