On 18 January 2022 Microsoft announced it would buy Activision Blizzard for 95 dollars a share in cash, about 68.7 billion dollars, the largest acquisition the games industry had seen. The deal closed on 13 October 2023. In the twenty-one months between, Microsoft negotiated with three regulators and got three different answers.
The European Commission cleared the deal on 15 May 2023 after Microsoft committed to let consumers in the European Economic Area stream Activision games through any cloud service, free of charge, for ten years. In the United States the Federal Trade Commission went to court to stop it; on 10 July 2023 a federal judge in San Francisco denied the agency's request for a preliminary injunction, finding it had not shown it was likely to win. In the United Kingdom the Competition and Markets Authority had already said no. On 26 April 2023 it blocked the acquisition. Not over consoles, a concern it had dropped in March, but over cloud gaming, where it put Microsoft's share at 60 to 70 per cent and rejected Microsoft's offer of ten-year licensing promises as a remedy it would have to police itself.
Microsoft did not appeal its way out. In August it restructured the deal: the cloud streaming rights to Activision's games outside the EEA, covering titles released over the following fifteen years, would go to Ubisoft instead. The CMA called the revised transaction "substantially different", opened a fresh investigation, and on 13 October cleared it. Microsoft gave undertakings that made the Ubisoft terms enforceable by the regulator, and closed the same day.
Four moves in that sequence are worth taking apart, because the counterpart in each of them is one you will meet: someone who cannot be paid more, only satisfied.
1. A regulator is a counterpart with a mandate, not a price
Most negotiation advice assumes the other side wants more of something you have. A regulator does not. The CMA could not accept a higher price, a side payment or a favour. What it needed was a decision it could sign: one that passed its statutory test, survived an appeal and read well in the press.
Microsoft's first remedy missed that. It offered more promise: ten years of Call of Duty on rival platforms, more titles, more platforms. The CMA's answer in April was that a set of obligations regulating Microsoft's behaviour for a decade would turn the authority into a supervisor, and it preferred to let the market work. The offer was bigger. It was not the kind of offer the counterpart could say yes to.
The lesson travels. Before you make a concession, work out what the person across the table has to be able to say afterwards, and to whom. A procurement officer has to show a process. A credit committee has to show a ratio. A compliance team has to show a document. Give them that sentence, and the number matters less than you think.
2. Find the objection that actually blocks, and solve only that
By April the CMA had narrowed its case to one thing. Console competition had been off the table since March. The block rested on cloud gaming alone: a market the regulator thought was about to take off, with Microsoft already the largest player in it.
Microsoft's August move answered that objection and nothing else. It did not promise again to behave. It removed the asset the objection was about. Cloud streaming rights outside the EEA went to a third party, Ubisoft, which could license them on to any cloud provider, including back to Microsoft. Structural, narrow, checkable.
Most refusals contain one load-bearing objection and several decorative ones. The decorative ones get argued because they are easier. When you hear "no", your first job is to find out which part of your proposal the no is attached to. Ask directly: "If the cloud piece were gone, would the rest pass?" Then answer that piece with something the other side can verify without having to trust you.
A no from someone with a mandate is a description of the mandate. Read it before you reply.
3. Concessions to third parties are currency
On 21 February 2023, in Brussels, Microsoft announced two ten-year agreements: one with Nintendo to bring Call of Duty to its players on the same day as on Xbox, one with Nvidia to put Xbox PC games and Activision titles on the GeForce Now streaming service. Nvidia publicly backed the acquisition afterwards.
Neither deal was with a regulator. Both were with parties the regulators were supposed to protect. That is the point. A promise to a regulator is a promise; a signed ten-year contract with a competitor is a fact, enforceable by someone with an interest in enforcing it. It changes what the regulator sees when it looks at the market, and it changes who turns up to say the deal is fine.
When the objection to your proposal is "what this does to others", go to the others. Make the commitment to them, binding, and let them tell the counterpart. Their voice is worth more at that table than yours.
4. The restructure gave the other side a way to say yes
By August the CMA had blocked the deal, defended the block, and confirmed in writing that the original transaction could not proceed. For the authority to clear that same transaction would have meant admitting in public that it had been wrong in April, as the only major regulator still standing in the way after the FTC's loss in July.
Microsoft never asked it to. It put a different transaction on the table, and the CMA could say, truthfully, that the first deal stayed blocked and this new one passed. Its own press release called the concession a "gamechanger". Same companies, same money, a different question, and so a different answer without a reversal.
Once someone has said no in public, the cost of yes includes their face. You can pay that cost for them. Change the object enough that their yes is to a new question, and give them the words: "What you rejected was X. This is not X."
At your own table
You will not buy a games publisher. You will sit opposite a procurement department, a landlord's lawyer, a bank, a works council, a compliance officer at a customer you want to sign. People with a mandate, who cannot be paid more and are not trying to win. The Microsoft case gives you the moves.
- Ask what the other side has to be able to say afterwards, and to whom. Build your offer around that sentence.
- When you hear no, find the one objection it rests on. Solve that one with something checkable, not with more promises.
- If the worry is about third parties, make binding commitments to the third parties and let them speak.
- If the other side has said no in public, change the question rather than asking them to reverse the answer.
The hard part is not the list. It is hearing a flat refusal and asking "which part?" instead of arguing, and the only safe place to drill that reflex is against a counterpart that is not real. On Pacct that counterpart is an AI with a mandate it will not sell, and the debrief tells you where you pushed on price when you should have worked on the mandate.
