 |
|
|
|||||
Shares in GSI Technology, Inc. (Nasdaq: GSIT) are soaring for the second day in a row. The company, which specializes in semiconductor memory solutions, saw its stock price skyrocket 155% yesterday. Today, GSIT shares are up another 39% in premarket trading as of the time of this writing. But why? Heres what you need to know. What is GSI? GSI Technology is a provider of semiconductor memory solutions. That means it specializes in developing memory chips and products that help process data. Though the company isnt as well-known as the bigger semiconductor memory solutions giants like Micron or Intel, it has been a staple of the semiconductor industry for three decades now. GSI was founded in 1995 and is headquartered in Sunnyvale, California. According to the company’s website, it has 158 employees and more than 120 granted patents. The companys most recent products of note are the Gemini-I and Gemini-II associative processing units (APUs). These APUs use Compute-In-Memory (CIM) architectures, which can process tasks inside the memory itself without the need to transfer the data to an external processor. Why are GSI shares surging? Shares in GSI Technology surged 155% yesterday after the company announced the publication of a paper by researchers at Cornell University. This paper, titled Characterizing and Optimizing Realistic Workloads on a Commercial Compute-in-SRAM Device, found that the companys associative processing units CIM architecture can match GPU-level performance of certain AI workloads but at a fraction of the energy consumption. And that energy consumption reduction is significantup to 98% less energy is required compared to GPUs. Additionally, the companys APU allow for the processing of retrieval tasks faster than traditional CPUs, which can reduce processing time by up to 80%. Whats so important about the Cornell paper? Like all companies, GSI Technologies makes claims about their products, which boast of their superior performance. But a company’s claims need not be backed up by independent research or scientific rigor. Cornells research paper effectively verifies large parts of GSIs claims about its Gemini-I and Gemini-II associative processing units, which should give potential customers a lot more faith in the productnot to mention the companys investors, too. The research was presented at the IEEE/ACM International Symposium on Microarchitecture, aka “Micro 2025.” What is the potential impact of GSIs breakthrough? AI computing is a very intensive task, requiring powerful GPUs and/or CPUs. Because these tasks are so compute-intensive, they require a lot of energy to carry out. More energy means higher costs. But if GSIs associative processing units can deliver the same AI performance at a fraction of the energy expenditure required over traditional means, one of the largest costs of computing AI tasks drops tremendously. This is good not only for a customers bottom line but for the environment as well. GSIT shares have had a stellar 2025 so far Upon GSIs announcement yesterday of the publication of the Cornell paper, the companys stock skyrocketed. GSIT shares closed up a staggering 155.32% yesterday to $12.97 per share. As of the time of this writing, in premarket trading this morning, GSIT shares are up another another 39% to $18.15 per share. However, this stellar performance of GSIT shares is nothing new for the companys stock price in 2025. As of yesterdays close, GSIT shares have surged 328% since the year began. Over the past 12 months, the companys stock price has grown 239%.
Category:
E-Commerce
Brian Chesky, the CEO of Airbnb, still admires Facebook. Not the Facebook of today, but the Facebook circa 2005. When it pretty much just told you someones birthday and let you poke ’em. It would still be a great product! exclaims Chesky. We’re not going to be that company [making it], but there’s still a need for it. But while Chesky doesnt want to build Facebook 2.0, he is laying the groundwork for Airbnb to become something much closer to a social network. Airbnbs fall updates launching today are but the first steps in a significant reframe of the experience of using Airbnbone that is moving it closer to social networking, and another that embeds it far more intimately with AI. The social aspects arrive through the introduction of Connections. Basically, when you book an experience with Airbnb, you will be able to opt in to share yourself. When other people book the experience, they will see your face. Then after the experience, they can see your greater profileand slide into your DMs to keep the discussion going. [Image: Airbnb] Beyond that, Airbnb is also doubling down on using AI agents for support processesin a move that Chesky claims speeds up the service, from roughly three-hours in phone calls to solve problems, to a few seconds. While introduced earlier this year for generic answers, now Airbnb support agents will be able to know deeper details about your tripsso they can respond with personalized answers: like if you are allowed to bring a dog to that apartment in Mexico City. These updates might seem minor or iterative, but to Chesky, a designer-entrepreneur with a penchant for extracting deep narratives from his own product, they are the first stages into redefining what Airbnb issomething of an AI-driven life experience platform that pushes you to meet others IRL. This conversation has been condensed and edited. Id like to kick things off talking about the social networking youre building into Airbnb. I think I see some original intent: I wouldnt want to sign up for an Experience if I don’t know who else is going. But how much of this update came from that problem, and then how much of it was born from other ambitions? So there’s two reasons we’re doing the social features on Experiences. The first is, we just noticed that a lot of guests use it to meet people. So yes, they want to get a vibe for who’s coming. We just show their profile avatars, but it gives a little bit of a sense of aliveness, a sense of, “Oh yeah, they’re kind of people my age, or kind of cool-looking people.” We also notice a lot of people create WhatsApp groups after the experience is over, because they may want to hang out after. It’s kind of difficult [to set that up]. So we wanted to make that easier. Part of it was just responding to how people were using it. That was the tactical reason to do it. [Image: Airbnb] There’s a much more strategic reason, which is, I don’t want Airbnb just to be a marketplace. I want it to be a community. A marketplace just sells you stuff. A community connects you to other people, other places, other cultures. And I think this is the beginning of us shifting from a marketplace to a community. Now, admittedly, these are modest features. I’m not saying this is, overnight, a social network in the real world. But we’re going to be announcing a lot more social features next year, and even more the year after. I think pretty soon it’s going to be a very social platform. Not like Facebook, social. Social in the real world. Its interesting you put it that way. I feel cautious about Airbnb social features, because the social aspects of Airbnb can already be so hit or miss. And so I find wondering if Ill want to engage that much with a more social Airbnb. I think everything will be very personalized in the future. And part of our AI strategy is that no one sees the same homepage, no one sees the same app. So the first thing we need to do is learn more about you. Like, do you want to be social? Not everyone does. If you don’t want to be social, you don’t have to opt into the social features, and we can show you things that are just less social. Not everyone wants the same level of social features. I’ll give you an example. We’re going to offer hotels in Airbnb. Some people will never stay in a hotel on Airbnb. We’re never going to show them hotels . . . we need to get more personalized. I guess the second thing, though, is that we do think human connection can be great, but you’re right. It’s hit and miss. So I think a bunch of what we want to do is do more verification and better matching of the kind of people that you might be inclined to want to meet. So it’s kind of a two-fold approach to personalization: Do you even want to meet people? And if you do, do we match you to the right people? [Image: Airbnb] With your initial launch, really, all I’m sharing is my avatar when I register for an Experience. It’s not until after the Experience that people can be like, “Oh, there’s Marks profile. I can scope him out, learn who he is.“ Yep. And you can opt out of the avatar and opt out of people messaging you. Longer term, Ive heard you frame Airbnb as a unifier during a time when the world is really divided. Is that too much weight to put on the platform? I feel like that’s a lot of weight to put on the platform! I think it’s a good mission. I don’t think it’s too much weight to put on the platform. I think what would be too much weight is us trying to solve a global problem on our own. That would be unreasonable. But if you think about the scale of the platform, nearly four million people a night stay in someone’s home, we’re probably already one of the biggest unifiers in the world. It is like the United Nations at kitchen tables. During the election, millions of Trump supporters were staying in the homes of Kamala Harris supporters and vice versa. And most of the time, they didn’t even realize that. So I think if there was ever a company to bring the world together, it would be a company like Airbnb that brings them together in the real world. But we’re under no delusion that we’re going to be the ones solving this problem. I do think there needs to be more unity in the world . . . and I want to be part of the solution, not part of the problem. I want to be the kind of company that helps get people off the phone, into the real world, and exposes other people to their cultures. It’s getting increasingly difficult to meet people. And increasingly, when people are colliding, they’re typically colliding with people they’re very similar to because [theyre in] these hermetically sealed social media bubbles. Or they collide, argue with each other on the internet, but they never change each other’s mind. So this is what we’re trying to do. And by the way, this is a multi-decade [ambition]. I’ve been doing this for 18 years. I would like to do this for another 18 years, or whatever. These are long-term views we’re taking. But I have to have a point of view. In the age of AI, what’s our purpose? I think our purpose is to help bring people together. As you’re approaching social networking, you have seen a lot go wrong in the industry. Are there things that are off the table right now for you? Are there things that you’re marking as guideposts to make sure you do it the right way? Yes. Well, number one, not having an ad-driven model. I think the business model needs to be aligned with the customers interest. And so that’s number one. Number two, I think there’s an expectation of safety on Airbnb, which I think there isn’t necessarily with social media. Parents want social media to be safe for their kids; I don’t know if the kids are asking for social media to be safe. But I think everyone that uses Airbnb has an expectation of safety. On social media, it’s kind of viewed as, like, safe and unsafe content. Safe speech is kind of political. Safety and everything else [on Airbnb] is not really a political issue. People expect it. And the reason people are okay with us holding them to safety standards is they want us to hold everyone else to safety standards. So I think the other thing is we want to do our best to design social interactions that we think are enriching to people and, I don’t want to make this interview about social media, but I think it’s safe to say that some social media improves my life, and some use of social media does not improve my life. And probably the biggest problem with social media is just the amount of time most of us use it. So not to say it’s good or bad, but it is probably bad if you’re using it many hours a day, because life should be experiencing the real world. Maybe that’s the biggest change of all, which makes this not social media or social networking: it primarily happens in the real world. We’re just the portal. By the way, one last point [on] social media: Notice the word social media. It used to be called social networking, but eventually your friends became your followers, intimacy became performing, and so really, there’s not much social networking anymore. As an aside, the old Facebook would still be a great product! I wish I had one product, and I knew what everyone’s birthday was, and I knew when somebody had a baby and they posted. But like, the old Facebook doesn’t even exist. No one’s on it, and it’s been optimized to be like a marketplace and other things. We’re not going to be that company, but there’s still a need for it. I know you don’t want this to be a social media discussion, but I think we’re already there. One other thing that occurs to me is that a lot of apps are pushing new social features lately. I made a short list before our call: Spotify, WhatsApp, Strava, Substack, Robinhood, and even Grindr have expanded social tools. Why do you think so many companies are doing this? I get your point about mission, but for most it feels like an engagement play. We’re doing social to the extent that we think it may enrich the experience for people that are looking for it. We’re not really doing it to engage people. We have a lot of traffic. We do need to convert more traffic to bookings to grow revenue, but we don’t really care how much time people spend on the app. We’re not really focused on how many times you open the app, because we’re not paid that way. We’re not paid for engagement with an advertising based model. Our incentives are a little bit different. And also, I don’t see social as primarily a revenue driver. I think it can be, but I’m primarily focused on something I think is far more important, which is making sure people have great experiences, or said differently, it’s great long-term revenue. People love Airbnb because they have great social experiences. Social features are not a way to make a quick buck in Airbnb. If I wanted to make quick bucks, you know, we would just go guns blazing on hotel expansion. Ive been thinking a lot about the two economies that we live in: which is basically everyone feeling pinched from the upper middle class downward. And then the upper echelon, the 1%. Were seeing more businesses catering to the higher end. You launched Reserve Now, Pay Later earlier this year in the U.S. 60% of customers are using it. How much is that about keeping Airbnb accessible to most people? And how are you thinking about that income divide right now when designing your platform? Let me start by saying, we had a choice to charge for Reserve Now, Pay Later. We could have charged interest, we could have charged fees, and we decided not to do that. It could have either been a way to be affordable and increase bookings, or it could have been a way to increase monetization. We chose to make it more affordable because we want as many people to use the feature as possible. And there are downsides! There’s revenue opportunities that we missed out on. We also don̵t have as much of a flow if people don’t prepay, we don’t hold as much money, we don’t make as much interest. So there is a cost. We think it’s worth the cost. I think Airbnb needs to remain a great, affordable brand and a value-based brand. I think we started off as a great brand. I think we lost some ground, especially during the pandemic. So there’s no cleaning fees in Airbnb any longer. All prices shown are the prices you get. We’re offering a lot of new pricing features. We’re trying to really make Airbnbs affordable. We measure the price increase of Airbnb versus hotels. For the last few years, hotel prices have appreciated faster than Airbnb, so we really myopically quickly focus on this. A lot of what we’re trying to do is just make Airbnb more affordable, more accessible. You’re right, in the Silicon Valley bubble, you’re surrounded by really, really wealthy people, and you can forget what life is like for everyday people. I mean, I grew up an everyday person. I grew up in upstate New York. My mom and dad were social workers. We lived in a $200,000 house. They made like $40,000 a year. And so Ive got to build a product for that Brian, not just today’s Brian or this little bubble that I live in now. And you know, we have luxury Airbnbs and all that, but the vast majority of our service has to be for everyday people, for middle class people. Youve talked about AI and your plan for Airbnb to evolve so the app is essentially just an assistive AI agent. How important is it for you to own the agentic experience? I think it’s really important. I mean, there’s many ways this whole thing could play out. I’m not opposed to a platform like ChatGPT becoming a platform where a lot of activity happens if the SDK is extremely robust. For example, the App Store: Apple does not make most of these apps in the App Store, and that’s because one company can’t make every app. But its okay that we’re inside of the App Store, because the SDK is so robust that we can do whatever we want. If the SDK is not very broad, then it’s probably not good. And we generally do want people to start on Airbnb. I think it’s important to note something. Almost all the technology in ChatGPT is now widely available to every other app in the world by API, so this technology is not proprietary. Now, maybe one day, the frontier companies might reserve the best models only for them and not make them available with APIs. But I don’t think travel ever needs the highest frontier model, and open source models are only three to six months behind a frontier model. And the average person using a travel application cannot discern the difference between a frontier model and a model three months behind. Plus you generally don’t want to use the frontier model for search anyway, but it is just too expensive and not optimized from a latency standpoint. So those models are for hardcore physics and research. They’re not for travel and lifestyle. I don’t think the frontier model is even the best model for it because of cost and latency. Also, we do want to be an AI company. In five years, or maybe seven years or however long it’s gonna take, every tech company is going to be an AI company. At some point, we won’t even ever use the word AI. Its just software. AI will just mean technology. In 1999 everyone said internet, and these were internet companies. And no one says internet company now. Everything’s on the internet. So I think AI is going to be so ubiquitous that everyone will be an AI company. Because you’ll have to be. If you’re not an AI company, it’s like, not using electricity, you’re just probably not going to exist. So the question is, who gets there first, and who’s the best at it? We want to be world-class at it. It’s really just about getting the best people of our generation. So we have some big hires that we’re making, really good talent that’s coming into Airbnb. I think what we want to do is we want to start with customer service. We made it personalized. We expanded action cards [to do things with a tap from the AI conversation], then we’re going to make it agentic. Agentic means it can take more than one action on your behalf. Then we’re going to bring it into [Airbnb] search. And then we’re going to connect the customer service and search into one app, under one agent and then, at that moment, the interface is essentially AI-native. So we design a new interface for [Airbnb], then it’s essentially an AI native app and an AI native company. And our goal is to become an AI company in, you know, the next few years. I think it’s interesting that youre broadcasting the exact UX evolution that youre planning. I mean, I don’t think it’s a secret. The only people I’d be worried about knowing it would be our competitors, but they chose a different strategy.
Category:
E-Commerce
Yelp users looking to learn more about restaurants, businesses, and other locations on the platform can now get information from an AI-powered Yelp Assistant. When logged-in users on Yelps iOS and Android apps visit particular business pages, they can now ask specific questions ranging from where to park to whether a restaurant offers vegan options. The answers are generated based on facts from reviews posted on Yelp, information provided to the platform by businesses, and businesses own websites, with relevant sections and even photos from Yelp reviews highlighted in the AI response. The assistant also provides a list of suggested questions to ask about a particular business. In addition, a new AI-powered feature called Popular Offerings separately highlights goods and services frequently mentioned and photographed in a businesss Yelp reviews. [Image: Yelp]In a demonstration for Fast Company, Akhil Kuduvalli Ramesh, Yelp senior vice president of product, highlighted how the Yelp Assistant could answer questions about a restaurants cuisine, parking, dog-friendly seating, and even the best times to avoid a long wait. The AI surfaced information from reviews and even user-submitted photos of canine-compatible outdoor tables. This isnt an AI thats hallucinating, says Kuduvalli Ramesh. This is an AI that is providing me answers, evidence first. Tapping into a rich vein of contentThe expansion of the Yelp Assistant, which launched last year with a focus on helping consumers find services from home repair to haircuts, comes as other tech companies from longtime rival Google to startups like Perplexity promote AI options for finding and booking tables at local restaurants, which has long been a key part of Yelps mission. But Yelp cofounder and CEO Jeremy Stoppelman says the companys wealth of business data, long trusted by consumers and partner services like Apple Maps, enables it to deliver reliable answers in an age of hallucinating AI. Were now at the point where you will be able to ask detailed questions about nearly every business on Yelp, he says. And its able to tap into the rich content, the reviews, and all of the survey informationeverything that weve been able to gather about that business. [Animation: Yelp]Stoppelman anticipates the Yelp Assistant will continue to grow more powerful, able to offer more detailed, personalized recommendations across businesses and categories. Already, thanks to Yelps AI advances, users can also now type or speak more involved natural language queries into Yelps search interface, with the AI able to parse requests like need help fixing a leaking faucet or dog-friendly brunch spot thats good for groups instead of simply matching on keywords. We would like people to basically query Yelp as though theyre talking to it, says Kuduvalli Ramesh. [Image: Yelp]And when Yelp users visit a particular restaurant, they can also now use the service to quickly find information about particular dishes by using the Yelp app to scan the menu with a new feature called Menu Vision. It can offer quick links to reviews and photos of Yelp-reviewed dishes, and the company plans to continue to expand the feature to include more options and information over time. Growing competitionYelps push into AI is far from the first time the company has incorporated new technology to meet new consumer tastes. In the heyday of Foursquare, Yelp added location check-in features. And as TikTok and other video platforms became increasingly important tools for restaurant discovery, Yelp enabled users to post short-form videos alongside text reviews and photos. The company, which for its most recent quarter reported $44 million net income on a record net revenue of $370 million, has also expanded into home services in recent years, taking on rivals like Angi and Thumbtack with its increasingly sophisticated tools to request quotes and communicate with home professionals. And in November 2024, Yelp also acquired RepairPal, which connects users with auto repair shops, and has been integrating its offerings into the core Yelp product. [Image: Yelp]Yelps AI also now offers guidance to service professionals on responding to customer requestsand awards visible badges to those with a pattern of helpful responses. A pair of paid AI phone tools, called Yelp Host and Yelp Receptionist, are now also rolling out to restaurants and other businesses, respectively, to field calls from customers. Yelp Host is able to book and modify reservations, including fielding special requests, while Yelp Receptionist can capture contact info and full call transcripts, then summarize the relevant information needed to get back to a potential customer. By the time the local business gets a lead, its a fully vetted lead, and its got an AI summary, says Craig Saldanha, Yelps chief product officer. They can listen to the transcript, but they can also read the summary in 30 seconds and essentially call you back with all of the information. Yelp Host and Receptionist are designed to let busy restaurants and businesses actually process phone calls, rather than simply amassing voicemails, while theyre closed for the day or assisting other customers. And those services, too, are enabled by the detailed information Yelp has amassed over more than two decades about specific businesses and different types of local merchants, Stoppelman says. It is grounded in the same underlying infrastructure, he says. And so, when you do sign up, we have a very good understanding of businesses, both in that category and then your business specifically.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||