News
SpaceX reveals Starship “marine recovery” plans in new job postings
In a series of new job postings, SpaceX has hinted at an unexpected desire to develop “marine recovery systems for the Starship program.”
Since SpaceX first began bending metal for its steel Starship development program in late 2018, CEO Elon Musk, executives, and the company itself have long maintained that both Super Heavy boosters and Starship upper stages would perform what are known as return-to-launch-site (RTLS) landings. It’s no longer clear if those long-stated plans are set in stone.
Oddly, despite repeatedly revealing plans to develop “marine recovery” assets for Starship, SpaceX’s recent “marine engineer” and “naval architect” job postings never specifically mentioned the company’s well-established plans to convert retired oil rigs into vast floating Starship launch sites. Weighing several thousand tons and absolutely dwarfing the football-field-sized drone ships SpaceX recovers Falcon boosters with, it goes without saying that towing an entire oil rig hundreds of miles to and from port is not an efficient or economical solution for rocket recovery. It would also make very little sense for SpaceX to hire a dedicated naval architect without once mentioning that they’d be working on something as all-encompassing as the world’s largest floating launch pad.
That leaves three obvious explanations for the mentions. First, it might be possible that SpaceX is merely preparing for the potential recovery of debris or intact, floating ships or boosters after intentionally expending them on early orbital Starship test flights. Second, SpaceX might have plans to strip an oil rig or two – without fully converting them into launch pads – and then use those rigs as landing platforms designed to remain at sea indefinitely. Those platforms might then transfer landed ships or boosters to smaller support ships tasked with returning them to dry land. Third and arguably most likely, SpaceX might be exploring the possible benefits of landing Super Heavy boosters at sea.
Through its Falcon rockets, SpaceX has slowly but surely refined and perfected the recovery and reuse of orbital-class rocket boosters – 24 (out of 103) of which occurred back on land. Rather than coasting 500-1000 kilometers (300-600+ mi) downrange after stage separation and landing on a drone ship at sea, those 24 boosters flipped around, canceled out their substantial velocities, and boosted themselves a few hundred kilometers back to the Florida or California coast, where they finally touched down on basic concrete pads.
Unsurprisingly, canceling out around 1.5 kilometers per second of downrange velocity (equivalent to Mach ~4.5) and fully reversing that velocity back towards the launch site is an expensive maneuver, costing quite a lot of propellant. For example, the nominal 25-second reentry burn performed by almost all Falcon boosters likely costs about 20 tons (~40,000 lb) of propellant. The average ~35-second single-engine landing burn used by all Falcon boosters likely costs about 10 tons (~22,000 lb) of propellant. Normally, that’s all that’s needed for a drone ship booster landing.
For RTLS landings, Falcon boosters must also perform a large ~40-second boostback burn with three Merlin 1D engines, likely costing an extra 25-35 tons (55,000-80,000 lb) of propellant. In other words, an RTLS landing generally ends up costing at least twice as much propellant as a drone ship landing. Using the general rocketry rule of thumb that every 7 kilograms of booster mass reduces payload to orbit by 1 kilogram and assuming that each reusable Falcon booster requires about 3 tons of recovery-specific hardware (mostly legs and grid fins) a drone ship landing might reduce Falcon 9’s payload to low Earth orbit (LEO) by ~5 tons (from 22 tons to 17 tons). The extra propellant needed for an RTLS landing might reduce it by another 4-5 tons to 13 tons.
Likely less than coincidentally, a Falcon 9 with drone ship booster recovery has never launched more than ~16 tons to LEO. While SpaceX hasn’t provided NASA’s ELVPerf calculator with data for orbits lower than 400 kilometers (~250 mi), it generally agrees, indicating that Falcon 9 is capable of launching about 12t with an RTLS landing and 16t with a drone ship landing.
This is all to say that landing reusable boosters at sea will likely always be substantially more efficient. The reason that SpaceX has always held that Starship’s Super Heavy boosters will avoid maritime recovery is that landing and recovering giant rocket boosters at sea is inherently difficult, risky, time-consuming, and expensive. That makes rapid reuse (on the order of multiple times per day or week) almost impossible and inevitably adds the cost of recovery, which could actually be quite significant for a rocket that SpaceX wants to eventually cost just a few million dollars per launch. However, so long as at-sea recovery costs less than a few million dollars, there’s always a chance that certain launch profiles could be drastically simplified – and end up cheaper – by the occasional at-sea booster landing.
If the alternative is a second dedicated launch to partially refuel one Starship, it’s possible that a sea landing could give Starship the performance needed to accomplish the same mission in a single launch, lowering the total cost of launch services. If – like with Falcon 9 – a sea landing could boost Starship’s payload to LEO by a third or more, the regular sea recovery of Super Heavy boosters would also necessarily cut the number of launches SpaceX needs to fill up a Starship Moon lander by a third. Given that SpaceX and NASA have been planning for Starship tanker launches to occur ~12 days apart, recovering boosters at sea becomes even more feasible.
In theory, the Starship launch vehicle CEO Elon Musk has recently described could be capable of launching anywhere from 150 to 200+ tons to low Earth orbit with full reuse and RTLS booster recovery. With so much performance available, it may matter less than it does with Falcon 9 and Falcon Heavy if an RTLS booster landing cuts payload to orbit by a third, a half, or even more. At the end of the day, “just” 100 tons to LEO may be more than enough to satisfy any realistic near-term performance requirements.
But until Starships and Super Heavy boosters are reusable enough to routinely launch multiple times per week (let alone per day) and marginal launch costs have been slashed to single-digit millions of dollars, it’s hard to imagine SpaceX willingly leaving so much performance on the table by forgoing at-sea recovery out of principle alone.
Elon Musk said today on X that Teslas will start to learn your individual preferences. This is something that he seemed to hint toward earlier this month when he said parking was by far the biggest reason drivers intervene with Full Self-Driving.
Musk made the comment in response to notable Tesla influencer Whole Mars, who said that his vehicle will sometimes disobey the settings he has enabled for his car. He responded to the post, stating that “The car will start to remember your specific interventions and match each person’s individual preferences.”
The car will start to remember your specific interventions and match each person’s individual preferences
— Elon Musk (@elonmusk) July 18, 2026
This is something that could be perhaps one of the biggest ways Tesla could minimize or even work closer toward eliminating interventions altogether. While FSD does a lot of things really well, many people intervene a vast majority of the time not due to major or critical safety errors.
Instead, many take over because the car is doing something that they do not like as a preference; it might park in a parking spot that is not preferred by the driver, it might linger too long in the left lane on the highway (a personal favorite), or it could even take a route that the driver does not like.
These all lead to interventions, but they are not triggered by a major safety issue. Instead, it’s just preference.
READ OUR REVIEW OF TESLA’S LATEST FSD VERSION:
Tesla Full Self-Driving v14.3.5 Early Impressions: new features and early performance
If Teslas could start to learn the personal preferences of the person who owns them, interventions will truly begin to be less frequent. Some of this is already pretty evident, in my opinion. Teslas use a neural network to learn behaviors and accumulate data to improve performance.
For months now, we’ve tracked FSD’s performance at “Except Right Turn” stop signs, something that is very common in Pennsylvania, but many of our readers located in other parts of the U.S. have never heard of. FSD handles one Except Right Turn stop sign very well, one that I travel past frequently. Others that I do not navigate through as often do not have as confident a performance. It seems like the cars might already be doing this to an extent.
🚨 Tesla Full Self-Driving v14.3 proceeds through an Except Right Turn Stop Sign pic.twitter.com/YemRSlens7
— TESLARATI (@Teslarati) April 8, 2026
That example is also for something that is a street sign and not necessarily a driver preference; however, I still feel it is worth mentioning because it only handles that commonly passed Except Right Turn stop sign with true confidence. Others it still seems to struggle with.
This could be one of Tesla’s big moves toward full autonomy, and it could be a pathway to truly unsupervised driving. Every day, millions of cars on the road travel at a human driver’s personal preferences with no incident. Why can’t autonomous vehicles still cater to a passenger’s preferences while being autonomous? Tesla seems to have the idea that it would be possible.
Florida Governor Ron DeSantis has sharply criticized legacy media outlets for what he describes as selective and biased reporting on vehicle accidents involving Tesla. In a recent X post, DeSantis questioned why headlines routinely spotlight the Tesla brand in crash stories, even when human error is the clear cause, while similar incidents with other automakers often receive generic treatment.
A prime example is the June 19, 2026, fatal crash in Katy, Texas. A Tesla Model 3 driven by Michael Butler struck a brick home at high speed, killing 76-year-old Martha Avila inside. Initial reports and headlines prominently featured “Tesla crash” and referenced the driver’s claim that an automated driving-assistance system was engaged.
Many outlets quickly speculated that Full Self-Driving or Autopilot were the cause of the crash, immediately blaming the suites for the accident shortly after it happened.
However, Tesla responded shortly after the accident with vehicle data that showed Butler manually overrode the system by pressing the accelerator to 100 percent, reaching 73 MPH in a residential area, more than double the speed limit. The accelerator remained floored after impact.
Tesla finally clarifies fatal Texas crash, confirms driver manually overrode acceleration
The National Transportation Safety Board (NTSB) later confirmed these findings, and Butler now faces manslaughter charges. His phone searches also included queries like “Tesla FSD too timid,” suggesting he may have intervened aggressively. Despite this, many headlines continued to center Tesla’s technology rather than the driver’s actions.
DeSantis highlighted a Washington Post headline, which was labeled, “Newly released photo shows wreckage of Tesla crash that killed grandmother.”
Do legacy media outlets typically use headlines involving the make of a car in a crash or is that only for Tesla?
It would be one thing if the self-driving malfunctioned but the crash was purely human-induced.
Seems like these outlets want to associate Tesla with crashes as… pic.twitter.com/EmfyeYiuv6
— Ron DeSantis (@RonDeSantis) July 17, 2026
The subheadline noted the driver overrode assistance and floored the accelerator, yet the brand name dominated the framing. He asked whether legacy outlets typically name the make of a car in routine crashes or reserve that treatment for Tesla to push a narrative.
This pattern appears widespread. Crashes involving Ford, Chevrolet, or Toyota vehicles frequently appear as “pickup truck slams into home” or “fatal car crash kills pedestrian” without brand specifics, especially absent new technology angles.
High-profile Ford F-150 or Chevy Silverado incidents tied to large sales volumes often escape brand-callout scrutiny. In contrast, Tesla stories consistently lead with the manufacturer, amplifying perceptions of risk despite data showing strong overall safety performance:
🚨 Why do Tesla Owners get so defensive over the narrative of crashes involving Teslas? https://t.co/aX7ogtjTCR pic.twitter.com/KO4QWaLOKl
— TESLARATI (@Teslarati) June 24, 2026
Tesla’s own 2025 Impact Report indicates vehicles using FSD logged 0.19 major incidents per million miles, roughly eight times fewer than the U.S. average. Models like the Model Y also rank among the safest in IIHS and NHTSA testing for occupant protection. Critics argue disproportionate coverage ignores these statistics and driver behavior factors, such as younger or more aggressive Tesla owners in some studies.
DeSantis frames this as part of a broader political agenda against innovative American companies like Tesla. By consistently naming Tesla while downplaying others, media outlets risk eroding public trust and shaping perceptions detached from the evidence of human error in most cases.
As autonomous technology evolves across the industry, consistent and factual reporting will be essential to separate real safety concerns from narrative-driven coverage.
Tesla entered two new markets this week by advancing its presence in Latvia (Europe) and officially launching operations in Uruguay (South America), marking a rapid dual-continent expansion.
These moves underscore the company’s strategy to tap into emerging EV markets with supportive policies, renewable energy grids, and growing demand for sustainable transport.
Latvia: Strengthening the Baltic Footprint
In Latvia, Tesla has built on its earlier registration of Tesla Latvia SIA in late 2025 with recent steps toward full operations, including job postings for a service center and representation in Riga. This aligns with broader Baltic expansion following Lithuania’s model of pop-up stores and service centers.
Coming to Latvia https://t.co/XNkQQJ2O6a pic.twitter.com/yS9kpcNky1
— Tesla Europe, Middle East & Africa (@teslaeurope) July 17, 2026
EV penetration in Latvia stands at around 7 percent for BEVs in new passenger car registrations. 2025 data showed 1,602 BEVs out of about 22,500 total, or 7.1 percent, with combined plug-ins nearing 19 percent. Growth has been steady but below the European average, supported by government subsidies and infrastructure development. Tesla models like the Model 3 lead local EV registrations.
Vehicles for the Latvian market will likely be sourced from Gigafactory Berlin or Gigafactory Shanghai. Charging infrastructure is robust for the region as well, with over 400- 2,000 public points, with Tesla Superchargers in Riga, Jūrmala, and along Via Baltica routes offering up to 250 kW.
Uruguay: Third South American Country
Tesla teased its Uruguay arrival with “Estamos llegando,” or, “We are arriving,” on social media, followed by an official presentation scheduled for mid-July.
Hola Uruguay 🇺🇾
Nuestros Model 3 y Model Y están cada vez mas cerca! pic.twitter.com/FR41fsA7um
— Tesla Latinoamérica (@Tesla_LatAm) June 30, 2026
The company established Tesla Uruguay SAS, homologated Model 3 and Model Y (three versions each), and appointed local leadership. This makes Uruguay Tesla’s third official South American market after Chile and Colombia.
Uruguay boasts one of Latin America’s highest EV penetrations, with battery-electric vehicles exceeding 20 percent market share recently, driven by tax incentives, high fuel prices, and a nearly 95-100 percent renewable electricity grid. Hundreds of Teslas already operate via grey imports, but official sales bring warranties, service, and support.
Vehicles will be imported from Gigafactory Shanghai, enabling competitive pricing for Model 3 and Model Y. Charging plans include Supercharger development alongside existing infrastructure, leveraging the country’s green energy advantage for affordable operation.
Tesla Superchargers follow Model 3 and Model Y to South American country
Tesla’s Dual Continent Expansion
Tesla’s simultaneous push into Latvia and Uruguay demonstrates efficient scaling: prioritizing service and infrastructure first, then direct sales in high-potential niches. In Europe, it fills Baltic gaps; in Latin America, it counters Chinese dominance while leveraging renewables.
This dual move signals Tesla’s ambition to accelerate global EV adoption amid varying regional paces. By addressing local needs, like subsidies in Latvia or incentives and green grids in Uruguay, Tesla not only boosts volumes but advances its mission of sustainable energy.
For investors and consumers, it highlights resilience and opportunity in diverse markets, potentially paving the way for further growth in underserved regions. With strong fundamentals in both, these entries could yield long-term gains as EV transitions mature worldwide.
Elon Musk says your Tesla will start to learn your individual preferences
Ron DeSantis calls out media bias in Tesla crash coverage
