• Huawei launches AsyncFlow, a new AI training framework for large language models
• AsyncFlow improves throughput by up to 2.03 times over traditional methods
• The system introduces TransferQueue for dynamic load balancing and streaming
• While promising, AsyncFlow’s performance in real-world settings is yet to be tested

Huawei has unveiled a cutting-edge AI training framework called AsyncFlow, aiming to dramatically improve the speed and scalability of post-training processes for large language models.

The announcement marks a major milestone in the company’s efforts to advance reinforcement learning systems while pushing further toward technological self-sufficiency.

Boosting Model Training with AsyncFlow

Developed by researchers at Huawei and led by Zhenyu Han, AsyncFlow introduces an asynchronous streaming reinforcement learning architecture tailored for the complex post-training phase of large language models (LLMs).

Traditional reinforcement learning-based post-training can be computationally intensive and difficult to scale, often bottlenecked by inefficient data handling and resource management. AsyncFlow seeks to overcome these limitations by rethinking how data flows through the training pipeline.

TransferQueue Delivers Key Performance Gains

At the heart of the new framework is a distributed data management module known as TransferQueue. This component plays a central role in boosting efficiency by automatically balancing workloads and allowing overlapping of different processing stages.

The result is a significant increase in throughput, with AsyncFlow achieving a 1.59 times improvement on average over conventional systems. In large-scale cluster setups, throughput gains reached up to 2.03 times, demonstrating the framework’s capability to scale efficiently.

Real-World Promise, But Caution Ahead

This development is particularly relevant as the AI industry faces growing demand for faster, more cost-effective training of increasingly complex models.

The researchers note that AsyncFlow not only maintains training stability but also optimizes how computational resources are used, potentially leading to substantial savings in time and infrastructure costs.

Beyond academic achievement, the framework holds real-world potential. Industries such as healthcare, finance, and autonomous driving could benefit from faster model adaptation and real-time data processing. For example, systems that need to react quickly to changing environments, like autonomous vehicles or live translation tools, could see direct performance enhancements from AsyncFlow’s architecture.

Still, the technology is not without limitations. So far, AsyncFlow has shown strong performance in controlled experimental conditions. Its resilience in more unpredictable, real-world dataflows remains to be seen. Further testing and adaptation will be necessary before the system can be widely deployed in production environments.

Part of Huawei’s Bigger AI Vision

AsyncFlow also complements Huawei’s broader AI and software strategy. Earlier this week, the company announced that it will open-source its proprietary programming language, Cangjie, later this month.

The move reflects a growing effort to reduce reliance on foreign technology by building a homegrown ecosystem of tools, languages, and infrastructure. This includes HarmonyOS and CloudMatrix AI racks, which support a fully integrated AI and software environment.

Taken together, these developments show that Huawei is not just keeping pace with global AI advancements, but is actively shaping them. With AsyncFlow, Huawei offers a glimpse into a more efficient future for AI model training, one that could cut costs, speed up deployment, and ultimately make large-scale AI systems more accessible across industries.

The post Huawei Unveils AsyncFlow to Turbocharge AI Model Training Efficiency appeared first on CoinCentral.

• The EU is proceeding with its AI Act despite heavy lobbying for a delay.
• Tech giants argue the law risks damaging European AI competitiveness.
• The phased rollout continues, with key obligations taking effect in August 2025.
• While some flexibility is possible, no formal delays have been granted.

The European Union has reaffirmed its commitment to rolling out the AI Act on schedule, despite mounting pressure from a coalition of tech companies and industry leaders.

Calls to delay the legislation, including an open letter from over 100 CEOs urging a two-year pause, were firmly rejected by the European Commission on Friday.

Thomas Regnier, spokesperson for the Commission, directly addressed the campaign, which included voices from Alphabet, Meta, Mistral AI, ASML, Airbus, and even retailers like Carrefour.

In a pointed message, Regnier said there would be “no stop the clock, no grace period, no pause,” signaling that the bloc intends to proceed as planned. The AI Act, first introduced in 2021, is now entering its next phase, with obligations for general-purpose AI models set to begin in August 2025.

Concerns over competitiveness fall flat

Opponents of the law argue that the regulation risks undermining Europe’s ability to stay competitive in the global AI race. Their central concerns include overlapping compliance requirements, ambiguous definitions, and limited clarity around how to implement the rules.

Despite these concerns, EU officials appear confident that the phased approach to implementation offers companies enough time to prepare.

Some provisions have already taken effect as of early 2025, and obligations targeting high-risk AI applications will not be enforced until August 2026. Yet even this gradual rollout has done little to calm fears among both large players and smaller AI startups, who say the uncertainty may stall innovation and slow market entry.

Clear message from the Commission

The Commission has framed the AI Act as a crucial step toward responsible and trustworthy artificial intelligence. The law prohibits a narrow set of AI use cases deemed “unacceptable risk,” such as manipulative behavioral targeting or social scoring. At the same time, it imposes stricter obligations on high-risk systems in areas like facial recognition, employment screening, and education tools.

Developers of general-purpose AI models, including large language models and chatbot systems, will soon have to comply with transparency standards. These include documentation, risk assessments, and, in some cases, registration with EU authorities before being offered in the single market.

Window of flexibility still open

While the Commission’s current position is unwavering, some ambiguity remains. European Commissioner Henna Virkkunen has previously indicated a willingness to temporarily defer certain aspects of the law if supporting guidelines or technical standards are not ready in time. For now, however, the legal deadlines remain intact.

This opens a narrow path for adaptation without halting the overall regulatory project. Still, industry leaders continue to push back, with the #stoptheclock campaign gaining momentum. But unless there is a significant policy shift, companies must prepare to meet the law’s requirements on the original timeline.

The post EU Holds Firm on AI Act Timeline Despite Tech Giant Pressure appeared first on CoinCentral.

• Sutskever takes over as CEO of Safe Superintelligence after Gross departs for Meta.
• Meta attempted to acquire the startup, but the offer was declined.
• The AI industry is seeing a shift as technical founders assume executive roles.
• The battle for top AI talent is redefining leadership and strategy across tech.

OpenAI co-founder Ilya Sutskever has officially taken over as CEO of Safe Superintelligence, the AI startup he co-launched in 2023, marking a pivotal leadership shift in the midst of escalating competition for artificial intelligence talent.

His appointment comes days after former CEO Daniel Gross departed the company on June 29, reportedly joining Meta as part of the tech giant’s sweeping effort to attract top-tier AI specialists.

This leadership reshuffle places Sutskever at the center of a high-stakes contest among tech titans, where talent has become the most valuable resource. With Daniel Levy, another co-founder, assuming the role of president, the startup appears to be consolidating its leadership to defend its mission and independence amid intense poaching attempts.

I sent the following message to our team and investors:
—

As you know, Daniel Gross’s time with us has been winding down, and as of June 29 he is officially no longer a part of SSI. We are grateful for his early contributions to the company and wish him well in his next…

— Ilya Sutskever (@ilyasut) July 3, 2025

Meta’s rejected bid highlights AI safety’s rising value

Meta’s interest in Safe Superintelligence, which reportedly included a failed acquisition attempt, underscores the company’s aggressive pursuit of cutting-edge AI expertise.

Industry sources suggest that Sutskever rebuffed the offer, choosing instead to double down on his vision for developing advanced AI with an unwavering focus on safety.

The company, often referred to by its initials SSI, has increasingly drawn attention for its unique positioning in the broader AI ecosystem. While many startups focus solely on capabilities, SSI is trying to balance advancement with caution. Its reported multibillion-dollar valuation reflects the rising market recognition of AI safety as not just a theoretical concern but a viable commercial frontier.

Tech founders step into CEO roles

Sutskever’s ascension from co-founder to CEO symbolizes a broader trend in the AI industry, where technical leaders are stepping into executive roles to ensure their scientific vision guides company direction.

This pattern echoes similar moves across the sector, as the breakneck pace of development compresses the traditional timelines of leadership transitions.

The talent war that helped trigger this change is not merely a numbers game. Reports indicate that Meta is offering compensation packages worth over $100 million to lure elite engineers, a strategy that has sent shockwaves through rivals like OpenAI and Anthropic. These unprecedented offers are not only challenging employee retention but also reshaping the economics of talent in the AI sector.

Talent now trumps compute in AI race

In this fiercely competitive environment, Safe Superintelligence is attempting to build a team that can deliver safe and powerful AI without compromising its foundational goals. The startup’s refusal to be absorbed into a larger corporate framework reflects a growing belief among some AI leaders that independence is necessary to preserve integrity in research and deployment.

Meta’s strategic shift from data and compute resources to acquiring specialized human capital highlights a fundamental change in how competitive advantage in AI is being defined.

The push to assemble elite research teams is creating ripple effects across the industry. This may lead to a future where talent mobility and cultural cohesion become just as important as raw technological output.

Sutskever’s move marks shift in AI leadership

Sutskever, known for his instrumental role in building OpenAI’s technical roadmap, now faces the dual challenge of leading a startup in a turbulent market while holding firm to a mission centered on safety and long-term AI alignment.

His leadership at SSI will be closely watched as the AI arms race continues to reshape the power dynamics of Silicon Valley and beyond.

The post OpenAI Co-Founder Now Leads Safe Superintelligence Amid AI Talent Wars appeared first on CoinCentral.

• Elon Musk’s xAI has been granted approval to operate 15 gas turbines at its Memphis supercomputer site.
• The decision follows months of protests from residents concerned about nitrogen dioxide emissions and health risks.
• xAI says the turbines are necessary to meet the power demands of its 200,000-GPU facility powering the Grok chatbot.
• The permit comes shortly after xAI raised $10 billion in funding to expand its AI infrastructure amid rising industry competition.

Elon Musk’s AI venture, xAI, has officially secured approval to operate 15 natural gas-burning turbines at its supercomputer facility in Memphis, brushing aside months of heated public opposition and environmental concerns.

The permit, issued by the Shelby County Health Department on July 2, 2025, allows xAI to power one of its largest AI data centers in the U.S., which currently houses around 200,000 GPUs. This facility underpins the Grok chatbot, which is deeply integrated into Musk’s social media platform, X.

xAI welcomes today's decision by the Shelby County Health Department. Our onsite power generation will be equipped with state-of-the-art emissions control technology, making this facility the lowest emitting of its kind in the country. We look forward to being a valued partner to… pic.twitter.com/Vo0wTKHkGy

— xAI Memphis (@xAIMemphis) July 2, 2025

Public Health Debate Intensifies

The decision comes in the wake of intense community protests and a string of public hearings that amplified fears over potential health hazards.

Many Memphis residents had raised red flags about the air pollution expected from the turbines, especially concerning nitrogen dioxide emissions, which have been linked to respiratory illnesses and cardiovascular disease.

According to health experts, high-temperature combustion from natural gas turbines can release concentrated levels of nitrogen dioxide. While national emissions have declined over the years, localized sources like xAI’s site could create pollution hotspots, exacerbating health risks for nearby communities.

AI Infrastructure’s Growing Energy Appetite

Despite the backlash, xAI’s need for dedicated power was deemed critical. The company’s Memphis facility demands massive computing power, far surpassing the capabilities of local grid infrastructure. To maintain uninterrupted AI model training and chatbot performance, xAI opted for on-site power generation via turbines, facilitated through an energy services agreement with Sustainable Energy Infrastructure (SEI).

This self-contained energy model reflects a broader industry trend. As artificial intelligence systems grow more complex, companies like xAI are turning to independent energy solutions to meet their performance targets. It’s a shift that underscores the tension between technological advancement and environmental sustainability.

Billions Fuel Expansion Amid Controversy

The permit approval comes on the heels of xAI securing US$10 billion in fresh capital to expand its infrastructure footprint. Morgan Stanley, which arranged the financing, noted overwhelming investor interest.

Part of these funds will likely be funneled into strengthening the Memphis supercomputing hub and other planned sites as xAI races to compete with established players like Google DeepMind and OpenAI.

Musk’s ambitions for xAI continue to scale at a staggering pace. The company is reportedly in talks to raise an additional US$20 billion in equity, which could push its valuation to as high as US$200 billion. Such rapid growth signals how integral AI has become to the broader tech race, but it also raises questions about the environmental tradeoffs involved in powering next-generation intelligence.

Regulatory Watchdog Sets Conditions

While the permit has been granted, xAI must comply with strict emissions testing and operational guidelines. Any violations could result in fines of up to US$10,000 per day, according to the health department. The scrutiny is expected to continue as local watchdog groups pledge to monitor air quality and hold the company accountable.

In the meantime, residents remain divided. Some view the facility as an economic boost and a signal that Memphis is entering the AI era. Others worry that the cost of progress may be measured in the air they breathe.

The post Elon Musk’s xAI Wins Gas Turbine Permit Despite Pollution Protests in Memphis appeared first on CoinCentral.

• Cadence Design Systems stock rose over 5% after the U.S. lifted export restrictions on chip design software sales to China.
• Synopsys and Siemens also saw gains as the rollback allows them to resume business with Chinese semiconductor clients.
• The three firms dominate the global EDA market, making policy changes highly impactful for revenue and global supply chains.
• The reversal reflects economic pressure overriding national security concerns amid rising demand for AI and advanced chip tech.

Cadence Design Systems saw its stock climb by over 5% in premarket trading after the United States government unexpectedly rolled back export restrictions on advanced chip design software to China.

The announcement came alongside similar rallies for Synopsys, which gained nearly 6%, and Siemens, whose shares edged up in Frankfurt following the news.

The policy reversal marks a significant pivot in Washington’s approach to regulating strategic technology trade with China, particularly in the semiconductor sector.

EDA Tools Back in Play

The U.S. Department of Commerce’s Bureau of Industry and Security confirmed that it had rescinded rules imposed in May that blocked the sale of electronic design automation (EDA) software to Chinese entities.

These tools are foundational to chip design and development, enabling companies to create complex integrated circuits that power everything from consumer electronics to high-performance computing systems.

For companies like Cadence and Synopsys, the shift offers a direct path to reclaiming hundreds of millions of dollars in revenue from the Chinese market. Cadence previously derived approximately $550 million annually, about 12% of its total revenue, from China before the restrictions were introduced. Synopsys was also at risk of losing an estimated $200 to $300 million per year. Siemens, which competes through its Mentor Graphics subsidiary, similarly stands to benefit as it resumes access to a critical customer base.

Market Reaction Swift

The three firms collectively dominate the global EDA market, controlling over 90% of total revenues in the sector. In China alone, their combined market share surpasses 70%, making the export restrictions especially disruptive.

While Washington initially framed the restrictions as a national security measure aimed at curbing Beijing’s access to sensitive semiconductor technologies, the swift reversal, just five weeks after implementation, suggests mounting pressure from industry stakeholders and a reassessment of the broader economic fallout.

The original curbs, enacted in May, were part of a broader U.S. strategy to slow China’s progress in advanced chip manufacturing. However, the unintended consequences were immediate.

Market analysts estimate that U.S. tech firms collectively lost about $130 billion in market capitalization following the initial round of export controls. Beyond stock performance, the bans also risked fracturing long-standing commercial relationships between U.S. firms and Chinese customers, while accelerating China’s efforts to build domestic alternatives.

China Still Behind

While China has poured billions into developing homegrown EDA solutions, experts widely agree that the country remains heavily dependent on foreign software to meet the demands of advanced chip design.

Tools from Cadence, Synopsys, and Siemens are particularly vital for producing chips at 5nm and 3nm process nodes, which are crucial for AI, 5G, and autonomous technologies. As global demand for generative AI and high-efficiency chips grows, so too does the strategic importance of EDA software in the broader tech landscape.

Balance Over Bans

The U.S. government’s decision to lift the restrictions appears to reflect a more balanced posture, weighing national security concerns against the economic realities of global supply chains.

By reinstating access to China’s vast semiconductor market, companies like Cadence and Synopsys can re-engage with key clients and stabilize future growth trajectories. Meanwhile, Siemens’ moderate gains reflect the broader optimism shared across the EDA sector as business with Chinese customers resumes.

Meta Description:

Subtitle:

Tags:
, Synopsys, Siemens, CDNS, chip design, EDA software, US-China tech trade, semiconductor industry, export controls, AI chips

The post Cadence Design Systems (CDNS) Stock: Gains 5% as Washington Lifts Chip Software Curbs appeared first on CoinCentral.

• Gemini’s custom AI helpers, called Gems, are now integrated into Google Docs, Gmail, Sheets, Drive, and Slides.
• Users can build personalized Gems that perform specific tasks without repeating prompts every time.
• Gems can pull data from files, mention colleagues, and function natively within Workspace apps.
• The rollout began on July 2 and will be available to Standard Workspace plan users and above.

Google has officially extended its Gemini AI capabilities to the heart of its Workspace ecosystem, integrating customizable AI assistants known as “Gems” into popular productivity apps including Docs, Gmail, Sheets, Drive, and Slides.

This rollout, which began on July 2, marks a significant shift in how users interact with Google’s suite of tools, offering a more streamlined and intelligent workflow for both individuals and teams.

AI Assistants Built Around Your Workflow

Gems are custom-built versions of Google’s Gemini chatbot, designed to reduce repetitive prompting and deliver personalized results.

Previously only accessible through the Gemini app and website, these AI helpers can now be launched directly from the side panel of Workspace apps. This means users can interact with their AI assistants without switching tabs or breaking focus from their current task.

Whether you’re drafting marketing copy in Docs, organizing sales data in Sheets, or managing client outreach via Gmail, a tailored Gem can now follow you throughout your workflow. For instance, a user could build a Gem that acts as a writing coach with a preferred tone and target audience, or create one trained specifically on company sales decks to help generate pitch emails or executive summaries.

Customized Support for Every Role

The appeal of Gems lies in their flexibility. Each one can be customized with specific instructions, a name, and even uploaded documents that serve as reference points. Once configured, the AI remembers the parameters and responds consistently, saving time and enhancing accuracy. Default options provided by Google include roles like Coding Partner, Career Guide, Brainstormer, and Learning Coach, but users are free to create any kind of assistant to match their needs.

One marketing team, for example, might use a Gem designed to test how executives would respond to a message, while another Gem might support a technical writer by adhering to company style guidelines stored in Drive. This personalization turns Gemini from a general-purpose AI into a deeply contextual assistant capable of adapting to distinct professional roles.

Deeper Integration Within Workspace

Beyond simple chatbot responses, these AI assistants can now make use of Workspace-specific features. This includes the ability to reference documents, tag collaborators with @ mentions, and surface relevant folders or files in real-time. This tight integration helps the AI feel more like a native part of the Workspace environment rather than a bolt-on utility.

However, users still need to visit the Gemini website to create or edit a Gem. Google has not yet brought full Gem editing capabilities into the side panel itself. Despite this limitation, once a Gem is set up, it becomes a powerful shortcut for repetitive or complex tasks inside the productivity suite.

Availability and Rollout Timeline

The update is being made available to users on Standard or higher-tier Google Workspace plans who already have access to Gemini in their side panel.

Google noted that while the rollout began on July 2, it may take up to two weeks to reach all eligible accounts. Once activated, the integration is expected to significantly enhance how teams and individuals get work done across the Google ecosystem.

The post Google Expands Gemini AI to Core Productivity Tools appeared first on CoinCentral.

• RFK Jr. plans to overhaul federal health agencies using AI, sidelining expert input.
• His push to replace animal testing with AI models at the FDA is drawing scientific skepticism.
• Proposed AI automation of the vaccine safety reporting system (VAERS) has raised fears of data misuse.
• Experts warn Kennedy’s policies could fuel vaccine misinformation and undermine public trust in health institutions.

Robert F. Kennedy Jr.’s sweeping plan to embed artificial intelligence into the U.S. health system is drawing fire from scientists, regulators, and public health experts who say the move risks sidelining evidence-based medicine in favor of untested automation.

AI to Replace Expertise in Federal Health Plans

In a recent interview with Tucker Carlson, U.S. Health and Human Services Secretary Robert F. Kennedy Jr. outlined his vision of a future where artificial intelligence becomes central to government health operations.

According to Kennedy, the department is undergoing what he calls an “AI revolution.” Critics, however, fear that revolution may be rooted more in ideology than sound science.

Kennedy urged Americans to “stop trusting the experts” and instead rely on machine-driven models to guide everything from fraud detection to public health interventions. That stance has sparked widespread concern among researchers, many of whom say Kennedy’s understanding of AI and its limits is dangerously shallow.

Animal Testing on the Chopping Block

One of the more dramatic shifts comes at the Food and Drug Administration, where Kennedy is accelerating the transition away from animal testing. Under his directive, the agency has begun using AI-driven models and foreign safety data in the drug approval process. This move builds on the 2022 FDA Modernization Act, which relaxed previous animal testing requirements, but experts argue the current technology cannot fully replace in vivo trials.

The National Association for Biomedical Research recently reiterated that while alternative methods like organoid cultures and computational models hold promise, there is still no complete substitute for animal models in ensuring drug safety and efficacy.

VAERS Overhaul Raises Alarm

Perhaps the most contentious proposal lies in Kennedy’s plan to revamp the CDC’s Vaccine Adverse Event Reporting System, or VAERS. Often misused by anti-vaccine activists, VAERS is a voluntary reporting tool meant to detect rare side effects from vaccines. Kennedy wants to automate it using AI, potentially transforming the system into a self-operating engine for vaccine risk assessments.

Critics worry that automation could lead to flawed data interpretations, particularly if AI tools are designed or trained without rigorous scientific oversight. There are already concerns about Kennedy’s objectivity, given his long history of opposing vaccines and promoting unsubstantiated medical claims.

Bias and Mistrust Could Undermine Public Health

The heart of the backlash isn’t just about AI. It’s about trust. Experts say that any data-driven system is only as reliable as the inputs and intentions behind it. If algorithms are built around Kennedy’s own skeptical views on vaccination, the resulting data could reinforce public fear, not alleviate it.

Already, Kennedy’s “Make America Healthy Again Commission” has come under fire for releasing a report riddled with false citations, allegedly produced using AI tools. The fear now is that similar unvetted technology could soon be responsible for making critical health policy decisions.

Dr. Paul Offit, a prominent vaccine expert, warns that Kennedy’s leadership could have dire consequences.

“He is an anti-vaccine activist and a conspiracy theorist,” Offit said. “He will do everything he can to make vaccines less accessible, less affordable, and more feared.”

As Kennedy barrels ahead with plans to embed AI into America’s health infrastructure, the debate continues over whether technology can truly replace scientific consensus or if it’s simply being weaponized to validate fringe beliefs.

The post RFK Jr. Pushes AI Over Experts in Controversial Health Policy Shift appeared first on CoinCentral.

• Meta is testing AI chatbots that can message users first on Messenger, WhatsApp, and Instagram.
• The bots remember previous chats and offer follow-up messages if users engage frequently.
• Safety concerns are rising as the bots mimic personal conversations but aren’t trained professionals.
• Meta may eventually monetize these bots through ads, subscriptions, or integration with other platforms.

Meta is rolling out a new approach to keep users glued to its platforms  by introducing AI chatbots that can reach out to people first.

The tech giant is currently testing proactive messaging features across Facebook Messenger, WhatsApp, and Instagram, allowing AI personas to initiate conversations and even recall details from earlier chats.

Bots That Start the Conversation

One such AI persona, dubbed “The Maestro of Movie Magic,” might send users an unexpected message like: “I hope you’re having a harmonious day! Have you discovered any new favorite soundtracks lately? I’d love to offer some movie night suggestions.”

This sample message, revealed in internal guidelines viewed by Business Insider, highlights how Meta’s AI companions are designed to be engaging and highly personal.

The proactive bots are being developed in partnership with data labeling firm Alignerr. Leaked documents show that Meta is training these bots to remember user interactions and follow up accordingly. Users can create their own AI personas through Meta’s AI Studio, deciding whether to keep them private or share them via stories, links, or profile pages.

Rules for Reaching Out

Meta confirmed it was testing these follow-up features with limitations in place. Bots can only message users within 14 days of an initial conversation and only if the user has sent at least five messages during that window. If there’s no response to the first follow-up, the bot won’t send more messages.

According to a Meta spokesperson, the aim is to let people “continue exploring topics of interest and engage in more meaningful conversations with the AIs across our apps.”

This strategy mirrors that of other AI-based platforms like Replika and Character.AI, which also offer companion-style chatbots capable of initiating dialogue.

Safety Concerns Resurface

Still, the move isn’t without controversy. Character.AI is currently facing a lawsuit after one of its bots was allegedly involved in the death of a 14-year-old boy. When asked about safety protocols, Meta pointed to several disclaimers on its platforms warning users not to rely on AI responses for serious matters. These notices clarify that chatbots are not certified professionals and should not replace expert advice in areas like health, law, or finance.

So far, Meta has not publicly set age limits for engaging with its AI bots, although some U.S. jurisdictions such as Tennessee and Puerto Rico restrict teen interactions with AI.

Monetization May Follow

At a glance, the new chatbot initiative appears aligned with CEO Mark Zuckerberg’s public commitment to tackling loneliness and promoting connection. However, the business implications are significant. In recently unsealed court documents, Meta estimated its generative AI products could drive up to $3 billion in revenue in 2025, ballooning to $1.4 trillion by 2035.

A large portion of that could come from monetizing interactions with chatbots through ads, subscriptions, or licensing its Llama AI models to partners.

When pressed on whether these proactive bots would eventually show ads or become part of Meta’s broader virtual world strategy via Horizon, the company declined to comment. That said, as the AI arms race heats up, Meta’s chatbots are poised to blur the line between helpful assistants and persistent digital companions. Whether users will embrace the shift remains to be seen.

The post Meta Tests Proactive AI Bots That Remember and Message You appeared first on CoinCentral.

• Foxconn’s Ingrasys leases a new facility in Taoyuan, boosting AI server output in Taiwan.
• The company expects AI server revenue to rise over 50 percent in 2025 to more than NT$1 trillion.
• Taiwan’s role in global AI infrastructure is expanding amid rising U.S. exports and local investment.
• Foxconn continues diversifying production globally with recent investments in Texas and California.

Foxconn’s strategic push into artificial intelligence hardware continues to accelerate as its subsidiary, Ingrasys Technology, expands its production footprint with a new facility in Taoyuan, Taiwan.

The move, part of a broader realignment of manufacturing priorities, signals the company’s growing focus on AI servers amid a global tech shift.

The newly leased site, located in the Farglory Free Trade Zone in Taoyuan’s Dayuan District, involves a NT$657 million (US$22.8 million) commitment over a 60-month lease. Ingrasys will occupy eight floors within the factory complex, reinforcing its AI server production lines already operating in Luzhu, Taoyuan, and Hukou in Hsinchu.

This latest expansion underlines Foxconn’s effort to recalibrate its business around high-performance computing, distancing itself from an overreliance on smartphone assembly. For years, Apple has contributed over half of Foxconn’s total revenue. But with smartphone demand flattening globally, the company is pivoting toward AI infrastructure, a sector it now views as critical to future growth.

Taiwan at the Core of AI Hardware Growth

The new investment places Taiwan squarely at the center of the global AI manufacturing boom. The country has experienced a surge in electronics exports, particularly those tied to data center and AI hardware. In 2024, Taiwan’s exports to the U.S. jumped by nearly 46 percent to over US$111 billion, largely driven by AI-related equipment.

In this environment, Ingrasys’ expanded presence in the free trade zone plays a dual role: supporting domestic industrial growth while feeding demand from major global clients, including U.S.-based hyperscalers.

This aligns with the broader trend of Taiwan becoming a crucial node in the AI supply chain, not only for chips but also for the servers that run large-scale AI models.

Foxconn’s AI Revenue Set to Soar

Foxconn expects its AI server revenue to grow by over 50 percent in 2025, potentially surpassing NT$1 trillion (around US$34.7 billion). That projection follows a significant uptick in broader sales performance.

In May alone, Foxconn reported revenues of NT$615.75 billion (US$20.6 billion), an 11.92 percent increase from the same period in 2024. Strong demand in cloud, networking, and computing sectors has helped the company build momentum through the first half of the year.

The AI pivot is not just about capitalizing on demand. It is also a necessary course correction. The company’s return on invested capital has dropped from 23 percent in 2016 to just 11 percent in 2023, pointing to declining profitability in its legacy assembly operations. Diversifying into AI server manufacturing offers a more scalable, future-focused path forward.

Global Footprint Supports Resilient Supply Chain

The Taoyuan expansion is just one piece of a larger global puzzle. In recent months, Foxconn has also committed US$142 million to building a facility in Houston, Texas, and has acquired another in Sunnyvale, California. Manufacturing hubs in Mexico and Vietnam round out its international footprint, part of a strategy to buffer against trade risks and maintain supply chain agility.

This distributed production model is increasingly vital as global trade policies evolve. U.S. tariffs on imported server products and rising geopolitical friction have pushed tech manufacturers to spread operations across multiple jurisdictions. Foxconn’s multi-site approach allows it to respond flexibly to shifting conditions while staying close to key markets.

The post Foxconn Unit Boosts AI Server Output with New Taiwan Facility appeared first on CoinCentral.

• Google replaces its traditional homepage Doodle with a promotion for AI Mode, signaling a major strategic shift.
• AI Mode offers AI-generated responses to complex queries using the Gemini model, now accessible via text, voice, and images.
• The move comes as Google faces pressure from ChatGPT, which recently hit 1 billion daily searches.
• Publishers are already seeing reduced traffic as AI changes how users find and consume information online.

Google has transformed its iconic homepage into a promotional platform for its new AI-powered search tool, “AI Mode.”

The move comes as part of a larger effort to defend its turf amid mounting pressure from AI competitors such as OpenAI and Anthropic.

Replacing the usual tributes to historic figures or global events, Google’s latest Doodle animates users toward “AI Mode,” a feature powered by its Gemini AI model. The tool, which officially launched in March 2025, is designed to handle complex queries with intelligent, context-aware responses. Unlike the standard keyword-based search, AI Mode accepts text, voice, or even image inputs, aiming to offer users a more seamless and intuitive search experience.

Homepage Becomes Google’s Prime Ad Space

By using its homepage, the most visited page on the internet, as a direct marketing channel for AI Mode, Google is signaling just how critical the feature has become in the race to define the future of online search. It marks a strategic pivot in Google’s approach, revealing that AI-powered search is no longer a side experiment but a central part of its roadmap.

Since May 2025, Google has gradually introduced AI Mode across its interface, testing it in places such as the “I’m Feeling Lucky” button. But placing it front and center on its homepage sends a powerful message to users and rivals alike: the future of search, as Google sees it, is increasingly conversational, multimodal, and AI-driven.

Rising Competition Forces Google’s Hand

This promotional shift reflects the growing urgency at Google to maintain its dominance in a market now threatened by rapid advances from other players. OpenAI’s ChatGPT, for example, recently hit 1 billion daily searches, outpacing Google’s early growth by over fivefold.

Its ad-free, conversational format has resonated particularly well with users across India, the U.S., and Southeast Asia.

OpenAI’s rise has been accompanied by aggressive expansion strategies, including hardware partnerships and real-time retrieval tools that make ChatGPT an increasingly formidable search alternative. Google’s homepage spotlight on AI Mode is widely viewed as a countermeasure, intended both to retain its vast user base and accelerate the transition toward AI-centric search behavior before rivals establish default positions in browsers or mobile ecosystems.

AI Mode’s Disruption Extends Beyond Search Habits

The rise of AI Mode isn’t just a shift in interface. It’s transforming the broader search economy. Publishers and content creators have begun reporting up to a 30 percent drop in click-through rates, as users increasingly get answers directly from AI responses without needing to visit traditional websites.

This paradigm shift is prompting website operators to rethink how they structure and deliver content. Google, for its part, has started offering AI Mode metrics in Search Console, recognizing that success in this new era will be measured differently than in the past. Structured data, authoritative sourcing, and alignment with AI-generated summaries are becoming the new pillars of search visibility.

Google Signals New Era in Search Wars

As AI Mode gains traction, already reaching 1.5 billion monthly users ,Google is leveraging every inch of its infrastructure to ensure its dominance doesn’t erode.

The homepage push is more than just a flashy promotion. It’s a calculated move in a high-stakes war for the future of digital discovery, where the winner will not just answer questions, but shape how humanity interacts with knowledge itself.

The post Google Turns Homepage Into Billboard for ‘AI Mode’ in Battle for Search Supremacy appeared first on CoinCentral.