DIRS.INFO The most relevant business & tech news

☆ OpenAI is taking a page out of Big Tech's playbook by reportedly building its own chips
▼ + stars: | 2024-10-31 | by ( Tom Carter | ) www.businessinsider.com time to read: +6 min

OpenAI is reportedly teaming up with Broadcom and TSMC to build custom AI chips, per Reuters. OpenAI could have its custom chips by 2026 but is reportedly dropping plans to build its own fabs. AdvertisementBuilding custom AI chips has long been the preserve of a select few tech companies — but OpenAI might be about to join the party. AdvertisementOpenAI's move, which will also reportedly see it incorporate AMD chips into its supply mix, means it would reduce its dependency on Nvidia, the market leader for AI chips. While it's unclear how much OpenAI's reported chip-building push will cost, creating custom AI chips doesn't come cheap.

Persons: OpenAI, —, Kate Leaman, Rahul Kulkarni, Maia, Meta, Mark Zuckerberg, Gil Luria, Davidson, Luria, Edward Wilford, Sam Altman, Altman, Pierre Ferragu Organizations: Broadcom, Meta, Google, Service, Reuters, BI, Microsoft, Apple, Tech, Amazon Web, AWS, Nvidia, Big Tech, Street, New, Research, New York Times

☆ Google VP Amin Vahdat leads the company in innovating with AI chips
▼ + stars: | 2024-10-24 | by ( Hugh Langley | ) www.businessinsider.com time to read: +1 min

With its core business under threat, Google has spent the past two years pouring resources into building its own AI chips. In charge of those efforts is Vahdat, one of Business Insider's 2024 AI Power List picks. Vahdat has been at the company for more than a decade and is today setting the direction for Google's silicon strategy. That also means working closely with Google DeepMind to take its breakthrough models and integrate them throughout Google's products, whether YouTube's creator tools or Google's search ads. See Business Insider's full AI Power List

Persons: Vahdat Organizations: Google, Amazon, Microsoft

☆ How Google makes custom chips used to train Apple AI models and its own chatbot, Gemini
▼ + stars: | 2024-08-23 | by ( Katie Tarasov | ) www.cnbc.com time to read: +7 min

He's been covering Google's custom cloud chips since they launched in 2015. Google was the first cloud provider to make custom AI chips. Microsoft 's first custom AI chip, Maia, wasn't announced until the end of 2023. But being first in AI chips hasn't translated to a top spot in the overall rat race of generative AI. Google TPUs still dominate among custom cloud AI accelerators, with 58% of the market share, according to The Futurum Group.

Persons: Google's, Daniel Newman, He's, Maia, wasn't, Gemini, OpenAI's ChatGPT, Newman, Amin Vahdat, Marc Ganley, Vahdat, they've, Stacy Rasgon, Rasgon, it's Organizations: Google, Apple, Apple Intelligence, Nvidia, Futurum, Amazon Web Services, Microsoft, CNBC, Systems, Unit, Bernstein Research, Broadcom, TSMC, Trillium, Taiwan Semiconductor Manufacturing Company, Intel, Samsung Locations: Mountain View , California, China, Taiwan, U.S

Share Share Article via Facebook Share Article via Twitter Share Article via LinkedIn Share Article via EmailInside Google's chip lab, where it makes custom silicon to train Gemini and Apple AI modelsGoogle was the first cloud provider to make its own custom AI chips, called TPUs, when they first came out in 2015 - a trend both Amazon and Microsoft followed years later. Now, Apple has revealed it uses TPUs to train its AI models, positioning Google chips as an alternative to Nvidia's market-leading GPUs. CNBC got an exclusive look inside the lab where Google makes its chips, where its top executive showcased TPU Version 6, and its new Arm-based CPU, Axion, both coming out later in 2024.

Persons: Apple Organizations: Gemini, Apple, Google, Microsoft, CNBC

☆ Apple's new AI is made in Google data centers
▼ + stars: | 2024-06-14 | by ( Hugh Langley | ) www.businessinsider.com time to read: +5 min

But Apple and Google had for months been working together behind the scenes, with Google giving Apple access to its data centers to train the iPhone maker's new AI models. For years, Apple has leaned on Google and Amazon's cloud services to store data for its products. For example, when Apple device users do iCloud backups they are often stored in Google's data centers. Most of the impressive features of AI models must be at least partly handled in massive energy-sucking data centers, which companies such as Microsoft, Google and Amazon have spent years building. Bloomberg previously reported that Google and Apple were also in talks to bring Google's Gemini AI to iOS devices.

Persons: —, OpenAI, Googlers, It's, Craig Federighi Organizations: Service, Apple, Google, Business, Apple Intelligence, Nvidia, Microsoft, Oracle, Bloomberg

☆ Wall Street showers an AI chipmaker we own with praise — and it's not Nvidia
▼ + stars: | 2024-06-07 | by ( Kevin Stankiewicz | ) www.cnbc.com time to read: +7 min

JPMorgan upped its growth forecast for the custom AI chip market and said Broadcom's dominance in the space is poised to continue. Essentially, these networking chips help make up the plumbing of data centers. The second line of Broadcom's AI business is custom processors, where the company is best-known for co-designing Alphabet's tensor processing units, or TPUs. Google-parent Alphabet and Broadcom have collaborated on the creation of the AI chips for years, with the sixth-generation TPU being announced last month . The AI chips made by market leader Nvidia and rival AMD are more general purpose.

Persons: Bernstein, Melius, —, Jensen Huang, Jim Cramer's, Jim Cramer, Jim, Hock Tan, Broadcom Lucas Jackson Organizations: Nvidia, Wall, Broadcom, Melius, JPMorgan, Wall Street, VMWare, Google, Amazon, Microsoft, Marvell Technology, Marvell, AMD, CNBC

☆ 'Going up against them is insane.' Brave startups are trying to usurp Nvidia's AI throne.
▼ + stars: | 2024-06-06 | by ( Alistair Barr | ) www.businessinsider.com time to read: +9 min

You could have invested a lot of money in AI chip startups several years ago. "Us going up against them is insane," said Thomas Sohmers, CEO of Positron AI, a chip startup that emerged from stealth a few months ago. "The model of AI chip startups needing to raise 10s to 100s of millions of dollars before they actually test product market fit is insane," he added. TigrisOpenAI's Altman may also be working on an AI chip startup of his own. So it stands to reason that Masa has been trying to raise money for an AI chip startup.

Persons: —, Graphcore, Thomas Sohmers, Nvidia Thomas Sohmers, Sohmers, Groq, Googlers, Puneet Kumar, Mark Hayter, Rivos, Andrew de los Reyes, Gavin Uberti, Chris Zhu, Nvidia's, OpenAI's Sam Altman, OpenAI's Altman, Altman, Masa, Softbank, Son Organizations: Service, Microsoft, Samsung, Business, Nvidia, MediaNews, Boston Herald, Getty, MIT, Google, Chrome, Matrix Capital, Intel, Harvard, Bloomberg News, Bloomberg Locations: Silicon Valley, Swiss, Groq, GroqWare, WeWork

☆ Google CEO Pichai says company will 'sort it out' if OpenAI misused YouTube for AI training
▼ + stars: | 2024-05-14 | by ( Jordan Novet | ) www.cnbc.com time to read: +4 min

Alphabet CEO Sundar Pichai said Google will "sort it out" if it determines Microsoft -backed OpenAI relied on YouTube content to train an artificial intelligence model that can generate videos. The New York Times later reported that OpenAI had transcribed over a million hours of YouTube videos. Asked if Google would sue OpenAI if the startup violated the search company's terms of service, Pichai didn't offer specifics. Pichai said Google has processes in place to figure out if OpenAI failed to comply with the rules. WATCH: Alphabet CEO on report OpenAI trained GPT-4 on YouTube: We have clear terms of service

Persons: Sundar Pichai, OpenAI, CNBC's Deirdre Bosa, Mira Murati, Murati, Pichai, Apple, Bosa, We'll, GPT Organizations: Microsoft, Wall Street Journal, New York Times, Google, Google's, Astra, Apple, Developers Conference, Bloomberg, Gemini Locations: U.S, Cupertino , California

☆ Smartphone makers are dreaming of a 'supercycle' driven by AI. Analysts disagree
▼ + stars: | 2024-02-27 | by ( Ryan Browne | ) www.cnbc.com time to read: +6 min

Samsung has gone big on generative AI with its Galaxy S24 Ultra smartphone. When was the last smartphone supercycle? Smartphone makers have been dreaming of a "supercycle" in their industry, driven by AI, after a bruising few years that saw device sales slow aggressively. 'AI phone era'Major smartphone players are betting that a supercycle is about to happen thanks to AI. "We're right now at the dawning of an entirely new era, an AI phone era," Kitto said.

Persons: Kim Hong, they're, Francisco Jeronimo, Jeronimo, Ben Wood, Wood, James Kitto, Kitto, Brian Rakowski, Rakowski, We've Organizations: Samsung Electronics, Reuters, Reuters BARCELONA –, Samsung, Google, Galaxy, Apple, Financial Times, Mobile, Congress, Huawei, Qualcomm, IDC, Twitter, CNBC, 3G, CCS, Samsung's, Google's, Android Locations: Seoul, South Korea, Reuters BARCELONA, Chertsey , England

☆ Big Tech's AI spending plans mean big business for 3 portfolio stocks, including Nvidia
▼ + stars: | 2024-02-02 | by ( Kevin Stankiewicz | In Kevinstankiewicz | ) www.cnbc.com time to read: +9 min

Some of Amazon's spending is likely to go toward its custom AI chips, known as Trainium and Inferentia. But the Seattle-based tech giant also buys Nvidia chips. On Thursday's post-earnings call, CEO Andy Jassy said AWS offers the "most expansive collection of compute instances with Nvidia chips." Alphabet's spending on AI chips has typically been split between its custom chips designed in partnership with Broadcom, known as Tensor Processing Units (TPUs), and Nvidia's offerings. Including networking products that stitch together parts of the data center, about 15% of Broadcom's semiconductor revenue was tied to generative AI spending in fiscal 2023.

Persons: Eaton, Brian Olsavsky, FactSet, Amazon's capex, Andy Jassy, Meta's capex, Susan Li, Li, Mark Zuckerberg, Zuckerberg, Meta, it's, —, Amy Hood, It's, chatbot ChatGPT, AMD's, Ruth Porat, Porat, Jim Cramer's, Jim Cramer, Jim, Florian Gaertner Organizations: Big Tech, Nvidia, Broadcom, Eaton, Microsoft, Meta, Google, Web Services, Facebook, Devices, OpenAI, supercomputing, TPUs, CNBC, YouTube, Photothek, Getty Locations: Eaton, Seattle

Google and Character AI did not respond to requests for comment. The demographic is helping the company position itself as the purveyor of more fun personal AI companions, compared to other AI chatbots from OpenAI's ChatGPT and Google's Bard. Character.AI is also in talks to raise equity funding from venture capital investors, which could value the company at over $5 billion, sources said. The talks with Google are ongoing and terms of the deal could change, said the sources, who requested anonymity as the discussions are private. Anthropic uses Google's cloud services as well as its latest version of TPUs.

Persons: Character.AI, Noam Shazeer, Daniel De Freitas, Billie Eilish, Character.AI's, Google's Bard, Andreessen Horowitz, Lina Khan, Krystal Hu, Anna Tong, Jeffrey Dastin, Bill Berkrot Organizations: Google, Reuters, Microsoft, . Federal Trade Commission, Thomson Locations: OpenAI, ., San Francisco, New York

☆ Why the case for owning Broadcom is compelling with or without its $61 billion takeover of VMWare
▼ + stars: | 2023-10-30 | by ( Kevin Stankiewicz | In Kevinstankiewicz | ) www.cnbc.com time to read: +6 min

Club holding Broadcom 's (AVGO) investment prospects look bright – with or without cloud software maker VMWare (VMW) under its corporate roof. For now, though, the companies remain apart after Broadcom's planned $61 billion takeover of VMWare did not close Monday, as previously scheduled. AVGO YTD mountain Broadcom YTD With VMWare, Broadcom would be a company with sales split roughly 50-50 between its traditional semiconductor division — an artificial intelligence beneficiary — and software, which typically is higher-margin revenue. Without VMWare, Broadcom would have freed up its capacity to reward shareholders in a different, but still meaningful, way. Even with the buyback silver lining, our hope is Broadcom completes the VMWare deal because it should boost Broadcom's earnings over time.

Persons: Broadcom's, CNBC's David Faber, Jim Cramer, Hock Tan, Jim, Broadcom's Tan, CNBC's Faber, There's, Jim Cramer's, Broadcom Lucas Jackson Organizations: Broadcom, VMWare, Club, Nvidia, Google, European Union, Financial Times, U.S, Intel, Semiconductor, Qualcomm, NXP Semiconductors, Microsoft, Activision, CNBC Locations: United Kingdom, Japan, Canada, China, California, Beijing

Tesla needs as many Nvidia chips as it can get, based on CEO Elon Musk recent comments on the company's quarterly earnings call. Tesla isn't the first to work on custom AI chips focused specifically on the company's bread and butter. And that's coming from the head of one of the most cutting-edge AI companies on the planet. Nvidia chips can be utilized by clients for many applications, making it easier for them to accelerate growth. The fresh Nvidia-Tesla conversation adds another layer to Oracle's report Monday, when viewed through the lens of Nvidia's AI leadership.

Persons: Tesla, Morgan Stanley, Morgan, Jim Cramer, Elon Musk, Musk, Sundar Pichai, Jim, Oracle, Jim Cramer's, Jensen Huang Organizations: Nvidia, Oracle, Broadcom, Google, NVIDIA, Citigroup, Oracle Cloud Infrastructure, Citi, Amazon Web Services, UBS, CNBC, Getty Locations: OCI, Taiwan

☆ Broadcom's post-earnings pullback doesn't shift the fundamentals. It's a buying opportunity
▼ + stars: | 2023-08-31 | by ( Zev Fima | ) www.cnbc.com time to read: +7 min

Excluding these AI-related sales, semiconductor sales have largely stabilized around the $6 billion level and management continues to think that a "soft landing" is still in the cards. In fact, networking revenue was up 20% versus the year ago period and now represents 40% of revenue at the company's semiconductor segment. Elsewhere within semiconductor solutions, wireless revenues (24% of segment revenue) were flat year-over-year, as was revenue for server storage connectivity (17% of segment revenue). Broadband sales (16% of segment revenue) moderated to 1% year-over-year growth following nine-consecutive quarters of double-digit growth. Industrial sales (about 3% of segment revenue) were down 3% year-over-year.

Persons: Hock Tan, Tan, Jim Cramer's, Jim Cramer, Jim, Broadcom Lucas Jackson Organizations: Semiconductor, Broadcom, Revenue, VMWare, Club, Management, Outlook Management, CNBC

☆ AI in focus: Here's what to look for when Salesforce and Broadcom report earnings
▼ + stars: | 2023-08-29 | by ( Kevin Stankiewicz | In Kevinstankiewicz | ) www.cnbc.com time to read: +7 min

The earnings from Salesforce and Broadcom follow a rough patch for technology stocks that began in mid-July. .SPX AVGO,CRM 3M mountain The S & P 500's performance versus Broadcom and Salesforce over the past three months. Salesforce For the three months ended July 31, Wall Street expects Salesforce's revenue to jump more than 10% year-over-year, to $8.53 billion, according to Refinitiv. Broadcom's AI exposure currently falls into two primary buckets. While Broadcom's AI networking business is no doubt fast-growing, Vinh noted there are headwinds within other pockets of its wide-ranging semiconductor business, such as traditional enterprise networking and broadband.

Persons: Wednesday's, SPX, Scott Berg, Berg, Wells Fargo, —, Salesforce, Needham's Berg, Hock Tan, Bard chatbot, John Vinh, Bard, Vinh, Jim Cramer's, Jim Cramer, Jim, Mike Blake Organizations: Club, Broadcom, Wall, Needham &, CNBC, Salesforce, Google, Jericho, UBS Locations: Irvine , California

And that $1 trillion of data centers is in the process of transitioning into accelerated computing and generative AI. Additionally, Nvidia announced AI-ready servers to support the VMware Private AI Foundation and help businesses customize and deploy generative AI applications using their proprietary data. Shares of AMD have also surged higher in this year's AI trade, though the role the company plays in the field is less clear. As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust's portfolio.

Persons: Jim Cramer's, Jensen Huang, Huang, ChatGPT, Nvidia's Huang, It's, We've, Grace Hopper Superchip, it's, hyperscalers, Jim Cramer, Jim, Eduardo MunozAlvarez Organizations: Broadcom, Jim Cramer's Charitable Trust, Devices, AMD, Moore's, Microsoft, Broadcom's, VMWare, VMware, Nvidia, AI Foundation, U.S . Federal Trade Commission, Intel, Tower Semiconductor, ARM, CNBC, View, Getty Locations: China

☆ RBC: These 19 stocks with an early lead in the AI revolution will continue to profit from rising demand in semiconductors, cloud computing, and more
▼ + stars: | 2023-08-10 | by ( Mark Reeth | ) www.businessinsider.com time to read: +3 min

RBC identified the stocks profiting from AI today while positioning themselves for future success. 19 stocks profiting from rising AI demandA number of themes leapt out from earnings reports and calls with management, the RBC analysts wrote. Companies leading the charge in AI investments enjoyed strong second-quarter earnings, thanks in part to those very investments. But AI isn't just bringing money in — companies are also spending more on AI than ever, RBC noted. After poring through earnings announcements from AI-related companies, RBC analysts identified the following 19 stocks that they believe are benefiting most from AI demand right now.

Persons: Jonathan Atkin Organizations: RBC, RBC Capital Markets, Companies, Microsoft, Google, NVIDIA, NVIDIA's, Nvidia, Oracle

☆ Nvidia wins again — plus two more takeaways from this week's mega-cap earnings
▼ + stars: | 2023-07-26 | by ( Zev Fima | ) www.cnbc.com time to read: +8 min

It was also one of several conclusions we made combing through the results and listening to the conference calls with investors. We are basically looking for results or management commentary that speaks to peer companies or the industry as a whole. On the call, management said it was "to support cloud demand, including investments in AI infrastructure." What is increased AI spending going to do for Meta's engagement and Reels monetization efforts? Nvidia chief Jensen Huang (centre L) poses for photographs before attending a press conference at Computex 2023 in Taipei on May 30, 2023.

Persons: it's, Sundar Pichai, We're, Jim Cramer's, Jim Cramer, Jim, Jensen Huang, Sam Yeh Organizations: Microsoft, Nvidia, Google, AMD, Intel, Jim Cramer's Charitable, CNBC, Afp, Getty Locations: Taipei

☆ Google reveals its newest A.I. supercomputer, says it beats Nvidia
▼ + stars: | 2023-04-05 | by ( Kif Leswing | ) www.cnbc.com time to read: +3 min

While Nvidia dominates the market for AI model training and deployment, with over 90%, Google has been designing and deploying AI chips called Tensor Processing Units, or TPUs, since 2016. On Tuesday, Google said that it had built a system with over 4,000 TPUs joined with custom components designed to run and train AI models. It's been running since 2020, and was used to train Google's PaLM model, which competes with OpenAI's GPT model, over 50 days. Google's TPU-based supercomputer, called TPU v4, is "1.2x–1.7x faster and uses 1.3x–1.9x less power than the Nvidia A100," the Google researchers wrote. For example, Google said that Midjourney, an AI image generator, was trained on its TPU chips.

Executives across the technology sector are talking about how to operate AI like ChatGPT while accounting for the high expense. What makes this form of AI pricier than conventional search is the computing power involved. Still, footing the bill is one of two main reasons why search and social media giants with billions of users have not rolled out an AI chatbot overnight, said Paul Daugherty, Accenture's chief technology officer. Technology experts also said a workaround is applying smaller AI models to simpler tasks, which Alphabet is exploring. The company said this month a "smaller model" version of its massive LaMDA AI technology will power its chatbot Bard, requiring "significantly less computing power, enabling us to scale to more users."

Search resuls for: "TPUs"

20 mentions found