AI Hype Meets Hard Reality: The Struggle of Launching AI Hardware Startups in a Tech Giant-Dominated World

To look at this article again, go to My Profile and then click on View saved stories.

Lauren Goode

The Challenge of AI Hardware Remains Despite Advancements in Generative AI

AI hardware startups are facing a tough time.

Following an extensive development period, the startup company Humane introduced a $700 wearable device in early April, which makes significant use of artificial intelligence. The unique selling point of the Ai Pin was its promise to eliminate the need for managing multiple apps by having an operating system capable of "identifying and utilizing the appropriate AI for any given situation," enabling it to stream music, translate various languages, and even calculate the protein content in a handful of almonds. With its lack of a conventional screen, the Ai Pin aimed to serve as a remedy to the overuse of screens, suggesting a future where smartphones become obsolete.

The Ai Pin has received widespread criticism. Julian Chokkattu from WIRED gave it a low score of 4 out of 10. Meanwhile, renowned YouTuber Marques Brownlee didn't hold back in his critique, even going as far as to label it the "Worst Product I’ve Ever Reviewed … For Now," despite acknowledging its impressive hardware design. In response to the backlash, the creators have slightly altered their initial claim that the pin could serve as a replacement for traditional phones. Humane's co-founder and CEO, Bethany Bongiorno, has taken to Twitter to engage directly with unhappy customers and enthusiasts alike. She's been offering apologies, promising upcoming enhancements, and sharing video demonstrations of the device's user interface, which interestingly projects information directly onto the user's palm with lasers.

Humane seems to have missed the mark with its recent product release, joining the ranks of others in similar situations. The more affordable Rabbit R1, initially marketed for $200 as a generative AI "portable buddy" and initially met with enthusiasm, has since been criticized as "lacking," "incomplete," "not fully developed," and "not dependable." Chokkattu from WIRED rated it a disappointing 3 out of 10. Additionally, there are concerns regarding the device's management of sign-ins for external applications like Uber.

The occurrence of early-stage hardware failures is not something new. Many emerging businesses have been guilty of setting high expectations through their marketing efforts, only to deliver subpar products. The challenge of competing in the hardware sector is significantly heightened in today's era dominated by major tech conglomerates and their expansive ecosystems. Developer Ben Sandofsky has speculated that the commitment of the Humane cofounders to the "Apple Way," or working in isolation, could be a contributing factor. According to a blog post he authored, they dedicated years to perfecting their product as if they were a large tech entity, utilizing $230 million in venture capital, rather than having access to a vast reserve of billions.

However, it seems that both Humane and Rabbit may have miscalculated once again. They were counting on the enthusiasm surrounding AI, especially with the rise of ChatGPT, to attract initial users and avoid becoming obsolete. Yet, they ended up heading full speed into a dead end, driven by the allure of AI. The reality is that the challenges of hardware development aren't lessened by the presence of generative AI.

Costly Failures

"For an AI gadget to truly excel, it's essential to master both the hardware and software aspects. The concern with certain startups is to what extent their software component is merely superficial," states M. G. Siegler, a partner at GV, the venture capital arm of Alphabet.

Siegler points out that established technology companies currently hold a greater edge, as they possess the capability to develop new offerings on their own platforms and can withstand financial losses during the refinement phases of product development. In contrast, startups are trying to introduce their innovative AI solutions from scratch. Meanwhile, giants like Meta, Google, Microsoft, and Apple have the advantage of leveraging their existing workforce and resources to integrate AI assistants into highly advanced wearable devices, produce smartphones featuring native generative AI capabilities, equip their laptops with special AI-dedicated buttons, and enhance their tablets with exceptionally potent AI processors.

"Large technology firms can afford to launch a hardware product up to five times, in contrast to startups, which typically have just one chance," states Jacob Andreou, a Greylock investor with a background in product development at Snap. "The likelihood of a smaller entity securing additional funding after a costly failure is slim."

Sheon Han

Benjamin Oliver

Gabrielle Caplan

Jaina Grey

Bongiorno, the leader and co-founder of Humane, stated that launching a first-generation product is a significant hurdle but mentioned that the company has rolled out "stability updates" to enhance their gadget. He expressed that the introduction of Ai Pin and its operating system, Cosmos, marks the inception of the journey into ambient computing, acknowledging that there's a substantial path ahead in this field. Separately, Rabbit's founder and CEO, Jesse Lyu, provided a statement indicating the company's commitment to rapidly and continuously refining their product based on the active feedback from their community. Lyu highlighted that since the initial batch of R1 devices reached their consumers, Rabbit has issued three software updates.

Humane and Rabbit are not the only new ventures integrating advanced AI technologies into innovative devices. Limitless AI, which was previously known as Rewind, has recently introduced a wearable device that functions as a memory aid by transcribing spoken words. Meanwhile, a startup named Iyo, which originated from Alphabet’s experimental division X, has developed smart earpieces. These devices aim to serve as a combination of therapist, coach, and tutor, and are expected to be voice-operated when they hit the market later this year. Additionally, an AI-powered navigation tool named Terra, which leverages technology from Google and ChatGPT, offers guidance for outdoor activities like walking and hiking. Terra's design will be shared openly to inspire individuals to create their own iterations of the gadget.

Many of these emerging companies are utilizing artificial intelligence to provide ways for users to interact without needing a screen, and to share information without the necessity of navigating through countless mobile applications. Additionally, some are optimistic that freely available, open-source AI frameworks will evolve to be more potent, simpler to tailor, and capable of operating directly on devices. They also anticipate that cloud-based AI services will become more efficient and less expensive to utilize as the technology progresses.

Even for an AI hardware startup that successfully develops a solution-oriented and functional product, the challenge remains to stand out in a market dominated by major corporations that significantly influence how consumers interact with technology. These startups must also persuade customers to adopt novel ways of engaging with their devices. In a recent discussion, Jason Rugolo, the innovator behind the latest AI device called Iyo One, pointed out the considerable influence that giants like Google and Apple have through their mobile ecosystems. "Their platforms are designed around their specific graphical user interfaces, which naturally become their focus," Rugolo shared with WIRED. "Our approach introduces a new application framework where dialogue is the main form of interaction, requiring us to innovate differently."

Rugolo points out that the terminology used in the promotion of these modern products is rapidly evolving, and the focus on "gen AI" might just be a temporary trend. "Technology plays a significant role in our operations. Fundamental to conversing with machines naturally are extensive language models, which in my view, fall under 'gen AI'," he mentions. "For instance, an app designed to improve hearing is probably going to utilize 'machine learning'. On the other hand, a contemporary translation application is more inclined to employ a large language model."

Cyclical Trend

According to Christina Warren, who is currently a senior developer advocate at Microsoft's GitHub and has a background as a tech journalist, the current rush towards AI devices harks back to the early 2010s trend of consumer wearable technology and gadgets financed through Kickstarter. This earlier wave was similarly driven by the advent of new technologies that simplified the manufacturing process.

"Warren recalls the array of tech products that captured attention: Google Glass, the Pebble Watch, Oculus Rift, and the Ouya gaming console, not to mention an Instagram photo frame he purchased. According to him, Kickstarter devices eventually turned into a bit of a joke, with the expectation that the majority would not succeed."

Sheon Han

Oliver Ben

Gabrielle Caplan

Jaina Grey

During that period, numerous devices were developed using modified versions of the Android Open Source Project as a foundation, with manufacturers customizing the system with their unique launchers or user interfaces. This approach is reminiscent of how current tech startups leverage APIs from services such as ChatGPT to develop bespoke software solutions on top of them. "Android significantly propelled the technological advancement of those gadgets, while Kickstarter emerged as the financial catalyst for their initial funding," Warren notes.

Numerous innovative devices that emerged in the 2010s have been either developed directly by or integrated into the operations of major tech firms. According to Warren, Pebble effectively realized its vision for an open-source smartwatch, yet its technology was eventually acquired by Google as part of Fitbit's purchase, resulting in a complex layering of wearable technology. Oculus, a company specializing in VR headsets, was an early acquisition by Facebook to bolster its VR ambitions. Amazon is credited with pioneering the smart speaker market through Alexa, while Apple's sophisticated version of a smartwatch emerged victorious. Throughout this process, each of these leading companies achieved trillion-dollar valuations and delved further into hardware design, even venturing into the manufacture of their own computer processors. For emerging companies like Humane and Rabbit, merely branding a product as "gen AI" is not sufficient to overcome these challenges.

Siegler emphasizes that for any AI hardware startup aiming for success at the moment, it's crucial to focus on building a strong brand identity and, above all, to maintain utmost simplicity.

"Declaring your intent to enhance the world can seem overly ambitious," he mentions. "Besides, smartphones already possess a lot of these capabilities. Therefore, it's essential to begin with something straightforward, such as a wearable device that utilizes a single AI model and carries a sense of intention."

Andreou suggests that the process of creating AI wearables might become less challenging, as AI hardware startups might collaborate with original device manufacturers—lesser-known firms that produce items for well-known consumer brands. This strategy would involve customizing or improving products with their software, thereby reducing the manufacturing expenses associated with the product.

"He emphasizes the importance of having a small team, ideally one or two individuals, within the company to oversee hardware operations while outsourcing the majority of tasks to minimize expenses," he suggests. He also foresees a trend where hardware startups will adopt a subscription-based approach to enhance their income. He notes that some companies have begun experimenting with this model—for instance, Humane offers a subscription at $24 a month. However, he cautions that for this strategy to be successful, the product must be effective and reliable.

Warren believes that emerging businesses with limited resources could benefit from integrating AI into their products by utilizing smaller, open-source AI models. These models can operate on devices independently and demand less processing power. However, she raises a critical point about the specific type of hardware necessary for this integration. According to her, even the manufacturers of the hardware appear uncertain about what should be developed.

Recommended for You…

Delivered directly to your email: Will Knight delves into the latest progress in artificial intelligence with Fast Forward.

He transferred all the digital currency from an online trading platform onto a USB stick—before van

Live deepfake romance scams are now a reality

Enthusiasm for Boomergasms is

Heading outside? Check out the top sleeping bags for any journey

Reece Rogers

Reece Rogers

Name: Julian Chokkattu

Elena Small

Benj Edwards, a writer for Ars Technica

Jessica Thompson

Steven Levy

Rogers Reece

Additional Coverage from WIRED

Evaluations and Instructions

© 2024 Condé Nast. All rights reserved. Purchases made through our site could result in WIRED receiving a share of the sale as part of our Affiliate Partnerships with retail outlets. Reproduction, distribution, transmission, caching, or any other form of utilization of the content on this site is strictly prohibited without prior written consent from Condé Nast. Advertisement Options

Choose a global website