| **BioNTech (Mainz, Germany) and Pfizer (New York, US), January–December 2020 | A smaller biotechnology company, a global pharmaceutical company, and about ten months from program initiation to the first temporary authorization of an mRNA vaccine.** |
BioNTech had worked on mRNA technologies since 2008, primarily for cancer immunotherapy, and had also begun applying its platforms to infectious disease. According to the account developed with BioNTech leaders in Joe Miller’s The Vaccine, co-founder and CEO Uğur Şahin became concerned in January 2020 that the cluster of pneumonia cases in Wuhan could develop into a pandemic. He and BioNTech co-founder Özlem Türeci spent the weekend of January 24–26 considering whether the company’s mRNA platforms could be redirected toward a vaccine.
BioNTech’s own retrospective timeline dates the initiation of its COVID-19 vaccine program, code-named Project Lightspeed, to January 27, 2020. The more detailed account of Şahin’s presentation to the executive team and the team’s internal deliberation comes from the insider history, not from a contemporaneous public announcement. At that point, the World Health Organization had not yet characterized COVID-19 as a pandemic, no mRNA vaccine had been authorized, and BioNTech had no marketed product. Committing people and capital to the program therefore involved substantial scientific and financial uncertainty.
The company proceeded with Project Lightspeed.
BioNTech advanced multiple mRNA vaccine candidates in parallel, but it did not by itself have Pfizer’s global clinical-development, regulatory, manufacturing, and distribution infrastructure. The companies already had a working relationship through an mRNA influenza-vaccine collaboration begun in 2018.
The chronology of their COVID-19 partnership matters. On March 17, 2020, Pfizer and BioNTech publicly announced a letter of intent to co-develop and distribute the BNT162 program outside China. They also said they had executed a Material Transfer and Collaboration Agreement so work could begin immediately, while financial terms and the detailed arrangements for development, manufacturing, and potential commercialization were still being finalized. On April 9, the companies announced those fuller terms.
The teams therefore did not simply exchange proprietary work with no agreement in place. They began intensive scientific collaboration before every commercial term had been settled, but with a prior relationship and interim legal scaffolding already in place. That distinction makes this a case about staged or bounded trust rather than blind trust.
Three features of the collaboration are especially useful for analyzing interorganizational teamwork.
Information sharing before all commercial terms were final. According to Miller’s insider account, Şahin told BioNTech staff to “share everything” with their Pfizer counterparts after the March letter of intent. The instruction described an unusually broad and rapid scientific exchange while fuller terms were still being negotiated. It did not mean that the exchange lacked a legal or governance framework: the March 17 announcement documents the material-transfer and collaboration agreement already in place. The teaching question is therefore not whether trust replaced contracting, but how a prior relationship, provisional agreements, shared urgency, and continuing negotiation made faster exchange possible.
Overlapping development work. Project Lightspeed overlapped activities that might otherwise have occurred more sequentially. The German Phase 1/2 trial began on April 23, the US trial began on May 4, and the pivotal Phase 2/3 trial began on July 27. Manufacturing capacity was expanded “at risk” before efficacy and authorization were known. Overlap exposed the partners to financial loss if a candidate failed, but it did not remove the independent data-monitoring, regulatory, safety, or efficacy requirements governing the trials. This distinction between accepting financial risk and weakening evidentiary standards is central to evaluating the acceleration responsibly.
Asymmetric capabilities with negotiated risk sharing. Pfizer and BioNTech brought different resources and faced different exposures. The April 9 agreement helps make the allocation concrete: Pfizer committed $185 million in upfront payments, including an approximately $113 million equity investment; the companies agreed to share development costs equally; and Pfizer would initially fund those costs, with BioNTech repaying its half during commercialization. Pfizer contributed global development, regulatory, manufacturing, and distribution capabilities, while BioNTech contributed the mRNA candidates and platform expertise. Trust operated alongside a negotiated allocation of money, responsibilities, and risk.
Insider accounts place the independent data-monitoring committee’s review of the first interim efficacy analysis for the BNT162b2 Phase 3 trial on November 8, 2020. Pfizer and BioNTech announced the next morning that the candidate had shown efficacy above 90 percent in that interim analysis. The announcement also made clear that the trial would continue, that additional safety and efficacy data were still required, and that a US Emergency Use Authorization request could not be made until the specified safety milestone had been reached.
Insider accounts describe intensive coordination between the companies before the November 9 announcement. That coordination rested on working relationships and decision processes developed over the preceding months, but the favorable interim result should not be treated as proof that every earlier decision was sound. A good analysis asks whether the process was defensible given the information and safeguards available at the time, including the possibility that the candidate could have failed.
The final analysis reported 95 percent efficacy on November 18. The United Kingdom granted the first temporary authorization for emergency supply on December 2, and the US Food and Drug Administration issued an Emergency Use Authorization on December 11.
Evaluate the March information-sharing decision prospectively. Based only on what the partners knew and had put in place by March 17—not on the vaccine’s eventual success—which features made accelerated sharing more defensible? Which risks remained? How did the prior influenza collaboration, the letter of intent, and the material-transfer agreement change the analysis?
Trust and contracts. The teams began work before every financial and commercial term was final, but they did not work without legal scaffolding. What can a provisional agreement accomplish, and what still depends on professional trust, shared norms, and day-to-day behavior? When would staged information sharing be preferable to either complete withholding or unrestricted disclosure?
Asymmetric capabilities and exposure. What did the April 9 financing and cost-sharing arrangements do to manage asymmetry between the organizations? Which asymmetries could a contract address, and which still had to be managed through governance and relationships?
Speed without evidentiary shortcuts. Distinguish overlapping trials and at-risk manufacturing from lowering safety or efficacy standards. What checks—independent monitoring, predefined endpoints, regulatory review, decision gates—allow teams to work in parallel without treating urgency as permission to skip verification?
Compare the decision processes without equating them. In the Challenger teleconference, Morton Thiokol requested about five minutes off the communication loop; testimony to the Rogers Commission says the caucus actually lasted approximately 30–35 minutes before management returned with a launch recommendation. The vaccine collaboration unfolded over months and included independent monitoring and regulatory decision points. What, if anything, can still be learned by comparing how the two settings handled dissent, expertise, uncertainty, and decision authority? What important differences make a simple comparison misleading?
Check for outcome bias. Imagine that BNT162b2 had failed despite the same process, or that Challenger had launched without an accident despite the same decision process. Which judgments about the quality of each process would remain unchanged? Which judgments, if any, should change?