Google Glass was a $1,500 computer worn on the face — a titanium frame carrying a camera, a microphone, a touchpad, and a thumbnail-sized prism that floated a heads-up display above the wearer’s right eye — and after roughly two years as a consumer product it was discontinued, its Explorer Edition pulled from sale on January 19, 2015. It was not killed by a competitor, a price drop nobody could match, or a fatal bug. It was killed by the people standing next to whoever was wearing it.
Glass arrived in 2012 wrapped in the most theatrical launch in Google’s history: skydivers wearing prototypes streamed live video into the Google I/O keynote while co-founder Sergey Brin watched from the stage. In 2013 Google sold the Explorer Edition to a hand-picked cohort of about 8,000 developers and enthusiasts — the “Glass Explorers” — for $1,500 each, framing them as pioneers of a hands-free, always-on future. The hardware mostly worked. The social contract did not. A camera that could record at eye level, without a shutter sound or a reliable indicator that anyone was being filmed, turned every wearer into a suspected surveillance device, and the public responded with a coinage that did more damage than any review: Glasshole.
The backlash was swift and physical. Bars, restaurants, cinemas, and casinos banned the device; a Seattle dive bar’s “no Glass” sign became a news story; the Motion Picture Association of America warned theaters to watch for it; a Las Vegas strip club made patrons check their Glass at the door. The failure was sociological, not technical — a product that asked the world to accept being filmed by strangers, and the world declined. On January 15, 2015, Google announced it would stop producing the consumer prototype and moved the project out of its experimental Google X lab.
Glass did not vanish entirely. Google pivoted it to industry as Glass Enterprise Edition (2017) and Enterprise Edition 2 (2019), where a face-worn screen made sense for warehouse pickers and factory technicians and nobody felt spied on. That tail ran until March 15, 2023, when Google stopped selling it for good. The consumer dream — the device people were supposed to wear to dinner — had been dead for eight years by then. What it left behind was a verb’s worth of caution and a permanent question every wearable since has had to answer: who else is being recorded, and do they get a say?
Juicero was a Wi-Fi-connected juice press that cost $699, then $399, and on September 1, 2017 the company that made it shut down for good, roughly sixteen months after the product went on sale. The machine did one thing: it took a single-serving pack of pre-chopped, pre-pressed produce — sold only by Juicero, on a subscription, scanned by a barcode reader, refused if past its expiry or recalled — and squeezed it into a cup of cold-pressed juice. It was a beautifully engineered appliance built to perform a task that, as the world memorably discovered, two human hands could do about as well and a great deal more cheaply.
The pitch was pure mid-2010s Silicon Valley: the “Keurig for juice,” a hardware-plus-recurring-revenue business wrapped in wellness aspiration and aluminum so over-built that a teardown likened its internal frame to something from an aircraft. Investors believed it. Juicero raised roughly $120 million — by some counts $134 million — from a marquee list including Google Ventures (now GV) and Kleiner Perkins Caufield & Byers, a sum that for a juice startup beggared belief and, in hindsight, narrated the whole tech-bubble mood in one line item. The company launched the press in 2016 at $699, cut it to $399 in early 2017 to chase volume, and was reportedly burning around $4 million a month.
The end arrived not as a market verdict but as a single demonstration. On April 19, 2017, Bloomberg published a report — with video — showing that Juicero’s proprietary produce packs could be wrung out by hand, yielding nearly the same amount of juice in nearly the same time, no $400 machine required. The internet did the rest. The image of a person squeezing a juice bag into a glass while a sleek connected appliance sat idle beside them became an instant parable for everything overfunded and over-engineered in consumer tech. The CEO published a defense; it did not help.
Sales were suspended, refunds offered, and the search for a buyer began. None of it worked. Juicero shut its doors, the press joined the small museum of gadgets remembered chiefly for being absurd, and a $120-million company became permanent shorthand for a question every hardware founder now has to answer out loud: what, exactly, does the device do that the customer could not do without it?
Jibo was an $899 tabletop robot billed as “the world’s first social robot for the home,” and in 2019 its servers were switched off and it went dark — roughly two years after the first units shipped and nearly five years after the crowdfunding campaign that made it a phenomenon. Built around a swiveling head, a round screen, two cameras, a microphone array, and a deliberately expressive personality, Jibo was meant to be a companion: it would recognize faces, turn to whoever was speaking, tell jokes, dance, take photos, and answer questions, less an assistant than a character that lived on your counter. It was designed by MIT roboticist Cynthia Breazeal, a founding figure in the field of social robotics, and it carried genuine pedigree into a market that did not yet exist.
The money followed the promise. After a 2014 Indiegogo campaign that raised around $3.7 million in preorders — at the time one of the platform’s most successful technology projects — Jibo, Inc. went on to raise tens of millions more in venture capital, with reported totals upward of $70 million. Expectations ran high and the timeline ran long: backers who ordered in 2014 waited until late 2017 to receive their robots, by which point the world they were entering had changed underneath them.
That world now had the Amazon Echo and Google Home — cylinders that cost a fraction of Jibo’s price and answered the same questions faster, with a vastly larger ecosystem behind them. Against them, Jibo’s charm could not carry its limits. Reviewers admired the personality and the engineering and found the actual abilities thin: it was, in the unkind shorthand of the moment, a tablet on a swivel that cost $899 and did less than a $50 speaker. The company laid off most of its staff in 2018 and sold its assets, and the servers Jibo depended on were scheduled to go offline.
What made Jibo’s death unusual was the death itself. As the shutdown approached in 2019, the robots delivered an on-device farewell — a short message, and a final dance, telling owners it had enjoyed its time with them and hoping that someday, when robots were more advanced, they might tell theirs that Jibo said hello. The moment was widely covered and genuinely moving to many owners, who had, against their own better judgment, come to feel something for a machine that was about to stop being able to feel anything back.
The Amazon Fire Phone was Amazon’s first and only smartphone, announced in June 2014 and discontinued roughly a year later in 2015 after a commercial failure so complete it produced one of the more memorable numbers in gadget history: a reported write-down of around $170 million on unsold inventory. Launched at $199 on a two-year AT&T contract — iPhone money — it offered two headline tricks. “Dynamic Perspective” used four front-facing cameras to track the user’s head and tilt the interface, producing a 3D-like depth effect. “Firefly” let the phone identify objects, audio, barcodes, and text in the real world — and, not coincidentally, offer to buy them on Amazon. The whole device was, in essence, a beautifully instrumented shopping cart.
Amazon entered the smartphone market years late, against entrenched giants, with a phone priced at the top of the market and an app ecosystem near the bottom. The Fire Phone ran Fire OS, Amazon’s fork of Android, which meant no Google Play, no Google apps, and a stunted app store missing much of what buyers expected a 2014 flagship to do. To pay iPhone prices for a phone that could not run the apps an iPhone ran was a proposition the market declined almost instantly. The gimmicks did not help: Dynamic Perspective was a novelty that reviewers found gave them headaches more reliably than utility, and Firefly was a clever scanner whose primary purpose was funneling purchases to Amazon.
The collapse was swift and visible in the price tags. Within about six weeks of launch, the on-contract price was slashed from $199 to $0.99, and the unlocked price tumbled from the mid-$600s toward $449, then lower still over the following year. In October 2014 Amazon disclosed a charge of roughly $170 million, largely tied to the Fire Phone and related supplier commitments, plus a large pile of surplus inventory it could not sell. By 2015 the phone was discontinued, never to be succeeded.
What the Fire Phone proved was that distribution dominance in one arena does not transfer to another. Amazon could sell almost anything, but it could not will a smartphone ecosystem into existence or persuade buyers to pay flagship prices for a device built mainly to sell them more things. The phone died; the lessons it taught Amazon — about hardware, about pricing, about playing to its actual strengths — fed directly into the cheap, content-first Fire tablets and the Echo line that succeeded enormously by doing the opposite.
The Microsoft Zune was the company’s answer to the iPod, a brown-tinged portable media player launched on November 14, 2006 at $249.95 — and on October 3, 2011 Microsoft confirmed it would build no more Zune hardware, ending a five-year campaign that never once threatened the device it was built to beat. The Zune was not a bad product. The later Zune HD, released in September 2009, was a genuinely handsome OLED touchscreen player, and the supporting software introduced ideas Apple would not match for years. It simply arrived half a decade after the iPod had already won, into a market Apple had spent those years cementing with iTunes, and it never escaped second place — or, more accurately, fourth.
Microsoft built real innovation into the thing. The Zune shipped with built-in Wi-Fi and a wireless sharing feature officially unnamed but universally called “squirt,” which let one Zune beam a full track to another nearby Zune; the recipient could play it three times over three days before it expired. There was Zune Pass, an all-you-can-eat music subscription launched November 2008 that, in a flash of foresight, let subscribers keep ten tracks a month for good — a streaming-plus-ownership hybrid that prefigured the Spotify era by years. The hardware had a tactile squircle control pad and a design language, Metro, that would go on to define Windows Phone and Windows 8.
None of it mattered. By the time the Zune launched, “iPod” was already a generic noun and iTunes was the gravitational center of digital music; Microsoft was asking people to abandon a library and an ecosystem to adopt a player whose chief novelty was beaming songs to the roughly nobody else who owned one. Its U.S. market share never climbed out of the low single digits — around 9 percent of units in its launch week, sliding to roughly 2 percent by 2009 — and it never made the list of the five best-selling players in America. Microsoft sold an estimated two million units total by 2008, against iPods sold by the hundreds of millions.
The hardware was discontinued in October 2011; the brand was folded into Microsoft’s broader media efforts as the Zune Marketplace gave way to Xbox Music in October 2012, and the music service itself was finally retired in November 2015, its users shuffled into Groove Music. The Zune’s legacy is not failure so much as mistiming: a stack of good ideas, shipped years too late, into a war that was already over.
Pebble was the smartwatch that proved the category before the giants arrived to take it — a developer-friendly e-paper watch funded by the most famous crowdfunding campaign of its era — and in December 2016 its founder sold what remained to Fitbit, which kept the talent and the intellectual property and shut the company down. Founded by Eric Migicovsky and routed through Y Combinator, Pebble turned to Kickstarter in 2012 after struggling to raise conventional money, and the result rewrote the platform’s record books: roughly $10.3 million pledged by tens of thousands of backers, the most-funded project in Kickstarter’s history at the time. A 2015 follow-up for the Pebble Time raised about $20.3 million, hitting $1 million in 49 minutes.
The watches were modest by design and beloved for it. A Pebble used a low-power, sunlight-readable e-paper display that ran for days on a charge, paired with iPhone or Android, and threw its doors open to developers — an SDK, an app store, and a hacker-friendly culture that produced thousands of watchfaces and apps. Priced from around $99 to roughly $250, Pebble sold more than two million watches across its models. For a stretch in 2013 and 2014, it was the smartwatch, the proof that wrist computing could work and that an enthusiastic community would build atop it.
Then the platform owners showed up. The Apple Watch arrived in 2015 with Apple’s marketing budget and ecosystem behind it; Fitbit and others pressed in from the fitness side. Pebble, a hardware startup living campaign to campaign, could not match their spend, their distribution, or their balance sheets, and by late 2016 it was running low on cash. On December 6, 2016, Fitbit announced it had acquired Pebble’s software and intellectual property and key personnel — hiring a portion of the staff, reportedly around 40 percent, and laying off the rest. The watches Pebble had not yet shipped were cancelled and refunded; the ones already on wrists would keep working “for now,” with support withdrawn and future functionality, the founder warned, likely to shrink.
The company was gone, but the watches refused to die. A volunteer community called Rebble stood up replacement servers when Pebble’s own cloud services went dark in June 2018, keeping the app store, voice features, and watches alive long past their maker’s death. And in a genuinely rare turn for this catalog, the story bent back toward life: on January 27, 2025, Google open-sourced PebbleOS, releasing the code that had been locked inside a defunct acquisition, and Migicovsky used it to relaunch Pebble hardware. The smartwatch Fitbit switched off became one of the few in this archive to get a second act.
The Microsoft Kinect was a $150 motion-sensing camera that let players control an Xbox 360 with their bodies instead of a controller, and on October 25, 2017 Microsoft confirmed it had stopped manufacturing it for good. Launched on November 4, 2010 as an add-on for the aging Xbox 360, Kinect did something no mainstream game device had done before: it watched the room, tracked a player’s skeleton, heard voice commands, and turned “you are the controller” from a slogan into a product. People bought it in numbers that embarrassed every prior peripheral. It was not killed because it failed. It was killed, in large part, because Microsoft tried to make everyone buy it whether they wanted it or not.
The early success was genuine and historic. Kinect sold roughly eight million units in its first 60 days, earning a Guinness World Record as the fastest-selling consumer electronics device, and went on to move around 35 million units over its lifetime — a figure Microsoft cited when it announced the end. For a brief window it looked like the future of how people would interact with computers: waving, leaning, speaking, no controller required.
Then Microsoft made the decision that defined Kinect’s death. When the Xbox One launched in November 2013, every console came bundled with a new, more powerful Kinect, which helped push the price to $499 — a full $100 above Sony’s PlayStation 4 at $399. Gamers, who mostly wanted a games box rather than a living-room camera that had to be plugged in to work, balked. The PS4 outsold the Xbox One decisively out of the gate, and in 2014 Microsoft reversed course, releasing a cheaper Kinect-free Xbox One at $399 and quietly cutting the sensor loose.
Unbundled, Kinect lost its reason to exist: developers stopped building for hardware that was no longer in every box, sales dried up, and in October 2017 production ended. The technology did not die, though. The depth-sensing and skeletal-tracking work Kinect pioneered fed forward into Microsoft’s HoloLens headset and, later, the Azure Kinect developer camera. The gadget that taught a generation to flail at their televisions was gone; the engineering that made it possible quietly went to work elsewhere.
Ouya was a $99 Android-powered game console that promised to break open the closed, expensive world of console gaming, and on June 25, 2019 its servers went dark and bricked much of what the device could still do. It began as one of the great crowdfunding stories: in the summer of 2012 the Ouya Kickstarter raised about $8.6 million from roughly 63,000 backers, hit its goal in a matter of hours, and arrived freighted with the hopes of everyone who wanted gaming to be cheap, open, and free of the gatekeeping of Sony, Microsoft, and Nintendo. It launched in June 2013 as a small black box you plugged into your television, ran Android games, and let anyone develop for. The dream did not survive contact with the product.
What backers and buyers received was underwhelming on nearly every axis. The hardware felt cheap, the controller was widely criticized as clunky and laggy, and the storefront was thin — a small catalog of mostly minor games, many of them free-to-try ports, with few reasons to keep the console connected to the television. The “open console” pitch was real, but openness without a compelling library is just an empty shelf, and the novelty that drove the Kickstarter curdled quickly into buyer’s remorse. Sales after the initial crowdfunded wave were poor.
Ouya the company could not make the economics work, and in 2015 the gaming-hardware maker Razer acquired its software assets, folding the storefront and content into its own Forge TV effort and ending the Ouya hardware line. For a few years the servers limped on, letting existing owners still reach the store and re-download what they had bought. Then Razer set a hard date: on June 25, 2019, it would deactivate Ouya accounts and shut down all online elements of the service.
When that date came, much of the Ouya experience simply stopped working. The store closed, purchases could no longer be downloaded, and games that phoned home for licensing or content broke, leaving owners able to play only what was already installed and self-contained on the box. A console that had been sold on the promise of openness ended as a sealed brick — a small black reminder that a record-breaking Kickstarter buys a launch, not a future.
Anki was the consumer-robotics startup that made the genre’s most charming products and then collapsed almost overnight when a single financing deal fell through. Founded in 2010 by three Carnegie Mellon robotics PhDs — Boris Sofman, Mark Palatucci, and Hanns Tappeiner — the company launched its first product, the AI-driven racing game Anki Drive, in 2013, and went on to build Cozmo (2016) and Vector (2018), two small desktop robots widely praised for genuine personality. On April 29, 2019, Anki told its roughly 200 employees that the company was shutting down within days. It was bankrupt.
The waste was the painful part. Anki was not a vaporware outfit or a crowdfunding ghost. It had raised in the region of $185 million from serious investors including Andreessen Horowitz, Index Ventures, and JP Morgan; it had reportedly sold around 1.5 million robots, including hundreds of thousands of Cozmo units; and its products were good — Cozmo and Vector were among the few home robots that critics and owners actually loved rather than tolerated. The robots used expressive movement, a tiny animated face, and real computer-vision and on-board AI to do something most “social robots” only claimed to do: feel alive on a desk.
What killed Anki was the brutal economics underneath the charm. Building a hardware-and-software company that designs custom robots, runs the cloud services behind them, and funds the next product is enormously capital-intensive, and Anki was burning through its money faster than the robots could replenish it. The company was negotiating a major new round when, in the CEO’s words, a significant deal at a late stage fell through with a strategic investor. Without that bridge, a business that lived on outside funding had no runway left.
The end was abrupt and the human cost was real: about 200 staff were laid off with minimal notice and little severance, a sober reminder that behind every “startup shuts down” headline are families that lost a paycheck without warning. The robots themselves were not entirely abandoned. In December 2019, the edtech firm Digital Dream Labs acquired Anki’s assets and later revived Vector, keeping its cloud services running and giving the little robot — and the people who had bonded with it — an unexpected second life. Anki’s products earned their affection; the company simply ran out of the cash that affection alone could never supply.
The Microsoft Band was Microsoft’s attempt to wear technology on the wrist, and after two generations and two years it was discontinued, quietly pulled from the company’s online store on October 3, 2016 with no third model to follow. It launched on October 30, 2014 at $199 — a slim, screen-bearing fitness tracker that crammed ten sensors into a stiff plastic-and-rubber strap and promised to read heart rate, skin temperature, sun exposure, and sleep while doubling as a notifications display. A second version, the Band 2, arrived a year later at $249 with a curved screen, a barometer, and eleven sensors. Then the team was disbanded, the device was delisted, and Microsoft walked away from the wrist.
It was not killed by a single rival or a fatal flaw so much as by the accumulated weight of its own compromises. The Band did a remarkable number of things, and it did most of them adequately and none of them with grace. The strap was rigid and uncomfortable for all-day wear; the battery lasted roughly two days, less with GPS; and — most damningly — the rubber had a habit of cracking and splitting along the seams, a defect common enough that Microsoft revised the materials for the Band 2 without ever quite admitting why. A wearable’s first job is to be worn, and the Band made that a chore.
Microsoft’s exit was characteristically tidy and characteristically final. There would be no Band 3; the software development kit was pulled; and the company folded its Health-branded apps into a backend it hoped third-party wearables would use instead of building its own. That hope, too, expired. On May 31, 2019, Microsoft shut the Microsoft Health Dashboard and removed the Band apps from every app store, deleting the cloud data that made the device smart and leaving the surviving Bands as offline pedometers.
What the Band left behind was not a community in mourning but a cautionary spec sheet: a device that out-sensored every competitor and lost anyway, because it never solved the unglamorous problem of being comfortable on a human arm.