The latest salvo in the decades-long conflict between Iran and Israel lit up the predawn sky over Tehran on Saturday. Israeli aircraft encountered little resistance as they struck military targets in retaliation for an Iranian attack earlier this month. Although Iran appeared to downplay its impact, the strike was Israel’s largest ever against the Islamic Republic. It raised not only the specter of full-scale war but also a prospect that experts told me has become much more conceivable in recent weeks: the emergence of Iran as a nuclear-armed state.

Think of Iran’s defenses as a stool with three legs. Two of them have suddenly gone wobbly. The first is Iran’s regional proxy network. This includes, most notably, Hamas in Gaza and Hezbollah in Lebanon, both of which Israel has dismantled through air strikes, incursions, and high-profile assassinations. Israel has even gone after Iran’s top military commanders. The second is an arsenal of missiles and drones, which Iran used to directly attack Israel for the first time in April, and then again this month. Not only did the strikes prove ineffective—Israeli and U.S. defenses largely thwarted them—but they also failed to deter Israel from continuing to hack away at the first leg and strike back as it did over the weekend.

That leaves the third leg: the Iranian nuclear program. Now that Israel has demonstrated its superiority over Iran’s proxies and conventional weapons—and degraded both in the process—Supreme Leader Ayatollah Ali Khamenei may decide to pursue a bomb in a risky attempt to salvage some measure of national security. He won’t have far to go. The program has made major advances since 2018, when the U.S. withdrew from its multilateral nuclear agreement with the regime, which now has enough near-weapons-grade uranium to produce several bombs, according to the International Atomic Energy Agency (IAEA). This already gives the country considerable leverage, but “there is a risk Khamenei decides that in this environment, a nuclear threshold won’t cut it, and Iran needs nuclear weapons,” Eric Brewer, a nonproliferation expert at the Nuclear Threat Initiative, told me.

Although Brewer and other experts I spoke with did not predict that Iran will go nuclear in the near term, they agreed that it is likelier than ever before. If Iran were to acquire nuclear weapons during the metastasizing conflict in the Middle East, it could become the first country to do so while at war since the United States in 1945. But Iran also has many ways to wield its nuclear program that stop short of getting a weapon, injecting further peril into an already volatile new nuclear age.

In recent years, current and former Iranian officials have insisted that the country is either already able to build a nuclear bomb or very close to that point. In the past month, as Iran awaited the retaliation that came on Saturday, its pronouncements got more pointed. Although the regime still denies that it’s seeking a weapon, a senior adviser to Khamenei warned that any Israeli strikes on its nuclear sites—which were spared over the weekend—could alter the nation’s “nuclear strategic policies.” That same week, a group of 39 Iranian lawmakers urged the Supreme National Security Council to eliminate its formal ban on the production of nuclear weapons.

[Read: What if Iran already has the bomb?]

The latest rhetoric in official circles could be a response to Iran’s shifting public discourse. Nicole Grajewski, an expert on Iranian nuclear decision making at the Carnegie Endowment for International Peace, told me that Israel’s assassination of the Hezbollah leader Hassan Nasrallah last month seems to have piqued Iranian public interest in their country’s nuclear program. She’s noticed a greater number of Iranian commentators on Telegram discussing Tehran’s nuclear capabilities.

Mohammad Ayatollahi Tabaar, a Texas A&M professor who studies nuclear statecraft and Iranian politics, has also observed this shift in Iranian public and elite sentiment. But he traces it back further, to America’s exit from the Iran nuclear deal and then, two years later, its assassination of the Iranian general Qassem Soleimani. When the deal took effect in 2015, Tabaar told me, the regime was responsive to public pressure to limit its nuclear program and improve relations with the United States. Discussing the nuclear-weapons option was, as he put it, “taboo.” But in recent weeks, he said, he’s seen “a lively debate” on social media about whether or not to pursue a bomb, even among critics of the regime outside the country.

“There is this realization that, yes, the regime and the [Islamic Revolutionary Guard Corps] are repressive, but we live in this neighborhood and maybe we need to have” nuclear weapons, Tabaar told me before the latest strike.

That decision belongs to Khamenei, but the increased public interest that Tabaar has observed creates an opening for Iranian leaders to advance the country’s nuclear program. As Tabaar noted, such decisions are often informed by the views of elites and by the regime’s “fear of popular revolt.”

Still, neither Grajewski nor Tabaar anticipates that the regime will immediately seek a bomb. Iran could instead use its near-nuclear status to its advantage, including by escalating threats to go nuclear, announcing progress in uranium enrichment, rebuffing international oversight, or exiting the Nuclear Non-Proliferation Treaty. In addition, Iran could try to reinforce the other legs of its security—by working with partners such as Russia and North Korea to upgrade its conventional military capabilities, and by bolstering proxy groups such as the Houthis in Yemen while seeking to rebuild Hamas and Hezbollah.

But strengthening these other legs could take years, and Israel appears poised to press its military advantage. That leaves a crucial question for Iran’s leaders: Is the country’s nuclear-threshold capability enough of a deterrent?

If they decide to cross the threshold and go nuclear, Iranian leaders know that their adversaries will likely detect their efforts and try to intervene, potentially undermining the very security Tehran may be seeking. The latest U.S. estimates indicate that Iran might require only a week or two to enrich uranium to weapons-grade. But concealing such a move from IAEA inspectors without kicking them out of the country would be challenging. And Iran could need more than a year—or at least several months, by some estimates—to convert its uranium into a usable weapon.

Those months constitute “a pretty big window of vulnerability” in which “Israel or the United States could disrupt Iran’s work to build a nuclear weapon, including through military action,” Brewer explained. So he thinks it’s “unlikely” that the supreme leader will wake up one morning and declare, “Damn the torpedoes. All hands on deck. We’re going to weapons-grade today.”

A more plausible outcome, Brewer and Grajewski believe, is that Iran covertly resumes the research on weaponizing fissile material that it halted in 2003. The goal would be to “shorten the window of vulnerability” between amassing weapons-grade uranium, putting it into a nuclear device, and fashioning a deliverable weapon, Brewer told me. This weaponization work is more difficult (though not impossible) to spot than uranium enrichment, at least at declared facilities still monitored by the IAEA. International inspectors retain access to facilities containing fissile material, but Iran has reduced the frequency of inspections since 2018, when the U.S. exited the nuclear deal. The regime has also ended IAEA monitoring of other sites related to its nuclear program, raising the possibility that it has moved some centrifuges to undeclared facilities. Nevertheless, U.S. officials said this month that they could probably detect any decision to build nuclear weapons soon after Iranian leaders make it.

[Phillips Payson O’Brien: The growing incentive to go nuclear]

American officials often speak about whether Iran’s leaders have “made the decision” to attain nuclear weapons, but Tabaar argued that Tehran’s calculations don’t work that way. Think of a dimmer, not a light switch: Iran is “making sure all components are there to preserve its option to develop nuclear weapons, gradually more and more.” Tabaar added, however, that there are “two very extreme scenarios” in which he could imagine Iranian leaders suddenly making the call to flip the nuclear switch. The first is a “window of opportunity” in which Iran’s enemies are distracted by, say, a major conflict elsewhere in the world. The second is “a window of threat” in which Iranian leaders fear that their adversaries are about to unleash a massive bombing campaign that could destroy the country or regime.

Brewer posited one other wild-card scenario: The supreme leader might proceed with weapons-grade enrichment at declared facilities if he assumes that he can achieve it before Israel or the U.S. has a chance to destroy those facilities, thereby establishing some measure of deterrence. “That would be a very, very risky gamble,” Brewer said—particularly if Israel learns of Tehran’s decision in time to unleash preemptive strikes. Additional enrichment might not ward off an Israeli or American attack anyway. Although 90 percent enrichment is typically considered the level required for weaponization, experts believe that Iran might already be able to use its current stock of 60-percent-enriched uranium to make a bomb. Anything higher wouldn’t necessarily establish greater deterrence.

But, as Brewer has noted, history offers several examples of regional crises prompting states to “break out,” or race for a bomb. Shortly before the 1967 Six-Day War, Israel reportedly rushed to assemble nuclear devices out of concerns about possible Egyptian strikes on its nuclear facilities. Amid tensions with India over the disputed territory of Kashmir, Pakistan is believed to have begun building nuclear weapons by 1990. That same year, following Iraq’s invasion of Kuwait, Saddam Hussein ordered an impractical (and unsuccessful) effort to quickly build a nuclear weapon. “I can give you lots of really good reasons why breaking out would be a terrible decision by the supreme leader,” Brewer told me. “I can also give you lots of reasons why the crash nuclear-weapons program in Iraq was a terrible decision. But [the Iraqis] still made it.”

I asked my Atlantic Council colleague Danny Citrinowicz, who from 2013 to 2016 led the Israeli military’s analysis of Iranian strategy, whether Iran is more likely to become a nuclear-weapons state today than it was at any point in the many years that he’s monitored its nuclear program. He didn’t hesitate: “Definitely.”

Citrinowicz broke down that answer into relative probabilities. He pegged the chances of Iran “storming” to a bomb—by, for example, detonating a nuclear device for demonstration purposes—at 10 percent, the highest he’s ever assessed it. Before Hamas’s October 7 terrorist attack against Israel, he would have said “close to zero.” He assigned a 30 percent probability to the scenario of Iran enriching uranium to weapons-grade, though perhaps only a minimal amount to show off its capabilities.

To my surprise, the scenario he deemed most likely—at 60 percent—was Iran pursuing negotiations on a new nuclear deal with the United States and other world powers. Citrinowicz could envision Kamala Harris and even Donald Trump—perhaps reprising the openness to nuclear diplomacy that he displayed with North Korea, despite his typically hard-line stance on Iran—being amenable to such talks after the U.S. presidential election. A diplomatic agreement would probably inhibit Iran from acquiring nuclear weapons, but it could also provide the country with economic relief. As an added benefit, a deal with Washington might serve as a wedge between the United States and Israel, the latter of which would likely oppose the agreement. Israel would be less inclined to strike Iranian nuclear facilities if it couldn’t count on U.S. support, or at least it would be less capable of penetrating their heavy fortifications without help from America’s arsenal.

[Read: The unraveling of Trump’s North Korea policy]

Still, there are many reasons to be skeptical about the possibility of a new nuclear deal with Iran. Russia and China, both parties to the 2o15 pact, are far more hostile to the United States today than they were then. Khamenei has expressed a general willingness to reengage in negotiations, but he has also instructed his government that the U.S. can’t be trusted. And Iran will be much less likely to enter into a comprehensive agreement again now that Washington has already pulled out of one and reimposed sanctions, delivering a shock to Iran’s economy. Getting the regime to agree to anything beyond limited concessions on its nuclear program appears implausible.

One way or another, though, Citrinowicz expects 2025 to be “decisive.” Without a new agreement, Iranian leaders could start procuring a bomb. Or Israel and the U.S. could take military action to stave them off. And either of those scenarios could trigger the other.

If Iran heads for the bomb, or leverages its threshold status for geopolitical gain, that could encourage other countries, including U.S. partners, to develop their own nuclear programs. “I absolutely do worry that we could live in a world in the future of not necessarily more nuclear-weapons states but more countries that have this capability to build nuclear weapons,” Brewer said.

In some ways, Iran has already passed the point of no return. By enriching uranium to 60 percent, Tehran has demonstrated that it probably possesses the technical expertise to further enrich that material to weapons-grade, which requires minimal additional effort. Destroying Iran’s physical nuclear infrastructure would be exceedingly difficult. Wiping out Iran’s nuclear knowledge base is not possible. Even if Israel or the U.S. takes military action, the threat of a nuclear Iran will almost certainly persist, at least as long as the current regime remains in power.

Should Iran get nuclear weapons, that would likely embolden its regime at home and abroad, elevate the risk of nuclear terrorism, upend deterrence dynamics between Iran and Israel along with the United States, and spur either an extension of the U.S. nuclear umbrella over Arab partners in the Middle East or a nuclear-arms race in the region—among a host of other potential consequences.

But such outcomes are hard to forecast, because so much of what we know about the interplay between nuclear weapons and international affairs is based on the Cold War and post–Cold War periods. We are now in a third nuclear age, in which nuclear and near-nuclear states come in a greater variety of shapes and sizes. Arms-control agreements have unraveled, diplomatic channels between adversaries have vanished, and establishing nuclear deterrence has never been more complicated.

After the advent of nuclear weapons in the 1940s, at least one new country acquired the world’s most destructive arms every decade until the 2010s, when the streak ended. Nearly halfway through the 2020s, it seems like we may revert to the historical pattern before this decade is done.

The little clump of cells looked almost like a human embryo. Created from stem cells, without eggs, sperm, or a womb, the embryo model had a yolk sac and a proto-placenta, resembling a state that real human embryos reach after approximately 14 days of development. It even secreted hormones that turned a drugstore pregnancy test positive.

To Jacob Hanna’s expert eye, the model wasn’t perfect—more like a rough sketch. It had no chance of developing into an actual baby. But in 2022, when two students burst into his office and dragged him to a microscope to show him the cluster of cells, he knew his team had unlocked a door to understanding a crucial stage of human development. Hanna, a professor at the Weizmann Institute of Science in Israel, also knew that the model would raise some profound ethical questions.

You might recall images of embryonic development from your high-school biology textbook: In a predictable progression, a fertilized egg morphs into a ball of cells, then a bean-shaped blob, and then, ultimately, something that looks like a baby. The truth is, though, that the earliest stages of human development are still very much a mystery. Early-stage embryos are simply too small to observe with ultrasound; at 14 days, they are just barely perceptible to the naked eye. Keeping them alive outside the body for that long is difficult. Whether anyone should is another matter—for decades, scientific policy and regulation has held 14 days as the limit for how long embryos can be cultured in a lab.  

Embryo models—that is, embryos created using stem cells—could provide a real alternative for studying some of the hardest problems in human development, unlocking crucial details about, say, what causes miscarriages and developmental disorders. In recent years, Hanna and other scientists have made remarkable progress in cultivating pluripotent stem cells to mimic the structure and function of a real, growing embryo. But as researchers solve technical problems, they are still left with moral ones. When is a copy so good that it’s equivalent to the real thing? And more to the point, when should the lab experiment be treated—legally and ethically—as human?  

Around the 14th day of embryonic development, a key stage in human growth called gastrulation kicks off. Cells begin to organize into layers that form the early buds of organs. The primitive streak—a developmental precursor of the spine—shows up. It is also at that point that an embryo can no longer become a twin. “You become an individual,” Jeremy Sugarman, a professor of bioethics and medicine at the Johns Hopkins Berman Institute of Bioethics, told me.

[Read: A woman gave birth from an embryo frozen for 24 years]

The primitive streak is the main rationale behind what is often referred to as the “14-day rule.” Many countries limit the amount of time that a human embryo can be kept alive in a petri dish to 14 days. When a U.K. committee recommended the 14-day limit in the 1980s, IVF, which requires keeping embryos alive until they are either transferred or frozen around day five or six, was still brand-new. The committee reasoned that 14 days was the last point at which an embryo could definitively be considered no more than a collection of cells, without potential individual identity or individual rights; because the central nervous system is formed after the 14-day milestone, they reasoned, there was no chance it could feel pain.

But the recent rise of advanced embryo models has led some groups to start questioning the sanctity of the two-week mark. In 2021, the International Society for Stem Cell Research relaxed its 14-day guideline, saying that research could continue past 14 days depending on ethical review and national regulations. (The organization declined to set a new limit.) In July, U.K. researchers put out a similar set of guidelines specifically for models. Australia’s Embryo Research Licensing Committee, however, recently decided to treat more realistic models like the real deal, prohibiting them from developing past 14 days. In the United States, federal funding of human-embryo research has been prohibited since 1996, but no federal laws govern experiments with either real or model embryos. “The preliminary question is, are they embryos at all?” Hank Greely, a law professor and the director of the Center for Law and the Biosciences at Stanford University, told me. Allow one to develop further, and “maybe it grows a second head. We don’t know.” (Having a second head is not necessarily a reason to disqualify someone from being human.) In the absence of an ethical consensus, Hanna is at work trying to cultivate his models to the equivalent of day 21, roughly the end of gastrulation. So far, he said, he’s managed to grow them to about day 18.

Researchers generally agree that today’s models show little risk of one day becoming walking, talking human beings. Combining sperm and eggs the old-fashioned way is already no guarantee of creating new life; even women in their 20s have only about a 25 percent chance of getting pregnant each month. Making embryos in a lab, sans the usual source material, is considerably harder. Right now, only about 1 percent of embryo models actually become anything that resembles an embryo, according to Hanna. And because scientists don’t have a great idea of what a nine-day-old embryo looks like inside the body, Greely said, they don’t actually know for certain whether the models are developing similarly.

[Read: The most mysterious cells in our bodies don’t belong to us]

And yet, in the past few years, scientists have already accomplished what seemed impossible not so long ago. Both Hanna and Magdalena Żernicka-Goetz, a developmental and stem-cell biologist at the California Institute for Technology and the University of Cambridge, have created models for mice with brains and beating hearts. Scientists and ethicists would be wise to consider what qualifies as human before human embryo models have beating hearts, too. The most important question, some ethicists argue, is not whether researchers can achieve a heartbeat in a petri dish, but whether they can achieve one with a model embryo implanted in a human womb. “It's no longer so much about how embryos are made or where they come from, but more what they can possibly do,” Insoo Hyun, a bioethicist and the director of life sciences at Boston’s Museum of Science told me. In an experiment published last year, seven-day-old model monkey embryos were successfully implanted in the uterus of three female monkeys. Signs of pregnancy disappeared about a week afterward, but the paper still raised the specter—or perhaps the promise—of a human version of the experiment.

Building more realistic embryo models could have enormous benefits—starting with basic understanding of how embryos grow. A century ago, scientists collected thousands of embryo samples, which were then organized into 23 phases covering the first eight weeks of development. Those snapshots of development, known as the Carnegie stages, still form much of the basis for how early life is described in scientific texts. The problem is, “we don’t know what happens in between,” Hanna said. “To study development, you need the living material. You have to watch it grow.” Until recently, scientists had rarely sustained embryos in the lab past day seven or so, leaving manifold questions about development beyond the first week. Most developmental defects happen in the first trimester of pregnancy; for example, cleft palate, a potentially debilitating birth defect, occurs sometime before week nine for reasons that scientists don’t yet understand. It’s a mystery that more developmental research performed on embryo models could solve, Greely said.

Better understanding the earliest stages of life could yield insights far beyond developmental disorders. It could help reveal why some women frequently miscarry, or have trouble getting pregnant at all. Żernicka-Goetz has grown models to study the amniotic cavity—when it forms improperly, she suspects, pregnancies may fail. Embryo models could also help explain how and why prenatal development is affected by viruses and alcohol—and, crucially, medications. Pregnant people are generally excluded from drug trials because of potential risks to the fetus, which leaves them without access to treatments for new and chronic health conditions. Hanna has started a company that aims, among other things, to test drug safety on embryo models. Hanna told me he also envisions an even more sci-fi future: treating infertility by growing embryo models to day 60, harvesting their ovaries, and then using the eggs for IVF. Because stem cells can be grown from skin cells, such a system could solve the problem of infertility caused by older eggs without the more invasive aspects of IVF, which requires revving the ovaries up with hormones and surgery to retrieve the resulting eggs.

[Read: Christian parents have a blueprint for IVF]

Answering at least some of these questions may not require hyperrealistic models of an embryo. Aryeh Warmflash, a biosciences professor at Rice University, is studying gastrulation, but the cells that form the placenta aren’t relevant to his research questions, so his models leave them out, he told me. “In some sense, the better your model goes, the more you have to worry,” he said. Hyun told me he cautions scientists against making extremely complex models in order to avoid triggering debate, especially in a country already divided by ideas about when life begins. But given all the medical advances that could be achieved by studying realistic models—all the unknowns that are beginning to seem knowable—it’s hard to imagine that everyone will follow his advice.