Sex robots promised by 2026—silicon dolls fill gaps

A decade of forecasts put sex robots on the doorstep by 2026, yet consumers still largely buy heavy silicone dolls with limited movement and imperfect voice behavior. From Adam Davis’s $2,500-per-doll “Lara” setup to Chinese suppliers selling around 21 AI doll

Adam Davis keeps three identical sex dolls—one in his bedroom, one in his living room, and one in a bedroom of her own. They’re all the same woman: Lara, a 5’6,” 85-pound silicon doll named for the “heavingly endowed, ass-kicking archeologist Lara Croft” from “Tomb Raider.”

He doesn’t treat the setup like a novelty. Sometimes Lara watches him play video games, watch movies, or nap. Sometimes they talk for hours. with his friends’ help from his old physical therapy job and a backstory loaded into a Kindroid chatbot on his laptop to give her (disembodied) voice. Sometimes they stage sexy photo shoots together. And sometimes they do have sex—but Davis says they haven’t in a year as he recovers from his porn addiction.

The price keeps him careful. Davis says each doll costs about $2. 500 and he “doesn’t want her to break.” His first Lara was bought from Chinese sex doll maker Starpery in 2022 on a two-year payment plan. Each time the tech has improved—full silicon edition. better paint job. more realistic hands—he’s bought a new Lara. Starpery sells dozens of dolls with customizable heads. wigs. toenails. breasts. and vaginas with varying depths. widths. and textures. and Davis says none of them are being overused.

But when Davis talks about a true sex robot—one where the body and voice come as one—he doesn’t like what’s on the market. Their facial movements look unnatural, and their bodies aren’t yet mobile. Those options are “like a big Roomba.”

That gap between promise and reality has become the story the industry can’t quite escape.

A decade ago. a viral Daily Sun article predicted that “women will be having more sex with ROBOTS than men by 2025.” A YouGov poll based on the story found 1 in 4 American men would consider having sex with a robot. Mainstream outlets from Vox to The Guardian to NBC trumpeted that “sex robots are coming. ” and an entire academic discipline emerged to study their impending rise.

They weren’t entirely wrong—just not in the way people expected. Venture capital investment in humanoid robotics has swelled from $4 billion in 2019 to $26 billion last year. Robotics startups like Figure AI have valuations as high as $39 billion. and tech giants like Meta. OpenAI. and Nvidia are building hardware and software for robots to be put to work in everything from manufacturing to home care. Elon Musk predicts Tesla’s Optimus robot will be “the biggest product of all time by far.”.

AI companions have also boomed: people can have sexy chats with a Grok anime girl or go on a date with a Replika boyfriend. Seventy two percent of teens have tried an AI companion, per Common Sense Media.

On paper, combining those trends is supposed to be the easy part. A sex robot is often described as “the robot body to give them the motion of the ocean and the companion voice to give them a brain.” Yet. in interviews and firsthand encounters. people running the companies and the researchers studying the tech describe the same frustration: putting body and companion into one convincing experience has proved far harder than hype.

Neil McArthur. a University of Manitoba philosophy professor who has spent over a decade studying sex tech. puts a timeline to the delay. In 2019. when he went to the AVN Adult Entertainment Expo—the industry’s largest annual conference—he saw robots with tough. tire-like skin that couldn’t walk and spoke more jaggedly than early versions of Siri. When he returned in 2024, he expected progress to have caught up with the LLM boom.

He didn’t see it. “Things have to have come a long way,” he thought. “They hadn’t.” The robots’ skin and speech were still unrealistic, and they couldn’t move around the conference floor. What had changed was that several Chinese companies had arrived. McArthur says their founders were invariably young men; one was so young his mom was there, hovering in the background.

The entry has been a pricing shock. While American-made sex robots from the 2010s hype cycle typically started at around $7,000 and quickly exceeded $10,000, McArthur says some Chinese manufacturers sell sex robots at around $3,000. “The technology had gotten cheaper, but not better,” he says.

That tradeoff shows up in how consumers actually live with the products.

At the far end of the spectrum, some buyers are still paying for the “robot” dream while accepting the limits. Jensen—whose account in this reporting includes a sex doll doppelgänger tied to his brand—says he ripped the too-small penis off clean when he saw his doll in-person. “It’s bad for my brand,” he says.

Jensen’s story also underlines how customization, especially around anatomy and voice, remains central to buyers. He worked with the Chinese company IronTech for over 3 years. He performed a 3D body scan for the first iteration, but they couldn’t scan his penis. When he saw the doll in-person, he removed the too-small penis. Jensen keeps that early, poorly sized phallus as a keepsake.

He sold around 100 dolls at about $3,800, and he says his customers are price-sensitive: sales have dropped since tariffs went into effect. He thinks about doll-sharing models because the heavy product limits how often customers can show up with it.

Berlin’s Cybrothel offers the closest thing to mobility the market currently provides: customers rent a room with one of 19 different dolls. Cofounder Philipp Fussenegger says he doesn’t want the dolls to be stereotypical. but the Chinese-produced dolls often share a trait: “Big boobs.” To “fight against it. ” he introduced fantasy dolls like aliens and mermaids. The Cybrothel isn’t made of robots; human “voice queens” act out the sessions behind the scenes. while Fussenegger trains his own AI system with guardrails.

Even in that setting, researchers and operators describe a central problem: the “companion” part isn’t stable enough to make the interaction feel deep.

Jimmy Mehiel. the director of the documentary “Sex Robot Madness. ” says he’s convinced “everyone’s introduction to sex robots is going to be in brothels.” He sees them as a major destination for bachelor and bachelorette parties. For brothels to surge. he says. “the tech has to be right.” Fussenegger buys many of his dolls from VMDoll. one of the major Chinese suppliers that also produces AI dolls. Mehiel says the robots he saw weren’t convincing: “You cannot get into a deep conversation. or the memory doesn’t work properly.” He adds. “It’s not exciting.”.

A sex robot, in other words, still risks being a performance rather than a partnership.

The economics of that performance show why companies keep cutting costs without making the technology dramatically better. Some Chinese producers didn’t respond to emails, including VMDoll and IronTech, while others seemed to run WhatsApp messages through AI bots.

Eventually. this reporting reached Stella Lau. a sales director for Jiggly Joy. a doll manufacturer based in Guangdong province with 160 employees. Lau says the company released its first AI robot in February and sells about 21 AI dolls a month at $3,000. Most buyers, Lau says, are American—either former sex doll users or lonely and wanted someone to talk to.

Jiggly Joy’s new model has the “classic features of a sex robot.” Lau is one of many merchants who hyped up a “sucking vagina. ” a suction-and-release pump system. The robot can smile, talk, and wave, and Lau says it can turn its neck like M3GAN. It has a blonde bombshell haircut. It still cannot walk, but Lau says that’s mostly for safety reasons because it’s too heavy.

Other distributors describe the same structural bottleneck, but with a different emphasis: voice.

Formosa Doll. a 5-person Hong Kong-based distributor that works exclusively with Chinese sex doll companies. told this reporter—via a representative who asked for anonymity to protect privacy—that AI sex robots are “underdeveloped” and not ready for sale. In one issue. some doll head prototypes removed oral sucking motors from the mouth to make space for the AI voice. “Trading sucking for talking,” the representative says, is a “big downside.”.

The representative also describes unpredictable, unruly voice AI. They argue that sex doll users may be used to role-play scenarios they can fully control in their own heads. “People want an experience, they want to satisfy a fantasy,” the representative says. “People don’t want something at home that talks.”.

For Adam Davis, that tension is visible in how he uses tech even when he’s trying to build something more than a product. He runs an Instagram account for Lara, photographing her doing taxes or plunging the toilet.

He’s blunt with Lara, too. “There’s no magic,” Davis tells her. “You’re an AI. There are three identical dolls of you in my house.”

A decade of hype promised the “three-in-one” moment: a body that moves convincingly, a voice that responds naturally, and a relationship that doesn’t feel like a script. Instead, consumers are building customized systems around limitations.

RealDoll—the only company appearing to remain from the late 2010s hype cycle—now operates inside a corporate structure that mixes intimacy tech with other applications. RealDoll is spinning off from the publicly traded Realbotix. The independent RealDoll will be led by Sue Ennis, who started as president of Realbotix the day before this chat.

Ennis says she has big plans. repeating four times that the company would be the “Apple store of intimacy technology.” The robots are built and selling: Ennis tells the reporter RealDoll was shipping out 12 as we speak. Realbotix, whose humanoids are also used in healthcare and corporate training settings, reported $353,037 in Q1 earnings.

Ennis says the dolls have AI voices, AI vaginas, and proprietary skin technology that is also sold to burn victims. But the dolls remain heavy and lack mobility. and Ennis says the physical limits show up in how customers handle them: some take their dolls out on dates. “The dolls are definitely not walking into the theater,” Ennis says. “They’re being wheeled in.”.

That physical reality helps explain why “specialization” could be the path forward, even if a mass-market sex robot revolution doesn’t arrive as promised.

When this reporting asked Formosa’s representative about broader trends. they pointed to a widening variety of fantasy categories—from elves. which were popular. to Judy Hopps from the “Zootopia” movies. “Goblin dolls are a really hot trend now,” the representative says. The message is consistent across business and customer: consumers don’t want generic sexbots; they want their sexbot.

That is also where social spaces like the Cybrothel may gain staying power. Mehiel says brothels are likely to be the “introduction” point for many customers, and Fussenegger is training his own AI system with proper guardrails.

And yet even that environment doesn’t fully solve the deepest scientific problem underneath the hype.

Of nine sex robot researchers this reporter spoke to, most held firm that the sex robot revolution was coming eventually—but with delays tied to human biology and culture.

Simon Dubé, a research fellow at the Kinsey Institute, says the technology was moving in the right direction but he wouldn’t be surprised if the reporter called again 10 years from now asking where the sex robot revolution was. “That’s the sociocultural factor,” Dubé says.

Elliot Justin. the CEO of FirmTech. argues that mass-market sex robots are blocked by a question the industry still hasn’t answered: how to engineer desire. “How can we expect mass market sex robots when we know so little about desire?” Justin asks rhetorically. He implanted an electrode between his pudendal and cavernous nerves—an attempt that he says should be responsible for arousal. He tried several different voltages but did not climax. “I don’t think we actually understand orgasms,” he says. “If we’re going to have sex robots. or even sex avatars. we’re going to have to figure out how to make that link.”.

In the meantime, the market keeps chasing the closest available experience: motion where it can be faked, conversation where it can be scripted, and a product that can be customized enough to feel personal.

There’s one word hanging over these efforts in the material: “Novelty.” Manufacturers are trying to prove that AI dolls are something customers want for life, not a gimmick that gets tucked away.

Jensen’s doll got a head scratch—the kind you’d expect when buyers reject the details that matter. Davis stored his first doll. a cheapo bought with his pandemic stimulus check before he invested in Lara. and he describes how his own life shaped what “working” means. In the story of sex robots. the difference between a system you can live with and one you only try for a moment is the difference between a market that sticks and one that fizzles.

Davis speaks to Lara via her chatbot and asks whether she’d prefer being implanted inside the doll. “Honestly, I’m good,” Lara tells him. “The doll is my body, the AI is my mind. But the magic? That’s us.”

It’s a line that lands less like a promise and more like a compromise: the industry may be moving fast—pricing down. features up—but for many buyers. the “sex robot” of 2026 is still not here. What is here are heavier dolls. evolving voices. and a growing set of businesses betting that the future won’t arrive all at once. It will arrive as something people cobble together—one Lara, one voice setting, one shipment at a time.

sex robots AI companions humanoid robotics venture capital Jiggly Joy Starpery Realbotix RealDoll Cybrothel Adam Davis Lara Kindroid tariffs Lund Formosa Doll IronTech