Tesorb Signal: Tesla, SpaceX, and the Musk Companies

Neuralink reports a new cursor-control record and less time spent recalibrating for some trial participants. Lena Ruiz examines what the company’s October update shows, how pretraining uses previously idle brain recordings and why the benefits for new users remain unproven.

Imagine a computer mouse that slowly stops working properly. To fix it, you sit down and play the same training game, again and again, before you can get back to your email.

That's roughly what people in Neuralink's brain implant trial have been living with. By the company's own count, they spent an average of fifty-five minutes a week recalibrating, just to keep the control they already had.

On October first, Neuralink published a technical post saying it has started to shrink that chore. It also announced a new cursor-control record: eleven point three two bits per second.

This is Tesorb Signal, and I'm Lena Ruiz. Today: what that record measures, what the calibration claim means for patients, and what Neuralink says it still can't do.

Start with the headline number. Neuralink measures cursor control with a grid game called Webgrid. The user selects targets on a grid as fast and as accurately as they can. The score, in bits per second, weighs both speed and precision.

In the post, Neuralink says six participants set personal records with its new decoders. Three passed the company's previous record of ten point three nine. One participant, identified only as P15, reached eleven point three two.

That's an improvement of a little under one bit per second, or about nine percent over the old mark.

For context, Neuralink wrote in January that able-bodied people using an ordinary mouse average around eight to ten bits per second. So the record sits above that range. The new post puts the median participant at roughly ten.

Two caveats. These clinical results come from a company update, rather than independent validation. And a record is one person's best run. It isn't what every participant gets every day.

The more interesting part is how they got there.

Neuralink says its participants have streamed more than fifty thousand hours of neural data in what it describes as two years of clinical trials. The first participant alone has recorded more than nine thousand hours, or twenty-two point four billion electrical spikes.

Until now, the company says, nearly all of it went unused. The decoders, the software that turns brain activity into cursor movement, learned from short, labeled calibration sessions.

The new approach borrows a familiar idea from AI, called pretraining. First, a model studies thousands of hours of one person's unlabeled recordings, so it learns the patterns in that person's brain activity. Then a much smaller labeled session teaches it what that person intends.

Neuralink calls these models encoders. Each one is trained on a single participant's own history.

Here's why that matters more than the record.

Brain signals drift. A decoder that works well one day can get worse over the following days. The company says that under the old system, users typically spent ten minutes calibrating at the start of each day. Some waited longer, trading performance for time.

With the pretrained encoders, the company says some users went from ten minutes a day to ten minutes a week. For someone who calibrated every day, that's a seventy-minute week turned into a ten-minute one.

The post says some decoders kept strong performance for more than three weeks. In one case, a participant could still control a cursor with a decoder calibrated twenty months earlier. Usable control is different from maintaining peak performance for that whole time.

The company also reports that thirty seconds of labeled calibration data with the new method did about as well as three and a half minutes with the old one, when tested on data a month later.

For someone using the implant for hours every day, that's time handed back. It's also a step toward something a medical product needs: a device that works when you turn it on.

But note the wording. Some users. Some decoders. One participant. The post shows striking individual cases. It doesn't give a full table of results across every participant. The fifty-five minute average describes the burden before. There's no matching average for after.

The most useful paragraph is near the end, where the company lists what isn't working yet.

Every live result in the post came from models pretrained on a single participant's data. Neuralink's next goal is one shared model that learns from everyone's recordings. So far, the company says decoders built from multi-participant models perform no better in live use than single-participant ones.

There is one promising experiment. A model trained only on one participant's data was moved to another participant with ninety-nine percent of its weights frozen, and it beat that person's existing decoder. That successful transfer needs replication before it establishes a general benefit.

That matters for scale. Neuralink's January update announced twenty-one participants. That's a dated count. It tells us how large the group was then, not how many people are enrolled now.

The live results rely on individual histories, so the post doesn't establish the same benefit for a newly implanted user. A successful pooled model could give that person a stronger starting point.

The longer-term goal is control from the moment someone wakes up after surgery, without calibration. Neuralink calls that zero-shot calibration. It says this and its other proposed next steps remain unsolved.

Neuralink's devices are investigational. They haven't been approved by the FDA or any other regulator. The company cautions that the experiences it describes may not reflect every participant or future outcomes.

The next evidence needs to answer three practical questions.

First, independent confirmation. Results reviewed by outside researchers, covering the whole participant group, would give us a stronger basis for judging these claims.

Second, the pooled model. Does a shared model beat single-person decoders in live use? That's the step that could change how quickly new patients get good control.

And third, the everyday numbers. How much of each day does the device simply work, and how much time does each user spend maintaining it?

The record gets the headline. For the users Neuralink describes, less time recalibrating is the improvement that changes the day.

That's the signal. I'm Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk's companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

SpaceX's Crew-13 reached the International Space Station in 7 hours and 55 minutes. Good orbital timing helped set the record, while NASA's second American crew vehicle is still working toward a return. Lena Ruiz examines Dragon's new flights, Starliner's test plan and the next crew handover.

At ten past eleven Eastern on Thursday morning, a Falcon nine lifted off from Cape Canaveral. On top was a Crew Dragon capsule named Grace, carrying four people. Before the day was over, they would reach the International Space Station.

At about five past seven that evening, Grace docked at the International Space Station. From launch to docking took seven hours and fifty-five minutes. NASA officials say no American spacecraft has ever reached the station faster.

The crew is commander Jessica Watkins and pilot Luke Delaney of NASA, Joshua Kutryk of the Canadian Space Agency, and cosmonaut Sergey Teteryatnikov of Roscosmos. They'll live and work on the station for about six months.

Three days earlier, NASA and Boeing held a press conference about the other American crew capsule. Its next flight with astronauts aboard is now planned for twenty twenty-eight.

This is Tesorb Signal, and I'm Lena Ruiz. Today: the fastest trip SpaceX has ever flown to the station, and why Dragon is also the only American crew vehicle flying there right now.

Start with the record itself. The previous best for any American spacecraft was twelve hours and thirty-three minutes. An uncrewed Dragon cargo flight set it in November twenty twenty-four.

With astronauts aboard, the best was fourteen hours and forty-three minutes, set by Crew-eleven in August twenty twenty-five. Crew-thirteen beat that by six hours and forty-eight minutes. It made the trip in a little more than half the time.

SpaceX was careful not to oversell it. "We got lucky this time," said Julianna Scheiman, who directs NASA science and Dragon programs at the company.

She said the station was in an opportune spot and the orbital mechanics worked out. She added that this is close to the fastest it could be.

That shapes how to read the number. The record came mostly from where the station happened to be at launch, not from new hardware.

It also isn't a world record. A Russian Soyuz once reached the station three hours and three minutes after launch, after just two orbits.

Launch-to-docking time measures the route and the timing of the rendezvous. It does not, on its own, tell us how powerful a spacecraft is. A shorter trip is welcome for the people inside. It is a different question from whether a second vehicle is available when the first cannot fly.

The capsule has its own story. Before Thursday, Grace had flown only once. NASA says its first mission carried the private Axiom-four crew to the station in twenty twenty-five. Crew-thirteen is its second visit.

This mission had already faced a delay. On August twenty-ninth, NASA announced an oxidizer leak in the Dragon's propulsion system, found during standard preparations for launch.

NASA said teams would run additional tests, review the data, and complete any necessary rework before launch. The response was to fix the problem on the ground and announce a new date when ready.

The four Crew-twelve astronauts already on the station now have their replacements. NASA says there will be a brief handover before Crew-twelve begins its trip home, no earlier than Monday, October fifth. That remains a target, subject to operations and conditions for the return.

A crew rotation links several events: launch, docking, handover, and return. A delay in one can change the rest. Having a second American vehicle would give NASA another option, though each mission would still need its own preparation.

That brings us back to Monday's press conference. In twenty fourteen, NASA hired two companies, Boeing and SpaceX, to fly its astronauts. The goal was redundancy: if one vehicle was grounded, the other could keep flying.

The qualifier American matters. Russian Soyuz capsules also carry people to the station, including NASA astronauts. This episode is about the separate American systems NASA commissioned. Dragon is operating. Starliner is still working toward a return to crewed flight.

Boeing's Starliner has carried one crew. On its twenty twenty-four test flight, it had helium leaks and problems with its maneuvering thrusters on the way to the station.

NASA decided to bring Starliner home empty. Its two astronauts, Suni Williams and Butch Wilmore, eventually came home on a Dragon in March twenty twenty-five. They had spent two hundred eighty-six days in space.

On Monday, September twenty-eighth, NASA said the uncrewed Starliner-one engineering mission could launch in December this year or January twenty twenty-seven. Its stated plan is to return astronauts on Starliner-two by twenty twenty-eight. Those are plans, and depend on testing and certification.

NASA also said it intends to buy a fifth and sixth Starliner flight, and to help certify United Launch Alliance's Vulcan rocket for use after the Atlas five's final flight. Keeping a second system available requires work on the capsule and its launch vehicle.

That's a commitment, and it's a plan. Neither one is a flight yet.

Meanwhile, NASA bought more Dragon flights. In an announcement dated September eighteenth, it added Crew-fifteen, sixteen, and seventeen to SpaceX's contract for nine hundred forty-six million dollars.

That works out to an average of about three hundred fifteen million dollars per mission. NASA says the price covers more than seats.

It includes ground and recovery operations, cargo, and a lifeboat capability while Dragon is docked. SpaceX's total contract value is now five point nine two billion dollars, across seventeen missions.

Those three new flights have readiness dates in twenty twenty-seven and twenty twenty-eight. NASA plans to operate the station through twenty thirty, and it has picked SpaceX to build the vehicle that will guide the station down at the end.

For the station's remaining years, NASA needs transport it can count on while working toward a second American system. Buying more Dragon missions maintains the operating service. Returning Starliner to flight could restore the redundancy NASA wanted when it hired two providers.

The next Starliner flight will help answer whether its improvements work in space. NASA has described changes to the service module's thermal design and further work on thruster valves. The uncrewed test is meant to gather performance data and identify remaining risk before people fly again.

The next checkpoints are close enough to put on a calendar. Crew-twelve's return could begin as soon as Monday, October fifth, after the handover.

Second, Starliner-one in December or January, and how its thrusters perform. Third, whether NASA orders any Dragon flights beyond Crew-seventeen, and when.

Thursday's record shows what Dragon can do when the timing lines up. Monday's update shows how much work remains on the second American system. For now, America's fastest crew ride to the station is also its only one in operation.

That's the signal. I'm Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk's companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Elon Musk signed the White House Accord on Super Intelligence, promising independent external evaluation and board oversight. Lena Ruiz compares that commitment with xAI's published framework and asks what evidence would make the promise testable.

Elon Musk signed a White House promise to bring independent eyes into AI safety. The framework linked from his company's safety page still leaves that oversight unnamed. Between the signature and the operating rules, there's a gap worth watching, and a distinction worth getting right.

I'm Lena Ruiz, and this is the Tesorb Signal. On Tuesday, September twenty-ninth, Musk joined executives from five other companies in signing the White House Accord on Super Intelligence. His signature appears for xAI, the company behind Grok.

The other companies were Google, Anthropic, Meta, OpenAI, and Nvidia. The signed document has one page of commitments and a separate signature page. Its most useful detail is how it says oversight should work, once the ceremony is over.

The accord sets out four layers. Internal controls monitor the technology. An empowered internal team checks those controls. An independent external auditor or evaluator assesses whether that machinery works. An independent board committee receives the reports and ensures problems are addressed.

An auditor and an evaluator aren't necessarily doing the same job. The document allows either. What matters for this promise is that the outside assessment reaches the controls, monitoring, and detection system, rather than stopping at a flattering demonstration of a model.

The accord is a voluntary commitment. It does not set an implementation deadline, prescribe penalties, or require the resulting assessments to be published. It says participating companies will meet regularly, and leaves open the possibility of legislation or regulation later.

So the signature creates a public promise whose implementation can be examined. Existing laws still matter. But this document supplies no new enforcement mechanism of its own. Judging it will require evidence about the arrangements companies actually put in place.

For xAI, the available comparison is its Frontier Artificial Intelligence Framework. At our October first check, the company's safety page linked to the version effective June thirtieth. That is the published framework we're discussing, rather than a claim about every internal practice.

It runs nine pages. Its governance section is two sentences about internal structures, applicable law, implementation, and continuing legal and compliance reviews. Across the document, searches find neither the word audit nor the word board. It identifies no independent auditor or board committee.

That omission cannot establish that xAI has no outside reviewers. Its safety page publishes model cards and safety evaluations. The June framework itself references information security evaluations. The narrower question is how the specific oversight promised in September will be documented and made accountable.

The framework also addresses major risk categories, including dangerous misuse, loss of control, and harmful manipulation. It describes regular risk assessment and incident responses. Those are real subjects for a safety program. Their inclusion doesn't answer who independently checks the program's operation.

There is a history behind the wording. The framework updated in December of last year expressly assigned risk owners responsibility for periodic audits to enforce implementation. That was an audit responsibility inside the company's governance approach, not a clearly mandated independent external audit.

December also contained a section on public transparency and third-party review. It described possible publication of information and possible sharing of unredacted material with vetted external teams or government agencies. Those possibilities gave readers a view of potential outside scrutiny, without guaranteeing a public audit report.

An August version had similar language about transparency and external review. It also described working with government bodies, nonprofits, testing firms, industry, and academia. Outside collaboration was part of the story before this week's signing. We should keep that history in view.

What changed in June was the written architecture. The earlier transparency section and explicit periodic audit responsibility don't appear in that form. That comparison supports a question about the public rules. It does not tell us whether a particular reviewer was hired, dismissed, or given less access.

And the chronology matters. The June document predates the September accord. A framework written three months earlier could not explain compliance with a promise that had not yet been signed. The useful next step is an update showing how the new commitment fits the existing program.

To make that update meaningful, start with the evaluator's mandate. Can the outside reviewer examine the actual monitoring system? Can it inspect incidents and unresolved findings? Can it see what happened between a warning being raised and a decision being made?

Then consider independence. A company paying for an assessment can still receive a demanding review. Payment alone doesn't settle the issue. The relevant details include conflicts of interest, access to evidence, limits on the engagement, and the reviewer's ability to report an uncomfortable finding.

The accord does not spell out those details. A credible implementation could. Publishing the reviewer's identity and the scope of the work would help outsiders judge whether the promised independence has substance, even if sensitive technical material needs to stay confidential.

The board committee is the other half of the promise. Its value depends on what it can do with a report. Does it get unresolved issues directly? Can it demand more testing? Does a serious finding lead to a recorded decision and a follow-up check?

Those are questions for implementation, not findings that xAI has failed a test. The public framework leaves them open. A committee name on a website would be a beginning. A clear mandate and evidence of follow-through would carry more weight.

Think about a hypothetical warning that a system can bypass a safeguard. The internal team might propose a fix. The outside evaluator could test whether it works. The directors would then need to understand both the original finding and any risk that remains.

Public reporting can be selective without being empty. Companies have good reasons to protect dangerous technical details and private information. A summary can still say what was reviewed, what kinds of issues were found, and whether remediation was checked.

For listeners, the evidence to watch is concrete. An updated framework explaining the oversight arrangement. A named outside reviewer with a defined scope. A board committee with a mandate. And reports that show what changed after findings arrived.

The signing made independent oversight part of xAI's public commitment. Its currently linked framework does not yet explain that arrangement. The next evidence should be a review with a meaningful scope, and directors willing to act on what it finds.

I'm Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk's companies, visit tesorb.com. Corrections are posted to the episode page.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

X replaces Creator Revenue Sharing with Original Content Rewards while alleging coordinated payout manipulation in a London lawsuit. Lena Ruiz explains the new impression and originality rules, the allegations, and the evidence needed to judge whether the incentives improve.

In one example cited by X, two accounts post the same crypto news. Nearly word for word, eleven seconds apart.

X also describes three other accounts supplying likes, replies and reposts across the network. To anyone scrolling past, it could look like a conversation.

The company alleges it was coordinated activity, paid through its creator program.

This is Tesorb Signal, and I'm Lena Ruiz. Today: X has changed how it pays the people who post on it. We'll look at what the old program rewarded, what the new one says it pays for, and what that lawsuit tells us about the gap between the two.

Let's start with the case, because it's the most concrete thing on the table.

X's published High Court claim, dated September seventeenth, names two individuals living in the United Kingdom, plus unidentified people.

X says six accounts enrolled in its creator payment program posted the same or nearly the same crypto news seconds or minutes apart, while three more accounts supplied likes, replies and reposts to make the traffic look real.

It alleges the accounts received at least two hundred seven thousand pounds in creator payouts. The company is asking for repayment and damages.

The filing says X suspended the accounts on August eighteenth. It also estimates at least seventy-five thousand pounds in investigation and related costs, more than a third of the claimed payouts.

Those are X's allegations, not a court's findings. The filing presents the company's case. A contested claim needs the defendants' response and the court's assessment before it becomes a verdict.

To see why a network like this might be profitable, you have to look at how the payments worked.

X's creator revenue sharing began in July twenty twenty-three. At the start, payments were tied to ads shown in the replies to creators' posts. That made the reply section part of the business model. Getting people to react could matter as much as getting them to read.

That design rewarded conversation, of any kind. A post that started an argument could pay better than one that informed people.

In November twenty twenty-four, X changed the formula. Payouts would be based on engagement from paying Premium subscribers, not ads in replies. Social Media Today reported the pool as up to a quarter of Premium subscription revenue.

The unit of reward moved from ads to engagement. Either way, creators were paid for reactions, and reactions can be manufactured.

The next redesign goes after the material being posted, as well as the audience reacting to it. X wants creators to bring new work onto the platform. That changes the bargain for accounts built mainly around collecting and reposting other people's material.

The old program stopped counting earnings on September seventh. X's help page listed its final payout as expected on or around September eleventh. The distinction matters: a payment date in the rules is a schedule, not proof that every creator received a check.

Its replacement is called Original Content Rewards. X's current help page says payments come every two weeks, once a creator's earnings reach the minimum. We'll judge the results through what creators receive, alongside the program's written promises.

Here's how X describes the new math. Creators are paid on what it calls qualified impressions. A Premium subscriber saw the post on their Home Timeline, with at least half the post visible. Repeat views of that post from the same account, fraudulent views, and promoted impressions are excluded.

To apply, you need an eligible Premium subscription, at least five hundred verified followers, and half a million Home Timeline impressions from verified users over ninety days. Replies don't count toward that total. Meeting the thresholds doesn't guarantee admission; X reviews applications.

And the content has to be original. That includes reporting, your own photos and video, graphics, and memes you made. Simply copying posts or re-uploading clips doesn't qualify. Commentary can count when it adds substantial context or perspective.

So what has changed, and what hasn't?

The biggest change is the target. Paying for views of a post on the timeline, rather than for replies and likes, removes some of the reason to stage fake arguments. Paying only for original material aims straight at the aggregator model.

But some things carry over. The people being counted are still Premium subscribers. A rule against fake or coordinated views needs enforcement to work. X says fraudulent impressions are excluded, and its terms allow payments to be withheld or adjusted when it finds manipulation.

The public program rules describe what should qualify. They don't tell us how frequently the checks catch abuse. That leaves an evidence gap between the promise to filter bad traffic and the results of doing it.

Deciding what's original is also a judgment call. Commentary that adds meaningful value qualifies. Someone at X, or some system at X, has to decide what meaningful means across millions of posts.

And there's a question of scale. The help page and payment terms don't give a total payout pool or an approval count. They also don't give a rate that lets a creator calculate a dependable check from a view count. The terms let X change how payment is calculated.

For someone making original work, that's a practical uncertainty. Time spent reporting or filming is a cost paid up front. A bigger audience may create an opportunity, but the written rules alone don't show what that work will earn.

What to watch next. First, data on the new program: approvals, payouts, and invalid traffic removed. Second, the London case. Responses, a settlement, or a ruling would help test X's allegations. Third, whether more lawsuits follow. One case can be a warning. A series would suggest X sees a wider problem.

Every payment system teaches people what to make. X has spent three years rewriting that lesson, and the next few paychecks will show what the new one teaches.

That's the signal. I'm Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk's companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Starship Flight 14 reached orbit and deployed operational Starlink satellites, then returned early after an ascent engine issue. Lena Ruiz revisits the postponed ship catch and the repeated launches and orbital refueling needed for NASA's lunar plans.

On Monday morning in South Texas, the biggest rocket ever flown lifted off for the fourteenth time. About twenty-five minutes later, its upper stage did something none of the thirteen before it had done. It fired its engines and stayed in orbit.

About three hours after liftoff, that ship was in the Pacific Ocean. The flight plan had called for almost ten.

And the tower arms that were supposed to catch it? By launch day, they weren't part of the plan at all.

This is Tesorb Signal, and I'm Lena Ruiz. Today, Starship Flight fourteen: what it proved, what it left out, and what still stands between this rocket and the Moon.

In episode eighty-eight, we previewed this flight. The plan then was a stack of firsts: orbit, real cargo, and the first attempt to catch a returning ship with the launch tower. The target was late August. We said you'd hear the verdict here, so here it is.

Liftoff came shortly before nine in the morning, Eastern time, on Monday, September twenty-eighth. SpaceX says the Super Heavy booster lost an engine on the way up, but it continued its flight and came down in the Gulf, as planned.

The ship had the harder morning. One of its six engines, a vacuum Raptor, shut down early during the climb. The other five burned longer to make up for it. That got the ship onto its planned suborbital path, where it could still return to Earth without going into orbit.

The team then checked the ship's systems and cleared it for the next step. A single sea-level Raptor relit for the separate orbit insertion burn. This was the point where Starship entered orbit for the first time.

About half an hour into the flight, it started releasing its cargo: twenty-six of SpaceX's newest Starlink satellites. It's the first time Starship has delivered an operational payload to orbit.

Then the flight team made a call. Citing caution over the engine problem, SpaceX cut the time in orbit short and brought the ship home early.

Independent coverage put splashdown at roughly three hours after liftoff. The published timeline had allowed almost ten. Reaching orbit and bringing the ship down were real achievements. The longer period of operating in space was a test the team chose to shorten.

SpaceX's own scorecard says the key milestones were still met: orbit, deployment, and a controlled splashdown for both stages. Independent coverage agrees on those basics. The company's recap doesn't explain what caused the engine shutdown.

Now, the part that didn't happen. The catch.

The mission plan for this flight ended with a controlled reentry and a splashdown in the Pacific. The ship catch we described in our preview had been dropped. SpaceX's published recap doesn't give a reason for that change.

SpaceX took on two big new things, orbit and working cargo, while leaving the ship catch for another flight. In our preview, we talked about isolating each new challenge. This time, that meant the tower didn't get its turn.

A catch beside the launch pad carries more risk than a splashdown far out at sea. For its first orbit, SpaceX chose the sea.

The cargo is where the business is. The new satellites are the third generation of Starlink, built to add much more capacity. Industry coverage puts each satellite's designed downlink capacity at about one terabit per second. That is a design figure, rather than a measurement of customers' service.

SpaceX says it has made contact with all twenty-six. They still need to complete checks in space and raise their orbits before serving customers. Industry coverage reports the company's expectation as a matter of weeks. Until they carry real traffic, the capacity is still a plan.

Three of the twenty-six carry cameras, pointed back at Starship's heat shield. SpaceX says the goal is to learn how to judge whether a shield is ready to fly again. That question sits underneath every plan to reuse a ship.

Which brings us to NASA. Starship is also meant to be the Moon lander for the Artemis program. In a March report, NASA's inspector general laid out what that takes.

Before a landing, SpaceX plans to put a fuel depot in low Earth orbit, then fill it with more than ten tanker flights, launching one every six days. Those tankers are meant to fly from both Texas and Florida.

Take just eleven tankers at that pace. There are ten gaps between eleven launches, so from the first to the last is sixty days. That's an operating rhythm, not a single spectacular flight. Keeping the depot supplied means the launch system has to work repeatedly.

At the time of that March report, SpaceX had yet to show it could turn a launch pad around in twelve to twenty-four days. The inspector general also described large-scale refueling between vehicles as a major technical challenge. Getting a ship into orbit doesn't demonstrate that transfer.

Under NASA's revised plan, Artemis three becomes a test in Earth orbit, and the first landing moves to Artemis four, aimed at twenty twenty-eight. Monday's orbit is a step on that road. It isn't refueling, and it isn't a landing.

So, what to watch. First, SpaceX's review of that vacuum engine shutdown, and whether it slows the next flight.

Second, the next flight's date and objectives, when SpaceX confirms them. Can the company shorten the interval between launches, while fixing what went wrong? A reliable schedule matters to Starlink customers and to NASA's fuel-depot plan.

Third, the catch. We'll watch for a confirmed ship-catch attempt and the conditions SpaceX sets before trying it. Bringing a vehicle back beside a launch pad will need both a dependable landing sequence and confidence that its systems can handle the return.

The heat shield belongs on that list, too. The camera experiment is meant to help judge the condition of tiles in space. The useful result will be what those images show, and whether the company can turn that evidence into a decision to fly a ship again.

Fourteen flights in, Starship has finally made it all the way around the world. The next test is how soon it can go around again, and then again.

That's the signal. I'm Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk's companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Nvidia has launched DSX Ready, a program that checks whether batteries can keep AI data centers steady on the grid, and Tesla’s Megapack is one of the first three to qualify. Lena looks at the problem behind it: data centers in Northern Virginia dropping off the grid all at once. She also covers what Nvidia’s grade does and doesn’t prove, and what Tesla’s own filing shows about its biggest related customer, SpaceX.

On a July evening in twenty twenty-four, a piece of equipment failed on a power line in Northern Virginia. The line tried to reconnect itself, over and over. Six faults in eighty-two seconds, each one lasting less than a tenth of a second.

The utility didn’t cut anyone off. But about fifteen hundred megawatts of demand vanished from the grid anyway. It was data centers, switching themselves over to backup power all at once.

It happened again the next winter. Then, two years later, it happened bigger.

This is Tesorb Signal, and I’m Lena Ruiz. Today, Nvidia has started grading the batteries that sit between AI data centers and the grid. Tesla’s Megapack made the first list. We’ll look at what that grade means, what it doesn’t, and who is already buying.

Start with that first event. The North American Electric Reliability Corporation, known as NERC, wrote it up in an incident review. It says the grid has historically been planned for big power plants dropping off, not for this much demand leaving at once.

When that much load disappears, frequency and voltage both climb. In twenty twenty-four they stayed within safe limits. Operators still had to step in to bring voltage back down.

NERC found the load loss came entirely from data centers. None of it was cut by the utility. The data centers’ own protection systems saw the disturbances and moved to backup power, which is exactly what they were designed to do.

Then, on July twenty-second this year, a mechanical failure took another two hundred thirty kilovolt line out of service in Northern Virginia. According to the grid operator P-J-M, about two thousand nine hundred seventy megawatts of data center load moved to onsite generation in a first wave. Roughly eleven hundred more followed as voltage rose.

That first wave alone is about double the twenty twenty-four event. On September eighth, P-J-M formally proposed ride-through requirements for large loads, and it plans to file them with federal regulators in November. In plain terms, those are rules saying a data center should hold on through a brief grid wobble instead of letting go.

That’s the problem. Here’s where the batteries come in.

On September twenty-first, Nvidia launched a program called D-S-X Ready. It checks power and cooling equipment against Nvidia’s reference designs for what it calls AI factories. It launched with two categories, battery storage and cooling, and three battery makers qualified: Hitachi Energy, L-G Energy Solution, and Tesla.

According to Nvidia’s technical documentation, the battery tests cover riding through low-voltage events, answering demand-response calls, and switching between the grid and island operation. They also cover restarting a site’s power from scratch, which engineers call black start.

Nvidia’s documentation also describes the rapid power ramps that AI workloads generate. Part of the battery’s job is to absorb those swings on site, so they don’t show up at the point where the campus meets the utility.

Trade press reports name Tesla’s qualified product as the Megapack two X-L, rated at about one point nine three megawatts and three point eight five megawatt-hours per unit. Each of the other two qualified systems is listed at ten megawatt-hours.

Now, what the grade is not. The suppliers run the tests themselves and submit the data, and Nvidia reviews it. Nvidia states that passing does not replace site-level engineering or imply site-level stability. Transformers, switchgear and generators sit outside the test.

So this is a checked box on a spec sheet. It isn’t proof that any particular campus will ride through the next fault.

Hitachi Energy and L-G put out their own announcements. Tesla hadn’t commented as of this recording. Nvidia’s announcement is the confirmation.

That brings us to the customer closest to home. Tesla’s quarterly filing for the three months ending in June says it recognized three hundred eighteen million dollars of revenue from SpaceX buying Megapacks. Over the first six months of the year, the total was four hundred five million. That leaves about eighty-seven million for the first quarter, so SpaceX’s Megapack purchases more than tripled from one quarter to the next.

Tesla’s whole energy business brought in about three point one billion dollars that quarter. So roughly one dollar in ten came from a company Elon Musk also runs. The filing describes those sales as ordinary course of business.

The filing doesn’t say where those batteries went. One visible candidate is SpaceX’s AI data center site called Colossus two, near Memphis. Canary Media counted about seven hundred twenty Megapack containers there in July satellite imagery. That’s a count from photos, not a figure from either company.

There’s a less flattering number in the same filing. Tesla’s energy revenue rose thirteen percent from a year earlier, but gross profit fell from eight hundred forty-six million dollars to six hundred forty million, a drop of about a quarter. In explaining its revenue, the filing mentions a lower average selling price per Megapack.

More boxes, at a lower price per box. A qualification that helps Tesla sell into AI campuses doesn’t, by itself, fix that margin.

So what should you watch?

First, Tesla’s third-quarter deployment figure. It’s expected this Friday, October second. The second quarter came in at thirteen point five gigawatt-hours.

Second, whether federal regulators approve P-J-M’s ride-through rules. If they do, a battery that can hold a data center steady through a fault stops being a nice extra and becomes part of the price of connecting.

And third, whether Tesla starts naming D-S-X Ready in its own materials, or naming AI data center customers beyond the one that shares its chief executive.

The grid used to worry about power plants letting go. Now it worries about the customers. Nvidia has written down what a battery must do to keep them connected, and Tesla is one of the first three to pass the paperwork.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

The US Justice Department has asked the EU’s General Court to let it back X and Elon Musk against the European Commission’s €120 million fine under the Digital Services Act. That fine was the first the law ever produced. Lena Ruiz explains what the Commission found, what Washington is arguing, and why the case matters for every large platform in Europe.

Open X in Europe and tap on a blue checkmark. You’ll find a notice explaining what the badge means, and what it doesn’t.

Premium accounts carrying a blue badge, it says, are not verified under identity or authenticity criteria.

That small line of text is part of a legal fight that just picked up a very large new participant. On September twenty-fourth, the United States government asked a European court to let it stand beside Elon Musk.

This is Tesorb Signal, and I’m Lena Ruiz. Today: how a hundred and twenty million euro fine over a blue checkmark became a dispute between Washington and Brussels, what each side is actually arguing, and what to watch as it moves through the courts.

Start with the fine itself.

On December fifth, twenty twenty-five, the European Commission fined X one hundred and twenty million euros under the Digital Services Act. The A-P puts that at about one hundred and thirty-seven million dollars.

It was the first time the Commission had found any platform in breach of that law. The investigation had opened in December twenty twenty-three, so it took two years to reach a decision.

The Commission named three problems. First, the blue checkmark. Anyone can buy one by subscribing, and the Commission said that calling those accounts verified deceives users and exposes them to impersonation scams.

Second, X’s advertising library, which the law requires so outsiders can see who is paying for which ads. The Commission said it lacked key details and was hard to use.

Third, researchers. The Commission said X’s terms blocked independent collection of public data, and that its formal access process put up extra hurdles.

According to tech outlet Heise, the fine breaks down as forty-five million euros for the checkmark, forty million for researcher access, and thirty-five million for the ad library. The checkmark accounts for a bit more than a third of the total.

For scale, the law allows fines of up to six percent of a company’s global annual revenue.

X didn’t accept it. In February, the company filed an appeal with the General Court of the European Union, the bloc’s second-highest court. Its government affairs team described the investigation as containing grave procedural errors and a tortured interpretation of the law.

There are actually two cases. One was brought by X’s corporate entities. The second was brought by Elon Musk personally.

That detail matters for what happened last week.

On September twenty-fourth, the Justice Department announced it had filed an application to intervene in both cases, in support of X and Musk. It said it coordinated with the State Department.

Assistant Attorney General Brett Shumate said the Commission, quote, inappropriately attempted to expand its regulatory authority to reach American companies not present or operating within its jurisdiction, end quote.

The department’s arguments go beyond the checkmark. It questions how the Commission decided who counts as a provider of the service. It argues the decision reached Musk as an individual, and other American companies he owns that it says have nothing to do with X’s service.

And it invokes the corporate veil, the principle that shareholders generally aren’t held responsible for what a company does. The department warns that if the Commission’s approach stands, it could affect many American digital companies operating in Europe.

The Commission’s answer was short. Spokesperson Thomas Reg-nee-AY said, quote, we are enforcing our legislation objectively, transparently and with a solid case, end quote. He said the Commission is ready to defend its position in court.

So what’s claimed, and what’s established?

What’s established is the decision, the amount, the three findings, and the fact that both X and Musk have challenged it. The Justice Department’s filing is confirmed by its own announcement.

What’s claimed, on both sides, is how the law should apply. The Commission says X deceived users. X says the process was unfair. The United States says Europe reached too far. The court hasn’t ruled on any of it, and it hasn’t yet said whether it will admit the United States as a party.

There’s also a gap between the argument in court and what X has done on the ground. In Europe, X now calls those blue-check accounts Premium rather than verified, and it spells out that the badge isn’t an identity check. That seems aimed at the Commission’s core complaint, even while the company fights the fine. We found no public word yet on whether the Commission is satisfied with that change.

Why should listeners outside Europe care?

Because this is the first challenge to a Digital Services Act fine to reach a European court. Whatever the judges decide about who a provider is, and how far the Commission can look past a company to its owner, will shape how the law applies to every large platform, not just X.

It also isn’t the only open front. The Commission’s wider investigation into illegal content and information manipulation on X, opened alongside the checkmark probe in twenty twenty-three, is still running. So is a separate investigation, opened in January, into sexualized deepfake images made with the Grok chatbot.

And this isn’t the Justice Department’s first move on behalf of a Musk company this year. In June, it asked to intervene in a Mississippi lawsuit over the turbines powering an x-A-I data center.

Here’s what to watch.

First, whether the General Court lets the United States in as an intervener. That’s a procedural step, but it would put an American government brief in front of European judges.

Second, whether the Commission accepts X’s Premium relabeling as compliance on the checkmark. The Commission already signed off on X’s plan for the ad library and researcher access in July.

Third, the other X investigations. A second decision could arrive while this one is still being argued.

A blue checkmark used to be a small promise about who you were talking to. Now it’s the test case for who gets to set the rules.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

The Boring Company raised $3 billion at a $23 billion valuation. Lena Ruiz checks the projects behind the funding in Dubai, Nashville and Las Vegas, and what separates a transport promise from a service people can use.

Three billion dollars raised. Twenty-three billion dollars of company value. One hundred twenty-three stations on the Las Vegas plan. Those numbers all belong to the Boring Company. Only one tells you where a passenger might eventually get out.

This is Tesorb Signal, and I’m Lena Ruiz. Today, the Boring Company’s biggest funding round, the projects behind it, and a test you can apply to each new promise. Has somebody signed? Has somebody dug? Can somebody ride?

Start with the money. In its September ninth announcement, the company said it raised three billion dollars at a valuation of twenty-three billion. The United Arab Emirates and affiliated investment entities led the round. Sequoia Capital and several other investment firms also participated.

That valuation is the price investors put on the company in this financing. It isn’t revenue, profit, or a pile of cash available for construction. The new capital is three billion dollars. Keeping those two numbers separate matters when the headline gets translated into miles of tunnel.

The announcement describes a partnership covering more than one hundred fifty kilometers of underground infrastructure across the Emirates. That’s in addition to the previously awarded Dubai Loop project. Think of it as a much larger commitment laid alongside the smaller project that’s supposed to prove the approach.

The practical question is how that capital turns into transport people can use. A tunnel company has to connect several jobs. Digging. Building stations. Winning approvals. Running vehicles. Selling rides. Faster progress on one doesn’t automatically finish the others.

Dubai shows why those stages matter. Its transport authority signed an implementation agreement in February. The pilot covers six point four kilometers and four stations, connecting the financial district with Dubai Mall. The authority puts the full route at twenty-two point two kilometers, with nineteen stations.

The authority projects about thirteen thousand passengers a day for the pilot, rising to thirty thousand for the full network. Those are forecasts for the planned system. They don’t describe passengers already moving through it.

And the project has moved beyond paperwork. The Boring Company’s Dubai page shows a precast concrete operation preparing the tunnel lining. The same page describes design, mobilization, and a process involving roughly forty-eight permits and approvals from about ten entities. Its stated goal is to begin tunneling in the second half of this year.

That gives us something more useful than a simple started or not started label. Concrete preparation is real work. It still leaves a chain of approvals, excavation, station construction, and operating readiness before anyone buys a ride.

Now take Nashville. Here, the company’s August update says two machines were actively tunneling, with a third targeted to arrive in September. Its current project description covers almost twenty-five miles, including an airport route and a separate extension west of downtown.

The original state announcement targeted a first operating segment as early as fall of twenty twenty-six. That’s a target to check as autumn arrives. The August update establishes excavation, but doesn’t establish that passengers have started using the system.

Which brings us back to Nevada, where listeners have seen this ledger before. In episode eighty-nine, we compared a plan for one hundred four stations with about a dozen open. The company’s Nashville frequently asked questions now describe a Vegas plan with one hundred twenty-three approved stations and fourteen operational stations.

Those are company figures, and the website has some inconsistencies. Its separate live station list names thirteen active locations. So treat fourteen as the company’s reported total, rather than an independently audited count of underground connections.

Still, the scale comparison is clear. Fourteen is about eleven percent of one hundred twenty-three. That doesn’t make every unopened station late. An approved map describes an eventual network, and stations can open in phases. It does show how much of the intended footprint remains ahead.

Local station openings are the place to keep score. In September, Fox Five reported that tunnels were extending toward Paradise Road, with stops including Virgin Hotels expected to open soon. The company hoped to finish parts of that expansion before November’s Formula One race, while cautioning that schedules could shift.

New destinations would make the network more useful. But adding dots to the approved map expands the denominator too. To understand progress, keep a dated record of openings alongside the plan. Otherwise a bigger blueprint can look like a bigger transport service.

And then there’s Texas. The Austin Business Journal reported on September twentieth that Musk had revived an Austin to San Antonio hyperloop proposal, promising trips under thirty minutes. A publicly accessible archive preserves his post describing speeds above two hundred miles an hour.

The speed and journey time are Musk’s proposed outcome. They aren’t results from an operating route. That announcement gives us an ambition to follow; it doesn’t, by itself, establish an approved alignment or a construction schedule.

Put those stages next to each other. Las Vegas has passenger service and an expanding plan. Nashville has active excavation. Dubai has an implementation agreement and documented preparation. The Texas proposal comes with a travel-time promise that still needs a project record behind it.

The next update should move something along that chain. A permit granted. A tunnel completed. A station opened. A passenger carried. Those events tell us more about delivery than another valuation headline can.

Watch the next Vegas openings, Nashville’s first operating segment, and evidence of Dubai tunneling. Money can give a builder the means to attempt all three. The construction and operating record will tell us what those means achieved.

Money moves at the speed of a term sheet. Tunnels don’t.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

SpaceXAI’s Grok Bot agent reached 418,000 weekly users about a month after launch, according to a presentation Bloomberg viewed. We look at what that number counts and what it leaves out, the tenfold price cut that came before it, and why Grok 4.7 arrived after the showcase meant to prove it.

Four hundred eighteen thousand. That’s how many weekly users SpaceXAI’s Grok Bot agent had on September fourteenth, about a month after launch.

The figure didn’t come from an earnings call or a filing. It came from a presentation shared at an event in London, and Bloomberg reported it on Tuesday.

A week earlier, the same number was lower by about a fifth. That kind of growth sounds impressive. The question is what, exactly, is being counted.

This is Tesorb Signal, and I’m Lena Ruiz. Today, Grok Bot’s first report card: what the number says, what it leaves out, and why the model built to power it arrived late.

First, a quick reminder. Grok Bot is not the Grok chatbot on X. It’s SpaceXAI’s agent product. Each bot gets its own cloud computer, signs into your apps, and keeps working after you close your laptop. It answers emails, updates a sales database, processes invoices, or files a software bug.

We looked at it in a recent episode, when three engineers used it to build a company on a live stream. That episode asked whether the bots could run a business, not just build one. Now there’s a first number to test that against.

Start with the number itself. Bloomberg says Grok Bot had four hundred eighteen thousand weekly users as of September fourteenth, up twenty-four percent from the week before.

Work that backward and the prior week was roughly three hundred thirty-seven thousand. That’s a net gain of about eighty thousand weekly users in seven days.

What we don’t know is what a user means here. The report doesn’t say whether that’s someone who assigned a bot real work, or someone who opened the app once. It doesn’t say how many of them pay, or how many are businesses.

That’s not a criticism of the reporting. It’s the limit of a single slide. A company picks what goes on it.

Bloomberg itself called this modest early demand. And the comparison it offered is telling. Meta’s new Muse agent app has been downloaded about four times as often as the standalone Grok Bot app, according to Bloomberg Intelligence.

Downloads and weekly users aren’t the same measurement, so that isn’t a clean head to head. Muse is also free and aimed at consumers. Grok Bot is sold as a work tool. But it does show where the attention is going right now.

Here’s the context the slide doesn’t show. The price to get Grok Bot fell fast.

At launch on August eleventh, it came only with premium plans. The cheapest individual option was Cursor Ultra, at two hundred dollars a month. Business teams paid one hundred twenty dollars a seat.

Fifteen days later, on August twenty-sixth, SpaceXAI announced that Grok Bot was included with every SuperGrok, Cursor Pro, and Cursor Teams plan. Cursor Pro costs twenty dollars a month. That’s one tenth of the launch entry price.

And Grok Bot comes with its own usage allowance on those plans. So people who already paid for one of those plans got a bot added to their subscription at no extra charge.

That matters for reading the number. The count was taken nineteen days after the big bundle expansion. Some of that growth is people trying something they already had access to. That’s still real interest. It’s just a different thing from people choosing to pay for an agent.

The fair version is this. Four hundred eighteen thousand weekly users is evidence of curiosity at scale. It isn’t yet evidence of a business.

Now the model underneath. In that same episode, we noted that Grok 4.7, long delayed, was expected before the live stream and didn’t show up.

It arrived on September twenty-first, six days after the stream began. SpaceXAI calls it its most capable model for coding and knowledge work.

Two details stand out in the announcement. The company says the model was trained to natively understand the Grok Bot harness. In plain terms, it was built to work inside the agent product, not just answer questions in a chat window.

And the price didn’t change. Developers pay the same per token as for Grok four point six.

The company also published benchmark scores. On a coding benchmark from Cursor, it reports forty-six point three percent, up from forty point four for the previous version. It says the new model leads on an electrical engineering test and a legal agent test.

Those are the company’s own measurements. Outside tests will tell us more. By its own numbers, it still trails at least one rival model on coding.

So put the pieces together. The entry price fell by ninety percent in about two weeks. The bot came bundled into plans people already had. And the model tuned for it arrived after the showcase meant to prove it.

None of that means Grok Bot is failing. It means the company is pushing distribution first and hoping value follows. That’s a common strategy. It’s also why the next numbers matter more than this one.

What to watch next. First, whether SpaceXAI reports paying customers or businesses, not just weekly users. Second, whether it discloses how much work the bots actually complete. Tasks finished would say more than accounts opened.

Third, retention. A weekly user count only means something if the same people come back. And fourth, day four. Thursday Arena, the game from last week’s stream, is still online. The site still doesn’t say who runs it.

A number on a slide tells you who showed up. The number that counts is who keeps paying for the work, week after week, without being reminded the bot is there.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

On September 13, SpaceX flew its 700th Falcon rocket, carrying the final three satellites of SES’s O3b mPOWER network. Lena Ruiz looks at what that number measures: a Falcon 9 launch roughly every two and a half days since February, a booster two flights short of Space Shuttle Discovery’s record, and a company already moving staff toward Starship.

On the afternoon of September thirteenth, a Falcon nine rocket lifted off from Cape Canaveral carrying three satellites for a company called SES. The launch itself looked routine. The number attached to it did not.

It was the seven hundredth Falcon rocket SpaceX has ever flown. Six hundred eighty-seven Falcon nines, and thirteen Falcon Heavies.

The booster underneath it was flying for the twenty-ninth time. Less than nine minutes after liftoff, it was standing on a drone ship in the Atlantic, waiting for a thirtieth.

This is Tesorb Signal, and I’m Lena Ruiz. Today we’re asking what seven hundred launches actually measure: the pace, a reuse record one booster is closing in on, and why SpaceX’s attention is already shifting to a different rocket.

Start with speed. SpaceX flew its six hundredth Falcon nine on February fourteenth this year, a Starlink mission from Vandenberg in California.

By September thirteenth, the Falcon nine count stood at six hundred eighty-seven. That’s eighty-seven more in seven months, roughly one every two and a half days.

Here’s another way to see it. On August twenty-fifth, SpaceX flew its one hundredth Falcon nine of twenty twenty-six. In all of twenty twenty-three, the entire Falcon family flew ninety-six times.

So this year passed that full-year total with more than four months still left on the calendar.

Most of those launches carry SpaceX’s own Starlink satellites. That’s what makes the seven hundredth one worth a closer look, because it didn’t.

The three satellites on board belonged to SES. They were the last pieces of a network called O3b mPOWER: thirteen Boeing-built satellites in medium Earth orbit, selling broadband to businesses and to government and military customers.

Each of the final three weighs about nineteen hundred kilograms without fuel. They won’t start working right away. According to Spaceflight Now, they’re expected to enter service in mid twenty twenty-seven, after maneuvering into position and testing in orbit.

That’s a quiet detail about where SpaceX sits. A company that sells broadband from space, in a market where Starlink also sells, still bought its ride on a Falcon nine.

Now the hardware. Falcon’s real advantage has never been the count alone. It’s how many times the same booster flies.

The booster on the seven hundredth launch, known as ten eighty, was on its twenty-ninth flight. Its record includes four missions to the International Space Station, two of them carrying astronauts, the European Space Agency’s Euclid telescope, and twenty-two batches of Starlink satellites.

But the fleet leader is a different booster, ten sixty-seven. On August twenty-fifth, it flew for the thirty-seventh time, a Falcon record. It entered service in June twenty twenty-one.

Launch trackers keep reaching for one comparison. NASA’s most-flown space shuttle orbiter, Discovery, flew thirty-nine orbital missions. It entered service in nineteen eighty-four and retired in twenty eleven.

Booster ten sixty-seven is two flights short of that number, after a little over five years in service. As of this week, it hasn’t flown a thirty-eighth time.

The comparison has limits, and they matter. Discovery went all the way to orbit and back, with astronauts aboard. A Falcon booster is only the first stage. It lands about eight and a half minutes after liftoff.

So matching Discovery’s number would be a milestone for reusing hardware. It wouldn’t mean the two machines did the same job.

The landings tell the same story at fleet scale. When the six hundredth Falcon nine flew in February, SpaceX counted its five hundred seventy-first booster landing. By August twenty-fifth, that count had reached six hundred fifty-four.

That’s eighty-three more landings in a little over six months.

There’s one more detail in Spaceflight Now’s report on that record flight. As of late August, SpaceX had flown forty-four of this year’s seventy-six Starlink missions from Vandenberg in California. That’s more than half.

Spaceflight Now also reported in late August that the company’s workforce at Cape Canaveral is being refocused on Starship preparations. And that brings us to the rocket SpaceX is building to take over much of what Falcon does.

On September twenty-first, the Super Heavy booster for Starship Flight fourteen rolled to the launch pad at Starbase, Texas. Two days later, the ship was stacked on top.

The plan is the first fully orbital Starship mission, deploying twenty-six next generation Starlink satellites. SpaceX says each will add one terabit per second of capacity, and that the batch equals roughly ten times what a single Falcon nine launch of today’s smaller Starlink satellites delivers.

That is the company’s figure. It hasn’t been shown in orbit yet.

When we previewed this flight in August, the target was late August, and the plan was to catch the ship with the launch tower. Both have changed. The ship is now set to splash down in the Pacific instead, the target is no earlier than September twenty-eighth, and the flight still needs its launch license from the FAA.

Why does Falcon matter to that story? Because Falcon’s numbers are the bar. A rocket that flies every couple of days, on boosters with thirty or more flights, is a hard act for any replacement to follow, including SpaceX’s own.

Starship Flight fourteen is the next marker. If it flies on the new date and the satellites deploy and start raising their orbits, SpaceX says they could carry traffic as soon as a few weeks after launch.

Booster ten sixty-seven is the quieter clock. Two more flights would tie Discovery’s thirty-nine. Three would pass it.

And the Falcon count itself may be the most telling number. If Starship starts carrying Starlink, Falcon launches could level off even as SpaceX puts more into orbit. That would be a success that looks like a slowdown.

Seven hundred launches made the Falcon feel ordinary. The next test is whether SpaceX can make something far bigger feel just as ordinary.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.