tag

Since the launch of ChatGPT three years ago, analysts and technologists — even a Google engineer and the company’s former chief executive officer — have declared Google behind in the high-stakes race to develop artificial intelligence.Not anymore.The internet giant has released new AI software and struck deals, such as a chip tie-up with Anthropic PBC, that have reassured investors the company won’t easily lose to ChatGPT creator OpenAI and other rivals. Google’s newest multipurpose model, Gemini 3, won immediate praise for its capabilities in reasoning and coding, as well as niche tasks that have tripped up AI chatbots. Google’s cloud business, once an also-ran, is growing steadily, thanks in part to the global rush to develop AI services and demand for compute.And there are signs of rising demand for Google’s specialised AI chips, one of the few viable alternatives to Nvidia Corp’s dominant gear. A report on Monday that Meta Platforms Inc is in talks to use Google’s chips sent shares of its parent Alphabet Inc climbing. The stock has added nearly $1tn in market capitalisation since mid-October, helped by Warren Buffett taking a $4.9bn stake during the third quarter and broader Wall Street enthusiasm for its AI efforts.Alphabet shares rose 1.5% to $323.44 in New York on Tuesday, sending the company’s market capitalisation near $4tn.SoftBank Group Corp, one of OpenAI’s biggest backers, fell 10% Tuesday on worries about the competition from Google’s Gemini. Nvidia shares dropped 2.6%, erasing $115bn in market value.“Google has arguably always been the dark horse in this AI race,” said Neil Shah, analyst and co-founder at Counterpoint Research. It’s “a sleeping giant that is now fully awake.”For years, Google executives have argued that deep, costly research would help the company fend off rivals, defend its turf as the leading search engine and invent the computing platforms of tomorrow. Then ChatGPT came along, presenting the first real threat to Google search in years, even though Google pioneered the tech underpinning OpenAI’s chatbot. Still, Google has plenty of resources that OpenAI doesn’t: a corpus of ready data to train and refine AI models; flowing profits; and its own computing infrastructure.“We’ve taken a full, deep, full-stack approach to AI,” Sundar Pichai, CEO of Google and Alphabet, told investors last quarter. “And that really plays out.”Any concerns that Google might be held back by regulators are dying away. The company recently avoided the most severe outcome from a US anti-monopoly case — a breakup of its business — in part because of the perceived threat from AI newcomers. And the search giant has shown some progress in the longtime effort to diversify beyond its core business. Waymo, Alphabet’s driverless car unit, is coming to several new cities and just added freeway driving to its taxi service, a feat made possible by the company’s enormous research and investment.Some of Google’s edge comes from its economics. It’s one of the few companies that produces what the industry calls the full stack in computing. Google makes the AI apps people use, like its popular Nano Banana image generator, as well as the software models, the cloud computing architecture and the chips underneath. The company also has a data goldmine for constructing AI models from its search index, Android phones and YouTube — data that Google often keeps for itself. That means, in theory, Google has more control over the technical direction of AI products and doesn’t necessarily have to pay suppliers, unlike OpenAI.Several tech companies, including Microsoft Corp and OpenAI, have plotted ways to develop their own semiconductors or forge ties that make them less reliant on Nvidia’s bestsellers. For years, Google was effectively its own sole customer for its homegrown processors, called tensor processing units, or TPUs, which the company first designed more than a decade ago to speed up the generation of search results and has since adapted to handle complex AI tasks. That’s changing. AI startup Anthropic said in October said it would use as many as 1mn Google TPUs in a deal worth tens of billions of dollarsOn Monday, tech publication the Information reported that Meta planned to use Google’s chips in its data centres in 2027. Google declined to address the specific plans, but said that its cloud business is “accelerating demand” for both its custom TPUs and Nvidia’s graphics processing units. “We are committed to supporting both, as we have for years,” a spokesperson wrote in a statement.Meta declined to comment on the report on Monday night.“We’re delighted by Google’s success,” a spokesperson for Nvidia said in a statement Tuesday. “They’ve made great advances in AI, and we continue to supply to Google.” The spokesperson added: “Nvidia is a generation ahead of the industry – it’s the only platform that runs every AI model and does it everywhere computing is done.”Analysts read the Meta news as a signal of Google’s success. “Many others have failed in their quest to build custom chips, but Google can clearly add another string to its bow here,” Ben Barringer, head of technology research for Quilter Cheviot, wrote in an email.Google has taken risks to get here.In early 2023, Google consolidated its AI efforts under Demis Hassabis, the leader of its London AI lab DeepMind. The reshuffle had some bumps, most notably a botched rollout of an image-generation product. For several years, DeepMind pursued research in areas like protein-folding that led to new commercial strategies (and a Nobel Prize) but contributed little to Google’s bottom line. Under the reorganisation, the AI unit is focused almost squarely on foundational models that keep pace with OpenAI, Microsoft and others.Hassabis, a renowned computer scientist, has helped retain key AI engineers despite multimillion-dollar offers from rivals. His boss, Pichai, has been willing to splurge on talent.Gemini 3 Pro has risen to the top of closely watched AI leaderboards on LMArena and Humanity’s Last Exam. Andrej Karpathy, a founding member of OpenAI, said it’s “clearly a tier 1 LLM,” referring to large language models. Google pitched the model as one that can solve complex science and math problems, and address nagging issues — such as generating images and overlaid text with incorrect spelling — that might deter enterprise customers from adopting AI services more widely.Consumer interest is harder to gauge. Google said last week that 650mn people use its Gemini app. OpenAI recently said ChatGPT hit 800mn weekly users. As of October, Gemini’s app had 73mn monthly downloads, well shy of ChatGPT’s 93mn monthly downloads, according to research firm Sensor Tower.Google is an advertising behemoth, but it has historically struggled to find other commercial models. Its cloud business reported third-quarter revenue of $15.2bn, up 34% from the prior year. Still, that remains in third-place behind Microsoft and Amazon Web Services, which posted more than double Google’s cloud sales in the most recent quarter. Counterpoint Research’s Shah said Google’s AI adoption with enterprises lags Microsoft and Anthropic.Meanwhile, OpenAI is targeting profits by selling a premium version of ChatGPT and adjacent software to companies. It’s cutting deals with chipmakers from Broadcom Inc to Advanced Micro Devices Inc to Nvidia to support its AI ambitions.Google’s TPUs are mostly attractive to a handful of companies with big computing bills, like Meta and Anthropic, said Meryem Arik, CEO of the AI startup Doubleword.And the chip industry is “not a zero-sum game with just one winner,” said Barringer.For one, AI developers can only access Google’s chips through the company’s own cloud service. They can use Nvidia’s graphics processing units, or GPUs, more flexibly. “As soon as you use TPUs, you’re locked into” the Google cloud ecosystem, said Arik.Being tied to a single supplier might have been something companies avoided. That’s no longer the case for Google, thanks to its advances in AI.“It’s definitely fair to say that Google is back in the game with Gemini 3,” said Thomas Husson, analyst at Forrester. “In fact, to paraphrase a quote attributed to Mark Twain, reports of Google’s death have been widely exaggerated, not to say irrelevant.” 

The Google team, including Executive Vice President of Generative Artificial Intelligence, Oliver Parker visited Al Jazeera Media Network’s headquarters on November 9 to discuss ways to collaborate in the fields of strategic development, artificial intelligence, and modern technologies, reported Al Jazeera Media Network.The delegation met with Sheikh Nasser bin Faisal al-Thani, Director General of Al Jazeera Media Network, where he highlighted that the Network is working towards establishing a global technological ecosystem to strengthen its position as a leading media institution in the field of adopting artificial intelligence technologies.In another meeting, the visiting delegation met with Ahmad Alyafei, Executive Director of Channels, Ahmad al-Fahad, Executive Director of Technology and Network Operations, Mounir Daymi, Executive Director of the Digital Division, Eman al-Amri, Director of Al Jazeera Media Institute, and the Network’s Strategy Review Committee.The Google team discussed with the Network Strategy Review Committee possible ways of co-operating and implementing Al Jazeera’s projects and creative initiatives to achieve the Network’s vision and goals.One of the most prominent projects presented at the meeting was the idea of the “The Core” project, the first integrated operational model that Al Jazeera is working to implement to integrate journalism with artificial intelligence technologies. It is redefining the formulation and production of news and making humans partners in the production process rather than just a user of technology and a recipient of news.The Google delegation included Yousri Mhedheb, Senior Digital Adviser of Google Cloud, Alex Rutter, EMEA Managing Director of AI, and Ghassan Kosta, Regional General Manager of Google Cloud, who expressed their pleasure to collaborate with Al Jazeera Media Network, as one of the leading media organisations in the region and the world.They toured Al Jazeera’s channels and newsrooms, and learned about the advanced technologies used in the production of media content.It is noteworthy that this meeting was preceded by a series of introductory meetings organised by the Network’s Strategy Review Committee with the Google team and its officials concerned with AI and cloud technologies, to discuss the mechanisms of joint co-operation between the two parties.

Google Photos has introduced a major update of AI-powered features, including advanced editing tools, a redesigned "Ask" button, and expanded global search capabilities, the company announced on Tuesday.The enhancements enable users to edit photos simply by typing or speaking commands - such as "remove John's sunglasses" or "make Sarah smile" - and apply changes across people and elements in the image. This "Help me edit" functionality is now available on iOS in the United States, following its earlier launch on select Android models.A highlight of the update is the introduction of the AI model Nano Banana, which powers creative transformations - users can now turn snapshots into fine-art-style portraits, comics, or vintage illustrations.The app's redesigned editor also adds an "Ask" button: while viewing a photo, users can tap to ask questions about its content ("What year was this taken"), discover related moments, or request edits via natural language.Meanwhile, the AI-powered search feature in Photos is being expanded to more than 100 countries and regions and will support over 17 additional languages including Arabic, French, German, Hindi, Japanese and Portuguese.The update signals Google's push to bring generative-AI and natural-language capabilities more deeply into everyday mobile applications, and positions Google Photos as not just a storage tool but a creative and intelligent assistant for imagery.

Google has announced the expansion of its AI Mode search experience to more than 35 new languages, including Arabic, extending the feature's availability to over 200 countries and territories worldwide. The company said the update is powered by its latest Gemini AI model, which provides advanced natural language understanding and enhanced handling of multimedia content. Google noted that Gemini's reasoning and multi-modal capabilities allow the search engine to deliver results that are more accurate and contextually relevant to each region's culture and language. AI Mode is designed to assist users with exploratory questions and complex tasks, offering more dynamic and conversational search interactions. According to Google, users engaging with AI Mode now tend to ask questions nearly three times longer than those in traditional searches, signaling a shift toward deeper and more analytical exploration. The company also introduced a new underlying technology called "Query Expansion," which breaks a user's question into multiple subtopics and runs numerous searches simultaneously. This approach allows users to explore a broader range of information compared to standard search queries. Google first launched AI Mode in the Middle East and North Africa region in English last August. The addition of Arabic marks a significant step toward providing more precise, culturally relevant, and linguistically accurate responses for Arabic-speaking users across various domains.

YouTube unveiled a suite of generative AI tools at its Made-on YouTube event, led by Veo 3 Fast, a streamlined version of Google DeepMind's video model. The system produces 480p clips with audio almost instantly, marking the first time sound has been added to Veo-generated Shorts. It is rolling out in the US, UK, Canada, Australia and New Zealand, with more markets to follow.The company also announced new AI editing features, including motion transfer from video to still images, text-based styling and object insertion, and Speech to Song Remixing, which converts dialogue into music using DeepMind's Lyria 2 model. Testing will begin in the US before global expansion.Another feature, Edit with AI, automatically assembles raw footage into a rough cut with transitions, music and interactive voiceover. YouTube said the tool is in trials and will launch in select markets in the coming weeks.All AI-generated Shorts will carry watermarks and labels, as YouTube seeks to accelerate creator adoption of Shorts while maintaining transparency.