On November 1, 2026, a report from The Information, carried by Reuters and summarized by outlets including Engadget, revealed that the
Meta humanoid robot Max is real hardware: a prototype home robot built in about three months with the help of Shanghai-based robotics firm Fourier, with a second version, Max 2, already in development.
Key fact: Meta says the project is internal research and that it has "no plans to sell these units" — so you cannot buy a Max, and no launch date exists. This guide separates what has actually been reported and verified from what is still rumor, walks through Meta's full robotics journey from 2025 to today, and then broadens out to the entire 2026 home-robot landscape — what today's humanoids can and cannot do, what they cost, and whether you should wait for one.
The Story That Broke on November 1, 2026
The news arrived on a Saturday, October 31 in the United States and November 1 in much of the rest of the world. The Information, citing people familiar with the project, reported that Meta had built a humanoid robot prototype internally codenamed Max together with Fourier, a Chinese robotics company based in Shanghai. The key takeaway: this is the first concrete evidence of actual Meta robot hardware, but the details come from a single paywalled report that Reuters said it could not immediately verify. No photos have been published, neither Meta nor Fourier has posted a statement in its online newsroom, and no independent demonstration video exists.
That single-source nature matters, and any honest article about Max has to say so up front. The Information's reporting was written by journalist Jyoti Mann and was relayed by Reuters, Engadget, The Next Web, Humanoids Daily and a wave of tech outlets within hours. The facts below are all attributed to that reporting chain. Where a detail has been confirmed by a second source — for example Meta's own spokesperson statement — we say so. Where something remains unverified, we label it clearly.
The basic outline is striking. According to the report, Meta and Fourier completed Max in roughly three months — an extremely fast cycle for humanoid hardware. Meta's vice president of product for robotics, Li-Chen Miller, reportedly spent almost two months at Fourier's Shanghai factory working on the prototype. In late August 2026, Meta showed Max to a group of its own employees at an internal robotics summit. And development of a second version, Max 2, began in September 2026. The most important fact for readers: Meta told The Information that "this effort with Fourier is focused on internal research and experimentation, and we have no plans to sell these units."
Max's Reported Specifications: What the Hardware Includes
Here is everything the reporting chain attributes to Max, assembled in one place. Treat each line as reported, not independently verified, because no third party has examined the machine.
| Attribute | Reported detail | Status |
|---|
| Internal codename | Max | Reported by The Information |
| Height | About 1.5 meters (roughly 5 feet) | Reported, consistent across outlets |
| Weight | About 48 kilograms (106 pounds) | Reported, consistent across outlets |
| Compute | NVIDIA Jetson Thor chip | Reported by The Information |
| Hands | Five-fingered hands able to grasp small objects such as bottles and toys | Reported |
| Face | Display that mimics human facial expressions | Reported |
| Audio | Built-in speakers, reportedly dual speakers in the chest | Reported |
| Build time | Roughly three months with Fourier | Reported |
| Sale status | Not for sale; internal research only | Confirmed by Meta spokesperson |
Key takeaway: Max is roughly the height and weight of a small adult — 1.5 meters and 48 kilograms — which makes it lighter than most rival humanoids and, in theory, easier to move around a home, but no independent benchmark, runtime figure, or task demonstration has been published.
Beyond the physical specs, the report describes Meta's design requirements for how Max should interact with people. These are internal targets rather than proven capabilities: recognizing a human voice from about five meters away, interpreting a wave or gesture from about three meters, maintaining eye contact during conversation, and signaling its internal state — listening, thinking, speaking, or uncertain — so people know what the robot is doing. It is important to be precise here: these are development targets described in the report, not independently verified capabilities that Max has demonstrated in testing. Interesting Engineering made exactly this distinction in its coverage, and it is the right way to read the story.
Fourier: The Partner That Built Max With Meta
Fourier is a robotics company headquartered in Shanghai, founded in 2015 and formerly known as Fourier Intelligence, according to its Wikipedia entry and company materials. The key fact: Fourier is not a newcomer — it started its humanoid robot project in 2019 and already manufactures its own humanoid robots, the GR-1 and the newer GR-3.
The GR-3 is Fourier's published flagship. Its documented specifications describe a 71-kilogram platform with 55 degrees of freedom and a three-kilogram single-arm payload. Those figures describe the GR-3, not Max, and the distinction matters: Max was reportedly built on hardware foundations related to Fourier's first three product generations, with Meta and Fourier redesigning the height proportions, head, display screen and camera and adding new hand and foot modules. The two robots are different machines, and the Max specifications remain independently unverified.
Why would Meta partner with a Chinese company instead of building everything in-house? Meta's own spokesperson gave the industry-wide answer in the report: "We face the same realities as the rest of the industry when it comes to sourcing components that are not readily available domestically." The key takeaway: even the world's largest tech companies still depend on specialized robotics hardware supply chains, and Fourier gave Meta working humanoid hardware in about three months instead of the years a from-scratch build would take. Reporting from earlier in 2025 also noted that Meta had talked with other potential partners, including Unitree and Figure AI, before Fourier became the partner named in the Max report.
Meta's Road to Robots: The Full Timeline From 2025 to Today
Max did not appear out of nowhere. Meta has been moving toward home robotics step by step for nearly two years, and the timeline shows a company that is serious about this as a long-term bet.
| When | What happened | Source |
|---|
| February 2025 | Bloomberg reported that Meta formed a humanoid robotics group inside Reality Labs, led by former Cruise CEO Marc Whitten; the goal was robots that could do household chores | Bloomberg, via TechCrunch and GeekWire |
| Early 2025 | Meta reportedly held early talks with Unitree and Figure AI about building prototypes together | TechCrunch, citing Bloomberg |
| September 2025 | CTO Andrew Bosworth confirmed the robotics group's creation to The Verge, crediting the initiative directly to CEO Mark Zuckerberg | The Verge, via Digitimes |
| Late 2025 | Li-Chen Miller, formerly head of Meta's smart glasses division, became the first product manager for the Reality Labs robotics team | Business Insider, via Digitimes |
| May 2026 | Meta acquired Assured Robot Intelligence (ARI), a San Diego startup of about 20 people building full-body robot control models and touch sensors; the team joined Meta Superintelligence Labs | Bloomberg, via Seoul Economic Daily |
| August 2026 | Wired reported Meta was testing robots inside its own data centers, plugging in cables and resetting servers | Wired, via AIStockWire |
| Late August 2026 | Max was shown to some Meta employees at an internal robotics summit, its first reported physical demonstration | The Information |
| September 2026 | Development of Max 2 reportedly began | The Information |
| October 26-27, 2026 | NVIDIA announced a $3 billion investment in Meta's equity offering and an expanded partnership on the Muse AI agent family | NVIDIA newsroom, via Unite.AI |
| November 1, 2026 | The Information's report on Max became public via Reuters and tech press | The Information, via Reuters |
Key takeaway: Meta's robotics push has been public and deliberate since February 2025 — hardware team, key hires, a startup acquisition, internal demos — and Max is the first reported physical result of that nearly two-year effort. The strategy also has a software-first flavor: Bosworth has emphasized that the primary challenge is software development rather than mechanical design, and Meta's reported broader ambition is to license its robot software platform to third-party manufacturers, similar to how Google's Android works for phones.
Max 2: The Second Prototype Already Underway
According to the report, Meta and Fourier did not stop with Max. Work on a second prototype, Max 2, began in September 2026, while the first version was barely a month past its internal debut. The key takeaway: a follow-up starting this quickly signals that Max was treated as a learning platform from day one, with the lessons feeding straight into a better second build.
The reported Max 2 upgrades are practical rather than flashy, and they read like a list of things the first prototype got wrong or left unfinished. The torso is being redesigned, the hands are being improved for finer object manipulation, a projector is being added to display visual content, and additional sensors are being included to help the robot climb and descend stairs. Meta is also reportedly recruiting mass-production engineers in Shanghai and Shenzhen, although the company says there are still no sales plans.
Stairs are worth pausing on. Any home robot that cannot handle stairs cannot be a whole-home helper — it is a single-floor appliance, and in a multi-story house that halves its usefulness. The stair-sensor upgrade suggests Meta's robotics team understands exactly which real-world barrier keeps humanoids out of real homes. No release date or even a target date has been set for any product, according to one of The Information's sources.
NVIDIA Jetson Thor: The Brain Inside Max
The report says Max runs on an NVIDIA Jetson Thor chip, and the choice is revealing. The key takeaway: Jetson Thor is NVIDIA's purpose-built "on-robot brain" for humanoids and physical AI — putting it in Max means Meta wants serious on-device intelligence, not a robot that streams everything to the cloud.
Jetson Thor is the top tier of NVIDIA's Jetson family, built on the Blackwell GPU architecture. The flagship module, the Jetson T5000, delivers up to 2,070 FP4 teraflops of AI compute with a 14-core Arm Neoverse CPU and 128 GB of LPDDR5X unified memory, in a power envelope of roughly 40 to 130 watts. It reached general availability on August 25, 2025, with a developer kit priced at 3,499 US dollars. A mid-tier T4000 module offers up to 1,200 FP4 teraflops with 64 GB of memory.
In July 2026, NVIDIA expanded the family with two smaller modules: the Jetson T3000, rated at 865 FP4 teraflops with 32 GB of memory and about half the size and power of the T5000, and the Jetson T2000 at 400 FP4 teraflops with 16 GB, both aimed at mainstream robotics and scheduled for the first quarter of 2027. One technical note for readers comparing numbers: Thor's headline figures are in FP4 precision, not the INT8 TOPS used for the older Jetson Orin line, so the two cannot be compared by simple arithmetic.
Which exact Thor module Max uses was not specified in the reporting. But the broader point stands: the same compute platform Meta is using for Max is also the platform NVIDIA positions for the whole humanoid industry, from factory robots to home assistants. On-device compute of this class is what would let a home robot run vision-language-action models locally — understanding a spoken request, seeing the room, and planning a grasp — without sending video of your home to a data center for every decision. For a company as privacy-scrutinized as Meta, that architectural choice is not a coincidence.
The $3 Billion NVIDIA Partnership: Context for the Robot Story
The Max report landed just days after an even bigger piece of news about the same two companies. The key takeaway: on October 26-27, 2026, NVIDIA announced it would invest $3 billion in Meta's equity offering while expanding the companies' partnership on Meta's Muse AI models and agents — so NVIDIA is now both Max's chip supplier and one of Meta's shareholders.
The deal sits inside a much larger financing. Meta is raising up to about $95 to $97.5 billion in equity to fund AI infrastructure, with 2026 capital expenditures expected between $135 and $145 billion. NVIDIA's $3 billion is part of the initial offerings, alongside non-binding cornerstone interest of up to $5 billion combined from NVIDIA, AMD and Broadcom. On the technical side, Meta plans to expand its deployment of NVIDIA's Vera Rubin GPU systems, Vera CPUs, Spectrum-X Ethernet networking and BlueField-4 processors, with the two companies working together to improve training and inference performance for the Muse family of models.
Does the $3 billion investment directly fund Max? The public disclosures do not say that. Analysts at MLQ noted that the confirmed transaction is narrower than the headline suggests: NVIDIA is funding part of Meta's equity raise while the two companies deepen a hardware and software partnership whose purchase volumes and economics remain only partly disclosed. The honest reading: the investment proves how strategically intertwined the two companies have become across AI agents and infrastructure, and Max's Jetson Thor brain sits comfortably inside that alliance — but no filing connects the $3 billion to the robot program specifically.
Muse Gets a Body: Why Meta Wants Humanoids at All
The most interesting line in the whole Max story is not a specification. According to The Information, CTO Andrew Bosworth has described humanoids internally as a physical embodiment of Muse, Meta's consumer AI agent. The key takeaway: Meta does not see Max as a gadget project — it sees a future in which its AI agent stops living only on your phone and starts acting in your physical home.
Muse is the center of Meta's consumer AI push. Meta introduced the Muse assistant as a personal agent for delegated tasks and planning, with capabilities including booking travel, filling out forms and making purchases, and it runs on the Muse family of AI models. Sessions from personal AI agents have exploded in 2026 — Arkose Labs reported that Muse sessions grew more than 30 times in September alone — as agents moved from reading the web to transacting on it.
The logic of giving Muse a body is straightforward. An agent that can book your flight is useful; an agent that can also see that your living room is messy, walk over, and pick up the toys is a different category of product. Meta's long-term goal, per the reporting, is robots that take on household chores. But here is the gap between the vision and the hardware: today's AI agents are impressive at digital tasks and still primitive at physical ones, and nothing in the Max report demonstrates that the robot can reliably do a single household chore unsupervised. The embodiment story is a strategy statement about where Meta wants to go, not a description of what Max can do today.
Why Meta Says "No Plans to Sell": The Strategy Behind the Prototype
It can seem strange that a company would build a humanoid in three months and immediately rule out selling it. The statement is deliberate, and there are at least four plausible reasons for it.
First, research before product. The key takeaway: Meta's spokesperson called the Fourier work "internal research and experimentation," which in industry language means the company is still learning what a home robot even needs to be — the prototype is a question, not an answer. Selling a half-finished humanoid would create support, safety and reputation risks far larger than any early revenue.
Second, the software-platform play. Reporting from 2025 on Meta's robotics strategy said the company was considering licensing its robot software to third-party manufacturers, like Google's Android model for phones. In that model, Meta does not need to sell a robot at all — it needs reference hardware that proves the software works, and manufacturing partners that ship it.
Third, the supply-chain reality. The spokesperson's line about sourcing components "not readily available domestically" is a quiet admission that the bill of materials for a humanoid is not yet a mass-market bill of materials. Rare-earth magnets, dexterous hands and high-torque actuators remain constrained and expensive across the industry.
Fourth, timing. With Meta Superintelligence Labs absorbing the ARI acquisition's full-body control models, and with Max 2 already redesigning the torso and hands, the company clearly believes its first prototype is not the right product yet. The honest summary: "no plans to sell" is both a genuine statement of current intent and a smart way to avoid being held to a launch date for hardware that is not ready.
The 2026 Humanoid Race: Who Is Actually Ahead
Max is one entry in the busiest year humanoid robotics has ever had. To judge Meta's position honestly, you need the full field. The key takeaway: as of late 2026, no company sells a humanoid that can reliably do your housework — the only home-focused robot you can actually order is 1X's NEO, and even that is a US-only pre-order with no confirmed customer home delivery.
| Robot | Company | Status in late 2026 | Listed price |
|---|
| Max / Max 2 | Meta with Fourier | Internal research prototypes; Max 2 started September 2026 | Not for sale |
| Optimus | Tesla | Q3 2026 letter says production began at Fremont; first builds used internally for training data; no price or sale announced | Not for sale |
| Figure 03 | Figure AI | Launched October 2025; 350+ units delivered by April 2026; logistics sequencing at BMW Spartanburg | Not for sale |
| NEO | 1X Technologies | US pre-order; 1X says US deliveries start 2026; the only humanoid sold as a household helper | $20,000 or $499/month |
| G1 / H2 / R1 | Unitree Robotics | On sale in the official shop; 5,500+ humanoid units shipped in 2025 | R1 from $4,900; G1 from $13,500; H2 $29,900 |
| Apollo | Apptronik | Pilots with Mercedes-Benz, GXO, Jabil and Google DeepMind; Apollo 2 unveiled June 2026 | Not published |
| Atlas | Boston Dynamics | Production Atlas unveiled at CES January 2026; all 2026 units committed to Hyundai and Google DeepMind | Not for sale |
| Digit | Agility Robotics | Leased to logistics customers; RoboFab facility with 10,000-unit capacity | Leased, not sold |
A few of these deserve a closer look. Tesla's Optimus is the program most often compared to Max. Tesla's Q3 2026 update letter, dated October 21, 2026, says Optimus production began at the Fremont factory and calls the robots from that line its first production-quality machines, with California capacity marked "Early Ramp" and Texas under "Construction." But Tesla's own filings are careful: its FY2025 10-K says the company has "yet to commercialize Bots," early builds go to an internal "Optimus Academy" for training data collection, and Elon Musk said on the April 2026 call that Optimus would probably be useful outside Tesla sometime in 2027. No price and no order page exist. Reporting puts output at several hundred units per week by late summer 2026 — real volume, but roughly ten times lower than earlier hype suggested, with hand assembly (each hand needs 100+ screws fitted manually) named as the bottleneck.
Figure AI's Figure 03 is arguably the most deployed humanoid in the West: 1.68 meters tall, 60 kilograms, 44 degrees of freedom, about five hours of runtime, wireless charging through coils in its feet, and a six-camera vision system. It is doing logistics sequencing work at BMW's Spartanburg plant. The widely repeated "$20,000 Figure" is a long-term price target, not a listed price — the robot is not on sale to the public.
Unitree is the outlier that will actually take your money today. The Chinese company lists the G1 from $13,500, the full-size H2 at $29,900 and the smaller R1 Air at $4,900 in its official shop. The honest caveat: what arrives is a demonstration machine or developer kit, not a butler — Unitree sells no list of household chores its robots can perform, and none of the makers publishes a euro price or confirms EU shipping. Still, with 5,500+ humanoid units shipped in 2025, Unitree is the only company with real consumer-scale sales today.
What Today's Home Robots Can Actually Do
Strip away the marketing videos and you are left with a short, honest list. The key takeaway: 2026's humanoids are genuinely good at a handful of structured tasks and genuinely bad at the unstructured mess of a real home — and knowing which is which will save you from disappointment.
In controlled settings, today's humanoids can walk on flat ground, climb some stairs with effort, pick up known objects from known positions, carry light loads, and follow simple scripted sequences. Figure 03 sequencing parts at BMW is a real deployment: repetitive, well-lit, with objects in predictable places. Agility's Digit moves totes in warehouses. Tesla's Optimus builds collect training data in Tesla's own facilities. 1X's NEO is designed around home tasks like tidying, fetching items and simple organizing, with tendon-driven actuators meant to be safe around people.
The pattern across all of these: the robot succeeds when the environment is prepared for the robot. Factories and warehouses are the natural first home for humanoids because they are structured — flat floors, consistent lighting, objects in fixed locations, no toddlers. A home is the opposite of structured, which is exactly why the home robot is the hardest problem in the industry and the last one likely to be solved.
What They Still Cannot Do: The Honest Limits
Here is the other list, and it is longer. No humanoid on the market or in a lab demo today can be trusted to do these unsupervised in a real home.
- Cooking a real meal. Handling heat, raw food, sharp tools and timing, all at once, in a kitchen the robot has never seen. No published demo shows this end to end.
- Cleaning a bathroom. Wet surfaces, chemicals, tight spaces, and judgment calls about what is dirty. This remains far beyond current manipulation and perception.
- Doing laundry start to finish. Sorting, loading, folding — folding alone is an unsolved robotics research problem for arbitrary clothing.
- Childcare or eldercare. The safety, liability and trust requirements are orders of magnitude beyond what any current system can meet. No maker claims this.
- Working a full day. Published runtimes run about two to five hours depending on the machine; the Unitree G1 manages roughly two hours. A robot that needs charging more often than your phone is not doing your chores.
- Handling the unexpected. A spilled drink, a pet darting underfoot, a door left half open — the long tail of home chaos defeats scripted routines.
- Guaranteeing your privacy. A home humanoid is a walking sensor array: cameras, microphones, depth sensors. No maker has published a full, audited data-handling regime for home deployment.
The key takeaway: if a video shows a humanoid doing something impressive, check whether the environment was staged, the task was scripted, or a human was teleoperating it — most viral robot videos involve at least one of the three, and none of them proves the robot can do the job in your home tomorrow. This applies to Max too: its internal summit demo was for employees, with no public footage and no published task list.
Prices and Timelines: A Reality Check
Money is where robot dreams meet reality. Here is what the numbers actually look like in late 2026.
| Option | What you get | Price | Reality check |
|---|
| Unitree R1 Air | Small humanoid developer platform | From $4,900 | Cheapest way to own a humanoid; a hobbyist machine, not a helper |
| Unitree G1 | Mid-size humanoid, depth camera and lidar standard | From $13,500 | Research and education favorite; shown as backordered in late 2026 |
| 1X NEO | Home-focused humanoid, pre-order | $20,000 or $499/month | US only; deliveries promised from 2026; no confirmed home delivery yet |
| Unitree H2 | Full adult-size humanoid, 1.8 m | $29,900 | Labs and showrooms; not a consumer appliance |
| Tesla Optimus | Not offered for sale | No price announced | Musk's long-term target "under $20,000"; public sales aimed at 2027 |
| Meta Max | Not offered for sale | No price announced | Internal research only; no launch date exists |
The key takeaway: a genuinely useful home humanoid will likely cost between $15,000 and $25,000 when it first appears — roughly the price of a car — and analyst projections put meaningful industry volumes in the late 2020s, not 2026. Bank of America projects about 90,000 humanoid units shipped industry-wide in 2026, scaling to 1.2 million by 2030, while Goldman Sachs projects a $38 billion humanoid market by 2035. These are projections, not guarantees, but they frame the timeline: 2026 is the year of prototypes and pilots, 2027-2028 the years of first real products, and the 2030s the decade of mass adoption — if the technology curve holds.
Safety, Privacy, and Regulation: The Questions Nobody Is Answering Yet
A 48-kilogram machine with five-fingered hands walking around your home raises questions that the industry has barely started to answer. The key takeaway: there is currently no established consumer-safety certification regime for home humanoids, and we could not verify any regulatory filing — such as an FCC filing — specific to Meta's Max, so any claim of one should be treated as speculation.
Start with physical safety. Industrial robots work behind cages; home robots will work next to children and pets. Makers talk about force-limited joints, soft coverings and tendon-driven designs — 1X's NEO uses tendon-driven actuators partly for this reason — but no independent body has published a pass/fail safety standard for a home humanoid the way crash tests exist for cars. Until such standards exist and machines are tested against them, every home deployment is effectively a beta test with the customer's family.
Then privacy. A home robot needs cameras and microphones to function — Max's own design targets include recognizing voices at five meters and reading gestures at three. That means a device in your bedroom and kitchen that sees and hears everything, built by Meta, a company with a long and difficult history on privacy. The questions to demand answers to, before any home robot enters your house: does perception run on-device or in the cloud, is video stored and for how long, who can access it, can you fully disable recording, and has any independent auditor verified the answers? As of this writing, no home-robot maker has published complete, audited answers.
On regulation, the honest position is that the framework is unwritten. Consumer robots will eventually face radio-emission rules, electrical safety standards and data-protection laws in every market they enter — but for Max specifically, our research found no publicly confirmed regulatory filing. Do not believe social-media claims of filings unless they link to the actual regulator's database.
Common Mistakes People Make When Following Robot News
Humanoid robotics is a hype magnet, and the same misunderstandings repeat with every announcement. Avoid these five.
Mistake 1: Treating a prototype as a product. Max is an internal research prototype with no price, no launch date and no sale planned. A prototype proves that a concept can exist; a product proves it can be manufactured, supported and kept safe at scale. The distance between the two is measured in years and billions of dollars.
Mistake 2: Believing every demo video. Teleoperation — a human piloting the robot remotely — is standard practice in impressive demos, and staged environments are the norm. The key takeaway: always ask who was holding the controls and whether the environment was prepared; if the maker will not say, assume the best-case interpretation.
Mistake 3: Mistaking AI concept art for real photos. No real photos of Max have been published, so any image you see claiming to show it is either Fourier's own GR-series hardware or an AI-generated illustration. Honest outlets label illustrations as illustrations — for example, our own free stock gallery clearly marks which images are AI-generated. If a post shows a "photo" of Max without a source, it is not real.
Mistake 4: Confusing a company's valuation with its robot's readiness. Apptronik raising hundreds of millions and NVIDIA investing $3 billion in Meta are financing events. They signal investor confidence, not engineering milestones. Money accelerates development; it does not certify that the robot works.
Mistake 5: Assuming the first mover wins. Unitree sells robots today while Meta and Tesla are still in research and ramp. But the history of technology — from phones to electric cars — shows that early sellers often lose to later entrants with better software and manufacturing. In humanoids, the software for reliable home autonomy does not exist yet at any company, so nobody has an unassailable lead.
Should You Wait for a Home Robot? A Practical Guide
Here is a step-by-step decision framework for a real reader in late 2026. The key takeaway: for almost everyone, the right move is to wait, watch the 2027 product launches, and spend nothing until a robot has a listed price, a published safety record and a return policy.
Step 1: Write down the three jobs you actually want done. Be specific: "fold laundry," "load the dishwasher," "tidy the living room." Then check the honest limits section above. If your jobs are on the cannot-do list, no robot on the market will help you, and waiting is the only rational choice.
Step 2: Set a budget ceiling and compare it to reality. A useful home humanoid will likely launch between $15,000 and $25,000, or around $499 a month on a subscription like 1X's. If that number shocks you, you are not the early-adopter market — and that is fine. Prices fall as manufacturing scales; the second generation is almost always cheaper and better.
Step 3: If you are a developer or researcher, consider Unitree. The G1 from $13,500 is the only humanoid you can actually buy today with an established community, and the R1 Air from $4,900 is the cheapest entry point. Buy it to learn and experiment, never as a household appliance.
Step 4: If you want a home helper specifically, join 1X's waitlist — but pay nothing yet. The NEO pre-order asks for a $200 refundable deposit. A refundable deposit is a reasonable way to signal interest; a non-refundable payment for an unshipped robot is not.
Step 5: Demand the safety and privacy answers before any purchase. Ask for the independent safety certification, the on-device versus cloud processing breakdown, the data retention policy and the kill-switch behavior. If the seller cannot answer in writing, walk away.
Step 6: Revisit the decision in the second half of 2027. By then, Tesla aims to sell Optimus to the public, 1X will have either delivered NEO units or missed its window, and Meta may have shown Max 2 or gone quiet. Real products with real reviews will replace today's speculation — and reviews from actual owners are the only information worth paying for.
Frequently Asked Questions
Is the Meta Max robot real?
According to a November 1, 2026 report by The Information, yes — a prototype exists and was shown to Meta employees in late August 2026. However, Reuters said it could not independently verify the report, no photos have been published, and neither Meta nor Fourier has officially confirmed the hardware, so treat the details as reported rather than proven.
Can I buy Meta's Max robot?
No. Meta told The Information that the Fourier collaboration is "focused on internal research and experimentation" and that it has "no plans to sell these units." There is no price, no order page and no launch date.
How much would a home humanoid robot cost?
Based on today's listed prices: a developer-grade Unitree G1 starts at $13,500, the home-focused 1X NEO is $20,000 or $499 a month, and Tesla's long-term target for Optimus is under $20,000. Expect the first genuinely useful home humanoids to land in the $15,000 to $25,000 range — roughly the price of a car.
When will I be able to buy a home robot that does chores?
No honest date exists. 1X says NEO deliveries start in 2026 in the US, Tesla aims for public Optimus sales in 2027, and Meta has no timeline at all. A realistic expectation: limited early products in 2027-2028, with reliable, fairly-priced home robots more likely in the late 2020s or early 2030s.
What is the NVIDIA Jetson Thor chip inside Max?
It is NVIDIA's flagship edge-AI computing module for robots and "physical AI," built on the Blackwell GPU architecture. The top T5000 module delivers up to 2,070 FP4 teraflops with 128 GB of memory — enough to run serious vision and language models on the robot itself instead of in the cloud. Smaller T3000 and T2000 modules were announced in July 2026 for the first quarter of 2027.
Who is Fourier, Meta's robot partner?
A Shanghai-based robotics company founded in 2015 that builds its own humanoid robots, including the GR-1 and GR-3. Meta reportedly chose Fourier for its ready humanoid hardware and spent about three months co-developing Max, with Meta's robotics product chief Li-Chen Miller spending nearly two months at Fourier's Shanghai factory.
What does "physical embodiment of Muse" mean?
CTO Andrew Bosworth has reportedly described humanoids internally as a physical body for Muse, Meta's consumer AI agent. The idea: Muse currently books travel and fills out forms on your phone; a robot body would let it act in the physical world — picking things up, tidying rooms, handling chores. It is a long-term strategy statement, not something Max can do today.
Is NVIDIA's $3 billion investment in Meta related to the robot?
Not directly, as far as public disclosures show. NVIDIA announced the $3 billion investment in Meta's equity offering on October 26-27, 2026, alongside an expanded partnership on Muse AI models and infrastructure including Vera Rubin systems. It makes NVIDIA both Meta's chip supplier and shareholder, which matters for Max's Jetson Thor brain — but no filing connects the $3 billion to the robot program specifically.
Which humanoid robots can I actually buy today?
Only two makers list prices you can pay online: Unitree in China (G1 from $13,500, R1 from $4,900, H2 at $29,900) and 1X in the US (NEO at $20,000 or $499 a month, pre-order). Figure AI, Tesla, Apptronik, Agility and Boston Dynamics do not sell to the public. And none of the purchasable robots is a proven household helper — they are developer kits, demonstration machines or pre-orders.
Is it safe to have a 48-kilogram robot walking around my home?
That is the industry's hardest unanswered question. Makers use force-limited joints and soft designs, but no independent consumer-safety standard for home humanoids exists yet, and no maker has published third-party safety test results for home use. Until such standards and tests exist, treat every home deployment as a beta test.
Will a home robot from Meta watch and listen to everything?
A home humanoid needs cameras and microphones to function — Max's design targets include voice recognition at five meters and gesture reading at three meters. Before buying any home robot, demand written answers: does perception run on-device or in the cloud, how long is data stored, who can access it, and can recording be fully disabled? No maker has yet published complete, independently audited answers.
Should I put down a deposit on 1X NEO now?
The $200 deposit is refundable, which makes it a low-risk way to hold a place in line if you are a committed early adopter in the US. Do not pay anything non-refundable for any unshipped robot, and do not expect a NEO to reliably run your household on day one — early units of every new robot category have been rough.
Conclusion: A Prototype, Not a Product — But a Signal Worth Watching
Meta's Max is the clearest sign yet that the home-robot race has a new serious entrant. A 1.5-meter, 48-kilogram prototype with five-fingered hands, an expressive display face and a Jetson Thor brain, built in three months with Fourier and already followed by a Max 2 program, shows that Meta's robotics team — formed less than two years ago — can now produce hardware fast. The key takeaway: treat Max as what it is — internal research with no price, no launch date and no sale planned — while treating the trend it represents as very real.
The honest state of the industry in November 2026: Tesla is ramping Optimus production at Fremont but has not sold a single unit to the public. Figure AI has the most deployed Western humanoid and still does not sell to consumers. 1X will take your pre-order for a home robot that has not yet reached a customer's home. Unitree will sell you a real humanoid today, but it is a developer machine, not a butler. And Meta, with Max, has joined the race at the prototype stage with the deepest AI-agent stack of any entrant.
For readers, the practical advice has not changed: do not buy hype, do not pay for unshipped robots, and judge every claim against independent evidence. The home robot is coming — the engineering progress in 2026 alone is remarkable — but the reliable, affordable, safe household helper is still years away. When it arrives, articles like this one will be here to tell you which one is actually worth your money. Until then, watch the 2027 launches, keep your wallet closed, and enjoy the most exciting hardware race in tech.
Sources
Every key fact in this article comes from the reporting linked below. The Information's original report is paywalled; the summaries and related coverage are open.
Comments
Post a Comment