TorontoStarts

SpaceX Reportedly Showed Investors a Grok-Powered AI Device Before Its IPO

Silicon Scoop · August 19, 2026

The Wall Street Journal says SpaceX showed investors a Grok-powered AI device before its IPO, while Musk denies the company is developing a phone. This fact-checked article separates verified reporting from speculation. https://torontostarts.com/2026/08/19/spacex-ai-device-prototype/

With Craig Major and Jon Irwin

Open audio in your browser
SpaceX Reportedly Showed Investors a Grok-Powered AI Device Before Its IPO

Estimated reading time: 8 minutes

Key Takeaways

  • SpaceX reportedly showed selected investors and other stakeholders a handset-like AI device before its June 2026 initial public offering.
  • The public did not learn about the prototype until July 1, nearly three weeks after SpaceX shares began trading.
  • The Wall Street Journal reported that the early-stage device used a proprietary operating system, incorporated xAI technology and contained a Qualcomm Snapdragon processor.
  • Elon Musk disputed the report, calling it “utterly false” and saying SpaceX was not developing a phone.
  • There is no announced product name, price, release date or guarantee that the reported prototype will become a commercial product.

The Device Story Arrived After the IPO—But the Demonstration Reportedly Happened Before It

SpaceX is already a public company. Its shares began trading under the ticker SPCX on June 12, 2026, following an offering priced at $135 per share. The offering formally closed on June 15 after the underwriters exercised their full option to purchase additional shares. SpaceX said it sold a total of 638,888,888 Class A shares and raised approximately $85.7 billion in gross proceeds.

That timing matters because the story of SpaceX’s mysterious AI device did not become public until July 1.

According to The Wall Street Journal, SpaceX had shown selected investors and other stakeholders an early prototype ahead of the IPO. In other words, the reported demonstration occurred while SpaceX was still privately held and courting prospective shareholders, but the wider public learned about it only after the stock-market debut.

That is the accurate premise: this is not a newly announced post-IPO product. It is a reported pre-IPO prototype that surfaced publicly after the IPO—and whose existence Elon Musk disputes.

What The Wall Street Journal Reported

The Journal described a device that resembles a slim handset but is intended to rethink how a person interacts with artificial intelligence. According to the report, the prototype:

  • was slimmer than an iPhone;
  • ran a proprietary operating system;
  • incorporated AI technology from xAI;
  • used a Qualcomm Snapdragon processor; and
  • remained at an early stage, with no guarantee that it would enter production.

Those details should all be treated as reported claims, not specifications confirmed by SpaceX, xAI or Qualcomm. SpaceX has not announced the device, published a technical sheet or demonstrated it publicly. There is also no verified name, price, release date, battery specification, connectivity plan or list of consumer features.

The description “handset-like device” is important. It may resemble a smartphone, but the Journal did not report a conventional phone launch. Whether it would support cellular calling, replace an iPhone or Android device, connect directly to Starlink, or operate mainly as a dedicated AI interface remains unknown.

Musk Denies SpaceX Is Developing a Phone

Musk responded to the Journal’s story by calling it “utterly false” and stating, “We are not developing a phone.” Reuters reported the denial and noted that SpaceX, xAI and Qualcomm had not publicly confirmed the reported prototype.

This leaves a direct and unresolved conflict:

  • The Journal says sources familiar with the investor presentations described a handset-like prototype shown before the IPO.
  • Musk says SpaceX is not developing a phone.

It is possible to imagine distinctions between a “phone” and another kind of AI device, but there is no public evidence proving that this semantic distinction explains the disagreement. The responsible conclusion is simpler: a reputable publication has reported the prototype, Musk has denied the phone claim, and neither side has provided the public with enough evidence to resolve the dispute conclusively.

That uncertainty is part of the story—not a detail to be edited out.

Why a Grok-Centred Device Would Make Strategic Sense

Even without a confirmed product, the reported concept fits a recognizable strategic pattern.

Apple and Google control the two dominant mobile operating systems and the app stores through which most consumer software reaches smartphones. A company that wants AI to become the primary interface—not merely another app—has an incentive to control more of the hardware and operating-system stack.

For Musk’s companies, the potential pieces are easy to identify:

  • Grok provides the AI interface and model layer.
  • xAI’s computing infrastructure handles demanding cloud-based AI workloads.
  • SpaceX brings large-scale engineering and manufacturing experience.
  • Starlink provides a global connectivity business.
  • X supplies a consumer platform, identity layer, payments ambitions and real-time information network.

A dedicated device could, in theory, connect those assets and reduce dependence on Apple and Google. However, the key phrase is in theory. The Journal’s report does not establish that the prototype will use a direct Starlink connection, that Grok inference will run locally, or that SpaceX has approved a commercial strategy for the product.

The existence of a Snapdragon processor would make some degree of on-device AI technically plausible. Qualcomm’s recent premium processors include neural-processing hardware designed to accelerate AI workloads. But a chip family alone does not reveal which model would run on the device, how large that model would be, what would remain cloud-based or whether Starlink would play any direct role.

The Financial Context: What SpaceX Investors Actually Bought

The device appeared in investor discussions during a pivotal transformation for SpaceX. The company acquired xAI in early 2026 and went public as more than a launch-and-satellite business. Its IPO materials presented launch services, Starlink connectivity and artificial intelligence as major parts of a much broader technology platform.

SpaceX’s public filings also show why the AI story attracts both excitement and scrutiny.

For 2025, SpaceX reported approximately $18.67 billion in revenue and a $4.94 billion net loss, according to reporting on its IPO filing. Starlink generated $4.42 billion in operating income, while aggressive investment in AI infrastructure placed substantial pressure on the consolidated business. Reuters reported that the AI segment posted an operating loss of about $6.4 billion for 2025, while AI-related spending accounted for roughly 61% of SpaceX’s $20.74 billion in capital expenditures.

The original version of this article incorrectly described $619 million as a Space-segment operating loss. SpaceX’s final prospectus instead reports that, for the three months ended March 31, 2026, the Space segment generated approximately $619 million in revenue and recorded an operating loss of approximately $662 million.

These numbers help explain the investor debate:

  • Starlink is a proven earnings engine within the company.
  • Space launch remains strategically important but capital-intensive.
  • AI represents a potentially enormous market, but it requires extraordinary spending and has not yet demonstrated the same profitability.

A Grok-centred device could eventually become one route into consumer AI, but that is analysis—not a revenue stream disclosed by SpaceX. No credible public report has established sales targets, production volumes, subscription pricing or financial projections for the reported prototype.

The IPO Timeline, Correctly Stated

Here is the chronology without the confusion:

Date What happened
June 4, 2026 SpaceX began its formal IPO roadshow.
June 11, 2026 The offering was priced at $135 per share.
June 12, 2026 SPCX shares began trading on Nasdaq and Nasdaq Texas.
June 15, 2026 SpaceX announced the IPO had closed, with approximately $85.7 billion in gross proceeds after the full overallotment option was exercised.
July 1, 2026 The Wall Street Journal published its report about the pre-IPO prototype; Musk publicly disputed the story.

The Journal has not publicly disclosed the exact date on which the prototype was shown. Its reporting places the presentation before the IPO. Therefore, it is safe to say the device was reportedly shown during the pre-IPO period, but not to assign it a more precise demonstration date without further evidence.

SpaceX’s own IPO closing announcement confirms that the offering had been completed weeks before the device story became public.

Why Dedicated AI Hardware Is Such a Difficult Market

The idea of an AI-native consumer device is not new, and recent history provides a warning.

Humane’s AI Pin promised to move common digital tasks away from the smartphone and into an always-available AI assistant. Rabbit’s R1 offered a dedicated handheld interface for AI services. Both attracted enormous attention, but neither established that consumers urgently needed a separate device for tasks their existing phones could already perform.

That history creates a simple challenge for any SpaceX or xAI product: What can this device do that a Grok app on a smartphone cannot?

To succeed, it would likely need to deliver clear advantages in several areas:

  • faster and more natural AI interaction;
  • useful functionality that does not depend on opening individual apps;
  • acceptable battery life and thermal performance;
  • reliable privacy and security controls;
  • a compelling connectivity advantage; and
  • a price or subscription model consumers consider worthwhile.

SpaceX would bring substantially greater resources than earlier AI-device startups, along with access to Grok, Starlink and a huge engineering organization. Those advantages improve the strategic possibilities, but they do not eliminate the brutal economics of consumer hardware, supply chains, distribution, customer support and platform development.

Building a proprietary operating system would raise the stakes further. An operating system needs more than an AI assistant: it requires security updates, hardware drivers, developer tools, application support, accessibility features, privacy controls and a reason for users to leave established ecosystems. The reported prototype may be an experiment designed to explore those questions rather than evidence that SpaceX has solved them.

Stop Leaving Money on the Table With Manual Work

While the world’s largest technology companies race to build AI into hardware, infrastructure and operating systems, many small and midsized businesses are still copying data, chasing approvals, sending repetitive follow-ups and assembling reports manually.

Flowgrammer helps businesses replace that repetitive work with practical automations and AI-enabled workflows—so teams can spend more time serving customers and growing the business.

👉 See how Flowgrammer automates your business

What Happens Next

The next meaningful development will not be another round of speculation. It will be verifiable evidence.

Watch for one of the following:

  1. A formal product announcement. SpaceX or xAI could reveal the hardware, its intended market and whether it will enter production.
  2. Regulatory or supply-chain evidence. Certifications, manufacturing contracts, developer documentation or additional corporate filings could indicate that the project has progressed beyond a prototype.
  3. Further confirmation from independent reporting. Additional sources or images could clarify what investors were shown and how advanced the project was.
  4. No further activity. Like many internal prototypes, the concept could be revised, absorbed into another product or abandoned without a public launch.

Claims about a developer edition, an enterprise release, direct-to-Starlink connectivity or an acquisition of a consumer electronics company remain speculation unless supported by new reporting.

The Bottom Line

The most defensible version of the story is also the most interesting: SpaceX reportedly showed selected investors a Grok-powered, handset-like prototype before its record-setting IPO, but the public learned about it only after SpaceX had gone public—and Musk denies that the company is developing a phone.

The reported hardware could signal an ambition to control the entire AI experience, from models and computing infrastructure to connectivity, operating software and the device in a user’s hand. It could also be an exploratory prototype that never becomes a product.

Until SpaceX provides evidence or additional reporting resolves the contradiction, this should be covered as a disputed report about an early-stage concept—not as a confirmed SpaceX phone launch.

Frequently Asked Questions

1. Did SpaceX show the AI device before or after its IPO?

The Wall Street Journal reported that SpaceX showed the prototype to selected investors and stakeholders before the IPO. The story became public on July 1, after SpaceX shares had started trading on June 12 and the offering had formally closed on June 15.

2. Is SpaceX now a public company?

Yes. SpaceX began trading under the ticker SPCX on June 12, 2026. The company announced the offering’s completion on June 15.

3. Has SpaceX officially announced an AI phone?

No. SpaceX has not announced a phone or consumer AI device. The available information comes from the Wall Street Journal’s sources, and Musk has disputed the report.

4. What did Musk deny?

Musk called the Journal report “utterly false” and said SpaceX was not developing a phone. He has not publicly supplied a detailed alternative explanation of what, if anything, investors may have seen.

5. What hardware is reportedly inside the prototype?

The Journal reported that the device used a Qualcomm Snapdragon processor. The specific Snapdragon model and its performance characteristics have not been disclosed.

6. Does the device run Grok locally?

That has not been confirmed. The reported device incorporates xAI technology, but there is no reliable public specification showing which AI tasks would run on the device and which would require cloud computing.

7. Does it connect directly to Starlink?

No direct Starlink capability has been publicly confirmed. Such integration would fit the broader corporate strategy, but it should not be presented as a reported specification without additional evidence.

8. Does the device have its own operating system?

According to the Journal, the prototype ran proprietary operating software. SpaceX has not publicly identified the operating system, its technical foundation or its intended developer ecosystem.

9. When will the device go on sale, and what will it cost?

There is no announced release date or price. The reported prototype may change substantially or never enter production.

10. Why does the prototype matter if it never ships?

It suggests that SpaceX may at least have explored a consumer interface connecting xAI technology with its broader ecosystem. It also raises questions about how the company presented its post-IPO growth ambitions to investors. But without confirmation, it cannot be treated as proof of a finalized consumer-hardware strategy.

Primary Reporting and Documents

Episode transcript

Read the full transcript

Craig: spacex reportedly showed investors a secret rock powered device before raising nearly 85,700,000,000 in the biggest ipo of the year, but elon musk says the entire phone story is utterly false. either spacex is quietly building a new gateway to musk's technology empire or investors were shown an experiment that may never leave the lab. so the wall street journal reports that spacex showed a selected, set of investors, a slim handset like ai prototype before its june ipo. it reportedly used xai technology, proprietary operating system, and a qualcomm snapdragon processor, but musk denies spacex is developing a phone. so if the report is accurate, musk may be exploring a device that unites groc, starlink, x, and proprietary operating system reducing his dependence on apple, google, and their app stores. what do you think is going on here, john?

Jon: yeah. believe, i don't believe musk for a second. i i i much i'm much more apt to believe the wall street journal. so, yeah, i i there there's something that he's working on. so, yeah, so it was described as a handset like device specifically designed around ai interactions. and so the reported characteristics included thinner than an iphone, obviously, a qualcomm snapdragon processor, their own proprietary operating system, native integration with xai models, which i wouldn't buy just for that right away. again, like you said, possible starlink, connectivity and an ai first user interface. and so the question is is, why build dedicated ai hardware? well, today's smartphones, they were designed around touch screens. right? it's crazy. i'm i still miss my blackberry bold and all those. but, generative ai changes that assumption that you should be using touch screens because modern ai systems can understand natural language. they can remember conversations. they can complete workflows. they can retrieve information, and they can automate actions. and so traditional operating systems were never designed around continuously running intelligent agents. and so an native operating system could instead make the ai itself the home screen, which i don't know if i'm ready for. do you think you're ready for an ai centric os that kind of you just talk to instead of touching and opening up apps and stuff like that? what are your thoughts?

Craig: i think it's the future. the more i work with my ai agents and the more ai agents i have, the more i want more channels to talk to them so i can get them to do more things at once. so, yes, you know, some of us are ahead of the curve on that. i won't say i'm ahead. there's lots of people farther ahead than me, but the point i'm making is, yes, i can see this. now you don't believe a word elon musk says, and that's perfectly fine. i think when all of this happens, everybody's gotta do what they gotta do to protect their company, their reputation, their stock, whatever. so that's fine. but elon specifically says he denies they're developing a phone. he didn't say we're not developing a hardware device. he didn't say, so i just wanna yeah.

Jon: so i

Craig: just okay.

Jon: okay. okay. you're you're gonna yeah. and that makes sense that elon would see something that specific. yeah. for sure.

Craig: to, you know, twist it so he can come back and say, hey. guess what? it's not it you can you can use it for a phone, but it's an ipad like device or it's an first controlled deck that can you know, talk to the data center, talk to all your cloud stuff, talk to your agents, use local based hard, ai that doesn't cost a bunch of money. now this is where we're heading. and i don't know if they're gonna do it here or not because we're just speculating here. but we're heading to open models or free models that run on your hardware and only go into the cloud for the bigger runs because they need to start distributing the cpu again. so if you're saying, hey. what's the weather like today? should i grab an umbrella? well, let's run that on local ai. you're not, doing a stock market analysis or creating you know, there's a bunch of stuff you can start taking these jobs. and when you start looking at, like, chatgtp's analysis on what people are using it for, most of it is just common everyday tasks, not actual work. so when you start saying that, okay, these data centers are costing billions and billions to run, but we have a plan to take a big chunk of the traffic that's coming to us off and doing it local so we can focus on the big stuff. that means our data centers get opened up and can serve more people and do all that stuff. and people can't afford to be paying $50 per million output tokens and more, just to get, you know, weather reports or, you know, jokes or whatever they're using. so i think there's a change here. we openai is doing something. apple's suing them for doing something. yeah.

Jon: we can cover that next time. yeah.

Craig: and, assuming that they took all these hardware people and they're taking design principles and stuff. so everybody's on a rush to do this ai centric hardware. but we've seen it fail with, what, the humane ai pin, the rabbit r one. but just because the rabbit r one failed doesn't mean it was the wrong idea. i was using microsoft tablets based on windows long before the ipad came out. the idea was good. the hardware wasn't there yet.

Jon: yeah. and yeah. i yeah. i agree. there is a move to something, and who knows what it is. like, when we made the leak from, no. was it touch screen? yeah. i guess they were no. they weren't yet. from the blackberry bowl to the iphone. that was a huge leap, and there was a lot of, negativity around it, and, it took a while. so i think we're in that era. i again, i guess because i can't picture what it would look like, i'm just, and i'm getting older. i'm getting set in my ways. so, yeah, it'll be interesting. i know just a sidebar. commodore just released go they're going backwards. they released a flip phone that, that, gets rid of all social media. it's a smartphone running on linux, but runs most android apps. but, i kinda like that. and there's companies releasing, blackberry type devices because people want the keyboard, the hardware keyboard back. it's going to be interesting to see what the next popular iteration or generation of device that would be ai centric will look like. is it going to be elon that releases it? is it gonna be apple? i doubt apple because they're not as innovative as they used to be. again, we can debate that anytime, but, yeah, it's interesting. what do you think? who do you have an idea in mind on who's going to release, the next iterative device that people are gonna fall in love with?

Craig: it's hard to say. i agree with apple's lost mo a lot of their design mojo. so we'll have to see, you know ipad was their last big device, and that was really still a steve jobs thing, wasn't it? yeah. and the apple watch, i think, was a tim, issue a tim device.

Jon: wasn't that johnny ives still?

Craig: well, i think it was tim and johnny ives. but, i mean, the point is i think it it that was not a, steve jobs device. it was in the new era after steve jobs. it's that's the only big device that's been popular after steve jobs. that's what all i'm trying to get at. so i don't know if they got anything left in the tank. everybody's trying to move this way when every one of my computers have some sort of ai agent on it. claude, openai, whatever now have their apps now connect to their agents that are on your desktop that yeah. run and control your desktop. so you can literally just say install this, do this, whatever, and it'll just do it on your machine. and once the world starts understanding that, the game will change. but because, like, literally, i'm setting up new client machines. one of the first thing i do is set up their ai agent on it to help me set up and document everything we do. so we've got it all configured and ready to go. and then having that connectivity, so no matter where i am, i can continue to work. i can get my status updates, reports. it has all that stuff. they're all moving towards that. they all have apps. they all have this thing now. so why not now move towards a hardware that can kinda separate, some of the requests? not everything has to go. not everything has to have all the context. not everything has to cost you. and why not put that in your own ecosystem where you have your own agent that's local on your hardware that controls all these other things. it is the future because people are realizing that he who controls the harness controls the world. and as soon as, they can build harness in the hardware that's on your device. they're removing the moat of the comp or creating the moat. the competition can't necessarily jump in. if you're running a spacex device, you're not gonna be getting the same, stuff as an openai experience or maybe an opencloud experience.

Jon: yeah. the one thing about this device that could be a differentiator, is the starlink connection. right? because, if you have a future ai centric device connected through starlink, it could theoretically provide global connectivity yes. remote connectivity, disaster communications, maritime coverage, and aviation connectivity. so, like, you'd because, like, oh, by the way, never take via rail. their wi fi sucks. and it it sucked for years. but, when i was on the train, i could barely use mobile data. i couldn't use wi fi. maybe having starlink could have been a much better option.

Craig: yeah. i think moving to satellite and all that kind of stuff. and, like, when i say he who controls the harness controls the world, humans aren't spending ai credits anymore. it's agents. and the agents are controlled by the harness that says, you know, do this, do that. here's what you do. here's my skills. here's the loop. my heart beats every 30 minutes. so when i say things like they wanna control the harness, the ai spend is gonna come from your device, not you. yeah. so meaning, you're gonna say, hey. go do this thing, and then you're gonna go for a walk, and your device is gonna go spend an hour doing whatever, booking tickets or break creating a report or whatever. it's spending the credits and spending the money. that's why they wanna control that device and that harness. and, the more they have a clean throughput from the hardware all the way to their back end ai, the better experience they can provide someone. so it's an interesting idea, but it's an expensive world to get into. elon could bypass this. so we've talked about this in the past where i thought it was a mixed mistake for openai to get into phones. but if they can direct to starlink and they can just bypass every, cell phone provider out there, at and t and all that stuff, then maybe this is, something they can get into.

Jon: maybe this is something viable. for sure, i agree. yep.

Craig: this could be the first clip that must attempt to control the complete ai stack from satellites and supercomputers to the device in your hand. but today, it remains a disputed prototype, not a spacex phone. if spacex launched a grok device with a global starlink connectivity, would you replace your smartphone, or would this become an ai gadget looking for a reason to exist? let us know what your thoughts are in the chat below.

More from Silicon Scoop

All episodes