NASA的Artemis II带来了壮观影像，但这次任务在政治与技术上都很脆弱：拍到月食、日落背光地球弧形与四名太空人安全回收，且飞得比任何人类更远，但公众被呈现的风险远低于实际。2022年无人Artemis Orion试飞中，隔热盾在数十处脱屑，且飞行器之后以每小时25,000英里（40,234公里）的速度重返大气层。2024年检察总长报告指出，若问题重复，可能导致载具或人员丧失。NASA最终改以模拟推得的轨道修正来降低风险，而未采取延迟发射、重新设计或新增试飞，导致重返大气安全仍带有未知机率，如同评论者所比喻的俄罗斯轮盘。
从起飞规划到执行，任务目标持续更迭。现行Artemis框架历时近十年，现至少包含五项任务，且在Artemis II之后，Artemis III定于2027年发射。成本治理成为主轴叙事：总支出约为1000亿美元，而且随著升温升至每次SLS发射超过40亿美元，NASA甚至停止继续追踪，整体开支与任务效益失衡。Orion在二十年间花费超过200亿美元，仍需要临时性的重返大气绕道补救；SLS被评为过重、推进不足、而在可重复使用火箭兴起时却近似一次性。从这种成本—风险比看，几乎所有技术风险都为了有限的即时科学产出而承担。
国会政治和既有产业结构维持了这条路线，外界质疑其是「自我回馈」的甜筒：成本加成契约（cost-plus）为超支提供持续获益。4月和3月修正后的建构方案则转向较务实路径：暂停“lunar gateway”耗资计划、加快发射节奏，并提出月球基地、月面通讯网、月球GPS、车辆与登陆器、表面反应炉等基础设施目标，同时重申“America will never again give up the moon”。同时，SpaceX等商业与更便宜更快的发射节奏正在使SLS-Orion旧模式快速老化，因此Artemis有可能从一个“boondoggle”过渡到更具商业化与可持续性的月球未来。
NASA’s Artemis II delivered spectacular lunar imagery but was politically and technically brittle: a lunar eclipse, a crescent earthset, and a splashdown by four astronauts who traveled farther than any humans before, yet public risk was far higher than implied. The 2022 uncrewed Orion test had a heat shield chipped in dozens of places, and the vehicle later reenters at 25,000 mph (40,234 km/h). A 2024 inspector-general report said a repeat could cause vehicle or crew loss. NASA chose trajectory changes based on simulation instead of a delay, redesign, or extra test, leaving uncertainty in reentry safety that even critics likened to Russian roulette.
From launch planning to operations, program goals kept shifting. The modern Artemis architecture has now run for nearly ten years and targets at least five missions, with Artemis III scheduled for 2027 after Artemis II. Cost governance became the dominant story: roughly US$100 billion has been incurred while costs kept escalating, and NASA eventually stopped tracking as launches exceeded US$4 billion each. Orion has cost more than US$20 billion over 20 years yet still required an ad hoc reentry workaround; the SLS configuration is judged overweight, underpowered, and effectively expendable despite private launch reusability trends. In that arithmetic, nearly all technical risk was borne against a shrinking set of immediate scientific returns.
Congressional politics and legacy industrial structure sustained this trajectory; one critic called it a self-licking ice cream cone, with cost-plus contracting rewarding overruns. March 2026 revisions shifted toward a more pragmatic path: pausing the lunar gateway, pushing faster launch cadence, and sequencing infrastructure goals like lunar base, lunar communications networks, lunar GPS, rovers, landers, and surface reactors while asserting that “America will never again give up the moon.” Meanwhile, SpaceX-era competition and cheaper, higher-cadence launches are making the old SLS-Orion model obsolete, so Artemis may be transitioning from a boondoggle to a broader commercialized lunar future. (Key numbers: 4, 3)