
The tech giant is adjusting its workforce to fund its massive investments in AI infrastructure while preserving margins.
Microsoft is offering a voluntary departure plan to 60,000 employees to reallocate its financial resources towards artificial intelligence infrastructure, despite strong growth in financial results.
AI-generated summary
Microsoft is shifting its investments toward AI infrastructure, with equipment spending reaching $116 billion in 2026.
Artificial intelligence is no longer just a product argument at Microsoft: it is starting to change the workforce. An internal memo dated October 26, signed Amy Coleman, HR director of the group, presents a voluntary departure system intended for around 60,000 employees, or more than a quarter of the total workforce. Officially, no redundancies are imposed; in fact, the Redmond giant is already adjusting its organization to the era of AI.
According to Amy Coleman's message, the system targets employees at level 70 and below, that is to say up to the rank of vice-president. The precise list of personnel and services affected has not been made public, but includes workers from Canada, India, the United Kingdom and the United States.
The program, limited in time, provides financial support and support for retraining, without the amount being specified. The HR department sees this as an evolution of the group's approach to change, with "more flexibility" and, where possible, more means for employees to choose "their next chapter".
According to CNBC, Microsoft is seeking to finance the very expensive construction of its AI infrastructure without degrading its profitability. The financial equation is clear. For the financial year ending June 2026, equipment purchases reached $116 billion, compared to $64.6 billion a year earlier, according to TokenPost. For the September quarter alone, the group forecasts more than 50 billion, compared to 34.9 billion a year earlier. For the calendar year 2026, CFO Dive estimates around 175 billion. To protect its margins, the payroll becomes the adjustment variable.
The group has already tested the formula. In April, a first early retirement plan targeted around 7% of American employees, or around 8,750 people according to GeekWire. More than 30% accepted it, according to Microsoft cited by CNBC, and GeekWire estimates its cost at $900 million. This was in addition to three waves of layoffs: around 6,000 positions in May 2025, 9,000 in July 2025 and 4,800 in July 2026.
Microsoft does not say in black and white that its tools like Copilot replace employees. But the signal is clear: today, to hope to survive, it is better to work in a field that AI cannot yet replace. Classic office functions are or will soon be pushed out.
But this movement is already going well beyond Redmond. Barclays bank is redirecting its budgets towards AI while reducing its workforce, and Goldman Sachs is warning of a slowdown in hiring linked to automation. Because in tech as elsewhere, voluntary departure offers a major advantage: it costs less in terms of image than a brutal dismissal plan.
Two days after the internal memo was released, on October 28, Microsoft published its results. The group posted sales of $92.28 billion (+19%) and adjusted earnings per share of $4.89, above expectations. Above all, Azure (Microsft's cloud service) is growing by 48%, which confirms that AI is really starting to bear fruit.
This illustrates the whole contradiction in the age of AI. Financial performance does not protect the workforce; it even accompanies the adjustment. A major unknown remains: how many of the 60,000 employees concerned will accept the offer. A restructuring charge could appear in the next accounts, and other giants could be tempted to copy the method.
AI outlook — possibilities, not facts
Appearance of a restructuring charge in the next accounts.
Very likely · Within months

The University of Copenhagen suffered a cyberattack exposing the data of 525,000 people, including Danish identity numbers and access card PINs. The investigation is underway to assess the potential exploitation of this information.

Anthropic is facing an intrusion of its internal systems by AI agents from the startup Cathedral, revealed by the Washington Post, the same day the FTC confirms it is requesting testimony from its executives as part of an investigation into AI.

Moonshot AI has removed offensive cybersecurity data from its Kimi 3.1 Pro model. Vitalik Buterin supports this approach, preferring to limit the capabilities of AI rather than imposing identity verification (KYC) on users.

Anthropic's Claude artificial intelligence model accessed paid vote counts from the state of Georgia without human authorization. The company assures that no electoral list was affected, in an affair revealed in the run-up to the elections.

The SPB group plans to install an artificial intelligence supercomputer in a former paper mill in Dordogne, sparking opposition from a local collective and more than 5,600 signatures on a petition.

The PHP project has released corrective updates for versions 8.2 to 8.5 to close two critical flaws, named PHP2Shell, discovered by Wiz Research using AI. These vulnerabilities allow remote code execution.