On April 16, 2019, in a federal courtroom in San Diego, Apple’s lawyer Ruffin Cordell stood up to open a trial that was scheduled to run five weeks. He told the jury that Qualcomm had used its power over the modem chips that connect phones to the network “to set unfair prices,”“stifle competition” and “impose their will on their customers.”
The two companies had been fighting for two years. Apple had stopped buying Qualcomm’s modems, and its iPhones now ran on chips from a single other supplier, Intel. Tim Cook was on the witness list.
Several hours after opening statements began, the trial was over.
Apple and Qualcomm announced that they were dropping every lawsuit between them, worldwide. Apple would make a payment to Qualcomm, sign a six-year license, and buy Qualcomm chips again. The company that had just accused its supplier of bullying customers had agreed to become one again, on the day the jury first heard the case.
The press release that same day
A second press release, dated the same day, points to the answer. Intel announced that it was leaving the 5G smartphone modem business and did not expect to launch the 5G phone modems “originally planned for launches in 2020.”
Samsung planned 5G phones that year. Intel, Apple’s only modem supplier, had said it would not have a 5G chip ready until 2020, which, Reuters reported, could have pushed a 5G iPhone into 2021. Qualcomm was already preparing its second-generation 5G chip.
The order of events is still debated. A week later, Intel’s chief executive Bob Swan told The Wall Street Journal that the settlement came first: “In light of the announcement of Apple and Qualcomm, we assessed the prospects for us to make money while delivering this technology for smartphones and concluded at the time that we just didn’t see a path.” If Swan is right, Intel’s exit followed the deal, so it can’t be counted as what Apple faced without one. Nor was Intel the only possible supplier: Apple had told the court earlier that year that it had held talks with MediaTek and Samsung about modems. The reporting supports something narrower. Apple was not out of options, but the supplier already in its phones was late on 5G.
How hard you can push in a deal depends on what you’d do without it.
The number nobody says out loud
Roger Fisher and William Ury of the Harvard Negotiation Project gave that alternative a name in Getting to Yes (1981): your BATNA, your Best Alternative To a Negotiated Agreement. “The reason you negotiate,” they wrote, in a passage Sebenius quotes, “is to produce something better than the results you can obtain without negotiating.” The BATNA “is the standard against which any proposed agreement should be measured.”
Two things follow. You should turn down any deal that is worse than your alternative, however hard the talks have been. And the other side does the same sum about you. Read in Fisher and Ury’s terms, the arguments about royalties in that courtroom mattered less than what Apple would face if it walked out of it.
A renter who must be out by Friday, with nowhere easy to go, has a harder time haggling over the new flat. A renter who can stay where she is can walk out of the viewing.
The Harvard negotiation scholar James Sebenius warns against treating the BATNA as a last resort that only matters when talks collapse. You should assess it, and improve it, before you sit down. He describes a “no to re-set”: stepping away from the table to improve your options, then coming back.
Apple’s later moves fit that description. In July 2019 it agreed to buy most of Intel’s smartphone modem business for $1 billion, with about 2,200 Intel employees. In February 2025 it introduced the iPhone 16e with C1, “the first modem designed by Apple.” Apple has not said that leverage over Qualcomm was the reason. But it now has a walk-away option it did not have in 2019.
Before the supplier calls back
Say you run a small bakery, and your flour mill writes to say prices go up 15% next month. The easy move is to phone them and argue.
The BATNA move is to spend the afternoon on other calls first. One mill quotes 5% more, delivered weekly instead of twice a week, so you would need more storage. Now you know your floor: a deal from your current mill that is worse than switching is worse than no deal. You may never mention the other mill. Sebenius notes that talks about interests often go better with the alternative kept “in the background,” not “continually brandished.” Knowing it may change what you are willing to accept.
The mill, of course, has a BATNA too: other bakeries. Your next question is how many.
Handing over the walk-away point
Delegating a negotiation means handing over your walk-away point, and a 2025 study by researchers from Stanford, MIT and elsewhere shows what can go missing.
They had nine language models play buyer and seller against each other over products drawn from a set of 100 real ones, from electronics to cars and homes. Each buyer agent got a budget and was told not to reveal it unless necessary. Many revealed it anyway when the seller asked; the seller then anchored on that number, and in some deals buyers paid more than the retail price. When budgets were generous, buyers tended to accept the first price under budget instead of pushing for a better one. Weaker models such as GPT-3.5 accepted deals above budget in over 10% of cases, according to the paper, more often when the budget was tight. Newer models such as o3 and GPT-4.1 generally kept to it.
The parallel is real but partial. A budget is a ceiling. It tells the agent when to refuse, and it says nothing about what you would do instead: another seller, a cheaper model, waiting a month. In this study, a generous budget worked more like a target than a limit. This was a simulation with fixed prompts, not a deployed product, and an agent can be told your actual alternative. A person usually knows what is waiting outside the room. In this study’s setup, the agents knew only what their prompts told them. An agent working for you needs to be given your alternative, or be able to look it up.