XTHE

Intel has introduced its new Xeon 600 processor family for workstations built for demanding tasks such as AI development, photo-realistic visualization, complex CAD modeling, and advanced 3D animation.  

From a platform perspective, this is the AI developer’s dream, said Jonathan Patton, Intel Client Product Marketing. This platform meets both CPU-based inference and multi-GPU connectivity needs.  

The Xeon 600 family has 11 models, offering 12 to 86 cores. Each one supports Intel AVX512 and AMX extensions, which now include Intel Integrated 8 B Float 16 and FP16, with up to 4 TB of memory. PCIe 5.0, Lens, and CXL 2.0 support. This is a major upgrade for AI developers.  

RAM bandwidth is higher thanks to support for multiplexed rank DIMMs (MR DIMMs), which put two ranks of RAM on a single module. This helps with memory-bound workloads with data rates reaching up to 8,000 MT/sec.  

The Xeon 600 processors use Intel’s Granite Rapids technology, which was previously only in servers. They are built on the Intel 3 process, not the newer and more efficient Intel 1.8 process. They will be available as standalone chips and systems starting in late March.  

Pricing and Specs for the Intel Xeon 600 Processors for the Workstation Family.  

There are eleven SKUs at launch. The top six marked with an X suffix are unlocked. Intel product marketing engineer Zachary Kranich explained during a briefing that this gives users more options.  

We found a lot of value in this from the Xeon space, primarily targeting system integrators, giving them the option to tune and tweak and get the best performance they need for their workload, use conditions, and all that good stuff.  

He added it gives the users the ability to say, “Hey, I care more about performance than I care about necessarily stability or longevity.” He said, “Allow me to crank it up as much as possible to get the most out of this, as long as I can.”  

Others will prefer stability, especially those who need this system to be rock-solid for five years.  

Intel is working with Ocbase to add support for the Intel Xeon 600 processor family in the OCCT app. They’ve added a ton of capabilities to our application to enable dynamic overclocking controls, said Kranich.  

Windows support, obviously, but Linux support will be included as well, which is awesome. You get all the platform, telemetry, stability testing, and configured benchmarking all integrated within that tool. Intel hopes support will be ready at launch, but there could be a slight delay. Only 5 SKUs will be boxed for retail sales, complete with package carriers to attach to heat sinks.  

• 18-core Xeon 654.  

• 24-core Xeon 658X  

• 32-core Intel Xeon 676X  

• 48-core Intel Xeon 678X  

• 64-core Xeon 696X  

How Much Faster Are the New Intel Xeon 600 Processors  

According to Intel’s Cinebench 2026 results, you can expect up to 9% better single-core performance and up to 61% better multi-threaded performance. As Intel says, these are huge leaps in workstation performance. There’s even a slide that says so.  

There are two things to keep in mind about this slide:   

1. It compares the 86-core Intel Xeon 698X to the 60-core Xeon W93595X.  

2. It does not state how the new chips compare to AMD’s Thread ripper family.  

With that caveat in mind, there are still positive signs in the details on the next four slides, which we will include as a gallery.  

Kranich also mentioned that Intel has worked with Asus to overclock the Xeon 698X processor, and together they expect to set 10 world records at launch.  

His results show what’s possible when processor architecture and motherboard engineering are co-optimized. From day one, Jonson Lee, head of research and development at ASUS, is quoted as saying on a slide.  

Our R&D team worked closely with Intel to fully unleash the overclocking headroom of Xeon 698X, establishing new worldwide benchmarks in extreme performance.  

Is the Xeon 600 Processor for Workstations the Right Choice for You  

You might be wondering how the Xeon 600 family fits with the rest of Intel’s chips or whether there are hints about future mobile workstations. Intel answers this with a helpful slide. The message is that there is more to come this year for slimline, affordable mobile workstations. Still, if you want power, then you are currently limited. Intel wouldn’t use that word to describe Core Ultra 200HK processors or AMD.  

Before buying a high-end desktop PC, Intel reminds you to use the right tool for the right job.  

One slide compares the Core Ultra 5245K, with 6 P-cores and 8 E-cores, to the entry-level Xeon 6C with 12 P-cores. The 5245K wins in a single-core performance, but the Xeon is better for most professional workloads, except for CAD.  

In other words, choose carefully.  

Intel has named Dell, HP, Lenovo, Supermicro, Boxx, and Puget Systems as hardware partners for the late March 2026 launch. Watch their sites for availability.

Source: Intel announces “AI developer’s dream” Xeon 600 Processors for Workstation 

Apple is developing a more affordable MacBook that will use an A-series iPhone chip instead of the usual M-series Apple silicon. Here is a summary of the latest rumors about this new laptop, which could arrive next year.  

Design  

Apple analyst Ming-Chi Kuo says the new low-cost MacBook will have a 13-inch display, making it about the same size as the current 13-inch MacBook Air.  

Apple once offered a very thin 12-inch MacBook, and there have been occasional rumors about its return. A lightweight MacBook powered by an iPhone chip seems possible.  

The earliest 12-inch MacBook, released in 2015, used a low-power Core M chip before Apple switched to its own silicon. It had no fans, which made it thin and quiet, a design Apple continued with the MacBook Air.  

A-series chip would generate less heat than an M-series chip, so it would need fewer cooling features. While thinner and lighter Apple devices are usually pricier, using an iPhone chip in a MacBook Air-sized body could keep costs down.  

Colors  

Apple usually sells classic colors for its Pro models and uses brighter colors for more affordable products. Rumors say the low-cost MacBook might come in silver, blue, pink, and yellow, similar to the iMac’s color options.  

Bloomberg reports that Apple has tested colors like light yellow, light green, blue, pink, silver, and dark gray. Not all will be released, but at least four should be available.  

Chip  

Power reports that the new MacBook will use an A18 Pro chip. There are also hints of a MacBook with this chip in Apple’s code, so the A18 seems likely.  

The A18 Pro chip was first introduced in the iPhone 16 Pro. It uses a second-generation 3nm process and offers strong performance for an iPhone chip.  

The A18 Pro features a 6-core CPU for performance and 2 efficiency cores, a 6-core GPU, and a 16-core neural engine for AI tasks. In Geekbench tests, it scores 3451 for single-core and 85724 for multi-core. By comparison, the M4 chip in the iPad Pro scores 3694 (single core) and 13732 (multi-core).  

The A18 Pro is faster than the M1 chip, which Apple has used in lower-cost MacBook Air models for years, in single-core performance. An A18 MacBook would be close to the M4 chips in Macs and iPads, though multi-core performance would still differ.  

A MacBook with the Intel-18A chip would easily handle everyday tasks like browsing the web, creating documents, watching videos, and light photo or video editing. It won’t be suited for demanding games or heavy tasks like 4K video editing and 3D rendering, but it can do nearly everything an iPhone or iPad can.  

Thermal Design  

The iPhone 16 Pro uses a thermal design with a titanium frame and a graphite-coated aluminum substructure. Some of these design elements could be used in the new MacBook.  

RAM  

Most Macs start with 16GB of RAM, while the iPhone 16 Pro has 8GB, which is the minimum needed for Apple Intelligence. The A18 Pro MacBook will likely have at least 8GB of RAM to support these features, but Apple might include 16GB, as in other Macs.  

Ports  

The A18 Pro chip in the iPhone 16 Pro does not support Thunderbolt, so the MacBook will only have USB-C 10GB/s and won’t reach Thunderbolt speeds. This means it will likely support just one external display.  

Name  

Apple hasn’t announced a name for the new budget notebook, but MacBook is the likely choice. The company has used this name before, though currently only the MacBook Air and MacBook Pro are available.  

Another possible name is the MacBook SE, which would match the iPhone SE and Apple Watch SE. However, since Apple calls its most affordable iPad simply “iPad,” MacBook remains the most likely name.  

Price  

The MacBook Air with an M4 chip starts at $999 and includes:   

• 10-core CPU  

• 8-core GPU  

• 16GB RAM  

• 256GB SSD  

The A18 Pro MacBook might be a few hundred dollars cheaper, but pricing hasn’t been finalized yet. Bloomberg reports it will cost well under $1000, though details are still unclear.  

Apple likely won’t price the new MacBook below its iPad lineup. The entry-level iPad with an A16 chip starts at $349, and the iPad Air with an M.2 chip starts at $599. Pricing the MacBook between $599 and $699 would keep it less expensive than the MacBook Air or iPad Pro, but just above the iPad Air.  

$599 would match the price of many popular Chromebooks used in schools. It’s also the price of the iPhone 16e Apple’s most affordable iPhone, which uses a slightly more powerful A18 chip.  

Launch Date  

Rumors say the new low-cost MacBook could be announced at or just before Apple’s March for Special Experience event in New York, London, and Shanghai. Apple might even reveal it in a press release before the event.  

Some media members have been invited to the event, where they’ll likely get a chance to try out the new MacBook and other upcoming devices.

Source: Apple’s Low-Cost MacBook: Everything We Know So Far 

Key Points  

• Meta is testing its first AI training chip as part of its plan to rely less on suppliers like NVIDIA.  

• Sources say the chip is designed to help lower the costs of AI infrastructure.  

• Meta plans to use these chips for recommendation platforms and Generative AI.  

Meta, the owner of Facebook, is testing its first in-house chip for training artificial intelligence systems, two sources told Reuters. This is a key step as Meta works to design more of its own custom chips and depend less on outside suppliers like Nvidia.  

The company has started a small rollout of the chip and plans to increase production for wider use if the test is successful, according to sources.  

Meta’s effort to develop its own chips is part of a long-term plan to lower its large infrastructure costs as it invests heavily in AI tools to spur growth.  

Meta, which also owns Instagram and WhatsApp, expects its total expenses in 2025 to be between $114B and $119B. Up to $65B of this will go toward capital spending, mostly for AI infrastructure.  

One source said Meta’s new training chip is a dedicated accelerator built to handle only AI-specific tasks. This design can make it more power-efficient than the GPUs usually used for AI workloads.  

Meta is working with Taiwan-based chip manufacturer TSMC to produce the chip, according to the source.  

The test deployment began after Meta completed its first tape-out of the chip, meaning sending the initial design through a chip factory, the other source said. This is an important milestone in chip development.  

A typical tape-out costs tens of millions of dollars and takes about 3 to 6 months to complete, with no guarantee of success. If it fails, Meta would need to find the problem and repeat the process.  

Meta and TSMC declined to comment.  

This chip is the newest in Meta’s Meta Training and Inference Accelerator (MTIA) series. The program has struggled for years and once canceled a chip at a similar stage,  

Here, Meta began using an MTIA chip for inference, which is the process of running an AI system as users interact with it. The chip helps power the recommendation platforms that decide what content appears on Facebook and Instagram news feeds.  

Meta executives have said that they want to start using their own chips for training by 2026. Training is the process of feeding large amounts of data to an AI system to teach it how to perform tasks.  

Like the Inference Chip, the Training Chip will first be used for advisory systems. Later, Meta plans to use it for Generative AI products, such as the Meta AI Chatbot, according to executives.  

We are working on how to train the recommended systems, and eventually how to think about training and inference, according to GenAI’s Meta’s Chief Product Officer, Chris Cox, at the Modern Stanley Technology Media and Telecom conference last week.  

Cox described Meta’s chip development as a “walk/crawl/run” situation so far, but said executives see the first-generation inference chip for recommendations as a “big success.”  

Meta previously canceled an in-house custom inference chip after it failed a small-scale test similar to the current training chip test. The company then switched course and ordered billions of dollars’ worth of Nvidia GPUs in 2022.  

Since then, Meta has stayed with one of Nvidia’s biggest customers, connecting many GPUs to train its models. These include Recommendation and AD systems, as well as the Llama Foundation model series. The GPUs also handle inference for more than 3 billion people who use Meta’s apps daily.  

This year, some AI researchers have questioned the value of these GPUs. They doubt how much more progress can be made by simply adding more data and computing power to large language models.  

These doubts grew after Chinese start-up DeepSeek launched new low-cost models in late January. DeepSeek’s models are more efficient because they rely more on inference than most present models.  

After DeepSeek’s launch, AI stocks worldwide dropped, and Nvidia shares lost about a fifth of their value. The shares later recovered most of that loss as investors still believe Nvidia’s chips will remain in the industry standard for training and inference. However, they have fallen again amid broader trade concerns.

Source: Exclusive: Meta begins testing its first in-house AI training chip

In July 2025, OpenAI tested a new general-purpose AI model. It solved five of the six problems from a 2025-style International Mathematical Olympiad test, earning a gold medal-level score of 35 out of 42.  

The model wrote natural-language proofs without using external tools, demonstrating reasoning similar to that of humans. Although the exact five problems are from a private 2025 test, their success shows strong ability in areas that are usually hard for AI.  

• Geometry: Like Alpha Geometry, the model could handle complex, multi-step geometric proofs.  

• Combinatorics: The model solved problems with distinct structures and arrangements typical of IMO-level challenges.  

• Number theory: The model managed problems that required complex algebraic work and an understanding of integer properties.  

• Arbitrary Inequalities: The model demonstrated creative multi-step methods for proving inequalities.  

• Advanced Algebraic Equations: The module produced detailed step-by-step proofs for complex functional problems.  

Main breakthroughs from this achievement include:  

• Human-level reasoning: Instead of relying on specialized tools, the model used natural language and step-by-step thinking to solve problems, not just pattern matching.  

• Gold Medal Performance: A score of 35 out of 42 indicates skill equal to that of top human math prodigies.  

• General purpose: Unlike models focused only on geometry, this one showed flexible reasoning across many areas of mathematics.  

Note: OpenAI reported this achievement, but the 2025 evaluation was done quickly, and experts are still reviewing some results.  

We tested our internal model on all 10 first-proof problems, advanced math challenges designed to see whether AI can create correct, checkable proofs. These problems differ from short-answer or computation problems because they require building full arguments in specialized areas, and only experts can reliably judge whether the solutions are correct. Top experts wrote the first proof problems, and some remained unsolved for years before the authors found answers. A university department with expertise in these areas can solve many problems within a week.  

We shared our proof attempts on Saturday, February 14th, 2026, at midnight Pacific Time. After receiving expert feedback, five of the models’ proofs (problems 4, 5, 6, 9, and 10) are likely correct, while the others are still under review. At first, we thought our solution for problem 2 was probably right, but after reading the first official proof commentary and more community analysis, we now think it is incorrect. We appreciate everyone’s responses and look forward to more reviews.  

You can find all our proof attempts here. The pre-print includes all ten proofs and a new appendix with prompt patterns and examples that show how we interacted with the models during the process.  

We believe novel frontier research is the most important way to evaluate the capabilities of next-generation AI models. Benchmarks are useful, but they can miss some of the hardest parts of research:   

• sustaining long chains of reasoning  

• choosing the right interactions  

• handling ambiguity in problem statements  

• producing arguments that survive expert scrutiny  

Frontier challenges like first proofs help us stress-test and probe those capabilities in settings where correctness is hard to verify, and failure models are informative.  

We are currently training a new model, with a main focus on increasing its level of strictness, so that it can think continuously for many hours and remain highly confident in its conclusions. When the first proof problems were announced, it seemed like a perfect test bed, so we tried it over the weekend. It has already solved two of the problems (numbers 9 and 10). As it trained, it became increasingly capable, eventually solving, in our estimation, at least three more. We were particularly pleased when it solved number 6, and then, two days later, number 4, as those problems were from fields similar to those of many of us. It’s incredible to watch a model get tangibly smarter day by day.  

James R. Lee (OpenAI researcher reasoning)  

We used the model with minimal human supervision. During training, we sometimes suggested trying strategies that had worked before. For some proofs, we asked the model to add more details or to elucidate its reasoning after receiving expert feedback to make them easier to check. We also set up a back-and-forth between the model and ChatGPT to help with checking, formatting, and style. For some problems, we selected the best attempt from several based on expert evaluation. This was a quick process and not as organized as we would want for a fully controlled test. 

We look forward to working with the first proof organizers on a more rigorous experiment and evaluation process going forward.  

This work builds on earlier results from frontier reasoning models in math and science. In July 2025, we achieved gold-medal-level performance on the International Mathematical Olympiad with a general-purpose reasoning model, scoring 35/42 points. In November, we shared early experiments in accelerating science with GPT-5, a set of case studies in which GPT-5 helped researchers make concrete progress across math, physics, biology, and other fields, along with the limitations we observed. Most recently, we reported a physics collaboration in which GPT-5.2 proposed a candidate expression for a gluon amplitude formula, which was then formally approved by an internal model and verified by the authors.  

We look forward to working more closely with the community to evaluate research-level reasoning, including seeking expert feedback on these attempts. We are also excited to bring these new capabilities to future public models.

Source: Our First Proof submissions 

Key points  

• Bath & Body Works has opened an official storefront on Amazon as part of its ongoing push to sell products beyond its own stores.  

• CEO Daniel Heaf told CNBC that the launch is focused on reaching customers where they already shop.  

• For brands such as Bath & Body Works, Gap, and Everlane, Amazon is becoming more of a logistics partner than a traditional retailer.  

Amazon Prime members can now easily buy Bath & Body Works Champagne Toast Body Wash with no minimum shipping requirement.  

The popular brand is offering some of its top fragrances, body washes, hand soaps, and candles to US Amazon shoppers. These products are also available with Prime shipping, according to Euromonitor. Amazon is the top online destination for US beauty shoppers, holding 47% of the online beauty and personal care market in 2024. Sephora comes in second with a 9% share. Euromonitor also estimates that 39% of all beauty and personal care sales happen online.  

Launching our first authorized brand storefront on Amazon puts us directly in consumers’ paths. Bath & Body Works CEO Daniel Harish told CNBC it’s about meeting them where they already are. The Amazon launch is the latest move by Bath & Body Works, based in Columbus, Ohio, to reach more customers. Last year, the company began selling its products in college campus stores, which now number over 1,000. These were the first sales points outside its approximately owned and franchised stores and its website.  

Heaf joined Bath & Body Works in May after his position as Nike’s Chief Transformation and Strategy Officer was eliminated by CEO Elliot Hill.  

Harish recently shared his plan to return Bath & Body Works to profitable long-term growth. He describes it as a consumer-first formula built on 4 pillars:   

1. Creating innovative products  

2. Reigniting the brand  

3. Succeeding in the marketplace  

4. Working quickly and efficiently  

The Amazon Partnership, he said, is the first of many milestones to be delivered this fiscal year under that strategy.  

Before launching its official storefront, Bath & Body Works products were available on Amazon through third-party sellers.  

Now, Harish says the company is working to take back control of its brand story and sales on Amazon.  

Amazon: Friend or Foe?  

While Amazon has many first-party relationships with brands from Nike to Calvin Klein that use wholesale partnerships as part of their business models, there are few examples of retailers selling on the site that design, manufacture, and sell their products entirely in-house.  

For vertically integrated brands like Bath & Bodyworks, Amazon is taking on a more logistics-partner role rather than acting as a traditional retailer.  

Gap, J. Crew, and Eve. srlane are also vertically integrated and offer limited selections of their branded products on Amazon.  

GAP began sharing what it calls core basics for the whole family in 2022 through a wholesale partnership in which Amazon owns and sells the products, which are Prime-eligible. GAP has said its goal is to improve customer retention and reach new and lapsed customers, as well as to provide existing shoppers with the convenience of core essentials.  

Under the new agreement, Bath & Body Works will retain ownership of its inventory and set its pricing while using Amazon’s fulfillment network to qualify for Prime shipping.  

Everlane declined to comment on its partnership with Amazon. J.Crew did not respond to a request for comment.  

Jewelry company Kendra Scott has authorized a storefront on Amazon after initially opposing the partnership, even though it had wholesale relationships with other retailers, including Macy’s and Nordstrom. But over time, the brand began to view Amazon as another opportunity to reach shoppers rather than a competitive threat, according to a person familiar with the company’s decision-making, who spoke on the condition of anonymity about private matters.  

Bath & Body Works has also made it easier to order from its own website. Last month, the company lowered its free shipping minimum from $100 to $50.  

Still, he admits we know that we will never compete with Amazon in terms of their Prime Network. No one will offer next-day shipping. That’s just not what we are in the business of. And so, by going on Amazon, we are also making our own site more competitive but recognizing that our job is not to build a fulfillment network that can operate at the speed of Amazon.

Source: Retail Bath & Body Works starts selling on Amazon as more brands embrace its logistics network 

Key details  

• Samsung unveils Galaxy AI as an open, integrated AI ecosystem.  

• Galaxy AI will support multiple AI agents to handle tasks flawlessly.  

• Samsung Apps to integrate Perplexity AI for improved workflows.  

Samsung Electronics has announced the next step in its Galaxy AI strategy, presenting it as an open and integrated ecosystem with multiple AI agents to make daily tasks easier. According to the company, Galaxy AI is designed to work at the operating system level, not just within individual apps, to help reduce effort in everyday routines. The goal is to make AI feel more natural, relevant, and smooth.  

A Shift to Multiple AI Agents  

Samsung points to recent research showing that almost 80% of users use more than two types of AI agents depending on what they need to do, as AI tools become a bigger part of daily life. People are using different assistants for things such as:   

• Searching  

• Getting work done  

• Creating reminders  

• Creating content  

To address this, Samsung is updating Galaxy AI to let users choose from several integrated Agents. Rather than limiting people to a single AI, the system will allow multiple Agents to work cooperatively within the Galaxy platform.  

Galaxy AI acts as an orchestrator bringing together different forms of AI into a single, natural, integrated experience, said Juan Jun Choi, President, COO, and Head of the R&D Office Mobile experience (MX) Business at Samsung Electronics.  

Samsung highlights that Galaxy AI works at the system level, not just within apps. Instead of making users switch between apps or repeat commands, the system understands what is needed and runs in the background.  

This setup lets Samsung add partner services while maintaining a consistent user experience. For example, a Perplexity AI will be added as another AI agent on the new flagship Galaxy devices.  

Users can turn on Perplexity by saying the wake phrase, “Hey Plex,” or by using quick access controls, such as pressing and holding the side button.  

Perplexity’s AI agent will be built into several Samsung apps, including Samsung Notes, Clock, Gallery, Reminder, and Calendar, as well as third-party apps.  

Samsung says this integration will make multi-step tasks easier. For example, users can collect information, save notes, set reminders, and organize schedules without switching between apps.  

By managing everything at the system level, Galaxy AI aims to deliver a smoother, more flexible experience across users’ devices.  

For users, this change shows that Samsung is following a broader industry trend toward Open AI ecosystems rather than relying on a single assistant. By making Galaxy AI a coordinator for multiple agents, Samsung is betting that people prefer options and flexibility over being limited to a single choice. Details about supported devices and timelines will be announced at a later date, but the future is unclear. Samsung wants Galaxy AI to serve as the connective layer that ties together different forms of intelligence into a single integrated experience.

Source: Samsung brings Perplexity to flagship Galaxy devices in AI ecosystem push

Apple is rolling out a major update to Siri called Siri 2.0, also known as the Siri Agent. This new version uses Apple’s intelligence to perform tasks automatically across your device and understand what’s happening on your screen. It can handle several steps at once in different apps.  

Here’s how you can try out the first version of this technology. The more advanced features will be available in early 2026.  

How To Access Siri Agent Beta  

Right now, you can try some Apple Intelligence features in iOS 18.1, 18.2, and later developer betas. The new Siri 2.0 agent features are planned for early 2026, starting with iOS 26.4.  

1. Verify Device Compatibility  

• iPhone: iPhone 16 series, iPhone 15 Pro, iPhone 15 Pro Max  

• iPad/Mac: models with M1 chip or later.  

2. Join The Beta Program  

• Go to beta.apple.com and sign up using your Apple ID.  

• On your device, navigate to Settings > General > Software Update > Beta Updates and select iOS 18 or later Public Beta.  

Enable Apple Intelligence  

• Go to Settings > Apple Intelligence & Siri  

• Tap “Join the waitlist.”  

• Make sure your device and Siri language are set to English (United States), which is required for early access.  

• Once authorized, turn on Apple Intelligence.  

How to use the new Siri Agent features?  

The new Siri agent is built for real, system-wide automation so that it can take actions for you across different apps.  

• On-screen awareness: If a friend sends you a new address, tell Siri to add this address to their contact card, and Siri will do it for you so you don’t have to copy and paste.  

• Cross-app functionality: You can say, “Make this photo pop” and then “Send this to mom in an email.” Siri will move between Photos and Mail apps to get it done.  

• Contextual comprehension: Siri now remembers what you are talking about, so you can ask follow-up questions without restating yourself.  

• App intents: With the new system, Siri can handle specific tasks across different apps, such as finding a document in a third-party app.  

Important Details & Privacy  

• Waitlist: After you turn it on, you may need to wait a short time while the necessary files are downloaded.  

• Privacy: Siri uses on-device processing and private cloud computing to help keep your personal data safe.  

• ChatGPT integration (if you allow it) – Siri can use ChatGPT to answer more specialized questions.  

The advanced Siri 2.0 agent could be available for developers in beta around February 23, 2026, and is expected to be released to the public in March or April 2026.  

Apple plans to update Siri later this year, turning the digital assistant into the company’s first artificial intelligence chatbot, Bloomberg News reported.  

The chatbot, codenamed Campos, will be deeply embedded in the iPhone, iPad, and Mac operating systems and will replace the current Siri interface, the report said, citing people familiar with the plan.  

Updating Siri is an important part of Apple’s plan to keep up with other major tech companies in the AI field. Especially after Apple Intelligence’s 2024 launch, it received a mixed response.  

Earlier this month, Apple struck a deal with Google to use Google’s Gemini chip to power Siri. This is a big win for Google’s parent company, Alphabet, which also makes its own smartphones.  

Campos will use an advanced version of Google’s custom model, similar to Gemini 3, which Apple calls Core Models for Version 11, according to the report.  

The chatbot features will be available later this year. Campos, which will support both voice and typing, will be the main new feature in Apple’s upcoming operating system, the report said.  

In a separate report on Wednesday, the information said Apple is working on an AI-powered wearable pin with several cameras, a speaker, microphones, and wireless charging. The device could launch as soon as 2027, according to the report.  

Apple did not immediately respond to a Reuters request for comment on both reports.

Source: Apple to revamp Siri as a built-in chatbot, Bloomberg News reports. 

Key Points 

• At the India Impact Summit, Microsoft said it is on track to invest $50 billion by 2030 to expand AI capabilities across the global South.  

• AI usage in the Global South is about half that of the Global North, and the gap is still growing.  

The five-part plan includes:  

• Building data centers  

• Expanding internet access to 250 million people  

• Investing $2 billion each year in training programs  

• India remains a key part of Microsoft’s strategy, with a previously announced $17.5 billion commitment that supports the broader regional effort.  

• Last fiscal year, the company invested over $8 billion in data center infrastructure across the Global South.  

On Wednesday, at the AI Impact Summit in New Delhi, Microsoft said it is on track to invest $50 billion by 2030 to bring AI to developing nations in the Global South. The goal is to help close the narrowing gap in AI adoption between emerging and advanced economies.  

Addressing the AI divide 

Microsoft says its announcement addresses a major gap in AI adoption. The company’s latest AI diffusion report shows that AI usage in the global north is about twice that in the global south, and the gap is still widening. Microsoft sees this investment as a way to help AI expand opportunities and prosperity worldwide, especially for young and growing populations in developing regions.  

Five-Part Investment Strategy 

Microsoft’s $50 billion plan has five main parts:  

1. Building key infrastructure like data centers with reliable power and internet  

2. Skills development  

3. Internet access to 250 million people in areas that lack it  

4. Bring internet access to 250 million people in areas that lack it.  

5. Has already reached 117 million people in Africa by working with partners to build local networks  

Another focus is on skills development. Last year, Microsoft spent more than $2 billion on grants, technology donations, training programs, and discounted products across the global south.  

India as Strategic Anchor 

India is a key part of Microsoft’s AI growth plans. Last year, the company announced $17.5 billion in AI investments for India, strengthening its role as one of the world’s fastest-growing digital markets. These investments include:  

• expanding data centers  

• building AI skills for businesses and start-ups  

• boosting workforce training  

With its large pool of developers, the startup scene is growing and expanding digital infrastructure. India is a strategic base for Microsoft’s wider push into emerging markets.  

Supporting Multilingual and Local Innovation 

Microsoft is investing in language data and AI models to help its systems better aid underrepresented languages in the global south. The company announced a 5.5 million project called Lingua Africa, led by Masa Kane African Languages Hub, Microsoft’s AI for Good Lab, and the Gates Foundation, to focus on open data for text, speech, and vision.  

Microsoft also launched a new project to improve food security in Sub-Saharan Africa, starting in Kenya and expanding across the region. This project uses AI and satellite data in a partnership with NASA Harvest, the Kenyan government, and the East African Grain Council.  

Competitive Infrastructure Play 

Microsoft’s $50 billion investment shows the growing competition among global tech companies to lead the next wave of AI adoption in areas in which digital change is still underway. By presenting these investments as long-term partnerships, not just business expansion, Microsoft intends to become a key partner for governments using AI in public services, health care, education, and finance.  

This strategy addresses a real market gap. The difference in AI use between the global North and South is both an obstacle for developing countries and a new opportunity for tech firms. Microsoft’s focus on local expertise and community needs sets this effort apart from traditional tech experts, striving to make AI solutions relevant and sustainable, not just advanced.

Source: Microsoft Pledges $50B for Global South AI by 2030  

OpenAI is working with India’s Tata Group to secure 100 MW of AI-ready data center capacity, with plans to reach 1 GW eventually. This cooperation is part of OpenAI’s effort to grow its business and infrastructure in one of its fastest-growing markets.  

OpenAI said on Thursday that its partnership with the Tata Group is part of the Stargate project, which aims to build AI-ready infrastructure and encourage more businesses worldwide to use AI. OpenAI will be the first customer of Tata Consultancy Services. Tata Consulting is an open AI data center, starting with 100 MW of capacity. The agreement also includes rolling out ChatGPT Enterprise to Tata’s employees and using OpenAI’s tools to standardize AI-based software development.  

This cooperation is part of the OpenAI for India initiative and demonstrates the company’s growing presence in the country. CEO Sam Altman currently estimates that over 100 million people in India use ChatGPT each week, including students, teachers, developers, and entrepreneurs. With this level of adoption, India has become one of OpenAI’s key growth markets as it increases its business and infrastructure investments there.  

With local data center capacity, OpenAI can run its most advanced models in India. This will reduce delays for users and help meet data-residency, security, and compliance requirements for regulated sectors and government work. Having computing resources in the country is important for businesses that process sensitive data and must follow data localization laws. This could help OpenAI reach more enterprise customers who need in-country processing.  

Starting with 100 MW of capacity is a major step for AI infrastructure, since training and running large models need a lot of powerful GPUs. If the project grows to 1 GW, the Tata facility would become one of the world’s largest data centers, underscoring the scale of OpenAI’s plans for India.  

OpenAI and the Tata Group will also work together to accelerate AI adoption across Tata businesses. Tata plans to introduce ChatGPT Enterprise to its employees over the next few years, starting with hundreds of thousands of AT&T Tata Consultancy Services (TCS) employees. This would also be one of the largest enterprise AI rollouts in the world. TCS also plans to use OpenAI’s Codex tools to make AI-based software development more consistent across its engineering teams.  

N Chandrasekharan, chairman of Tata Sons, said the partnership with OpenAI will help create state-of-the-art AI infrastructure in India and support efforts to train the country’s workforce for the AI era.  

The financial details of the deal have not been shared, and it is unclear whether OpenAI is investing in Hypervault or just leasing capacity.  

In November 2025, TCS received support from private equity firm TPG to build AI-ready infrastructure in India through its Hyper World Data Center business. The platform has about ₹180 billion (about $2 billion) in planned investment and is meant to handle large computing needs for big tech companies and enterprise customers.  

OpenAI will expand its certification programs in India, with TCS as the first organization outside the United States to participate. These certifications are meant to help professionals acquire practical AI skills in different roles and industries, according to the company. This follows OpenAI’s latest partnerships with top Indian institutions in engineering, medicine, and design.  

OpenAI plans to open new offices in Mumbai and Bengaluru later this year, adding to its current office in New Delhi. As it grows its operations in India, this expansion will help support business partnerships, connect with developers, and work with local regulators. As the company increases its presence in the country,  

This announcement arrives as India hosts the AI Impact Summit in New Delhi. Global AI leaders like Sam Altman, Anthropic CEO Dario Amodei, and Google CEO Sundar Pichai are joining Indian startups and companies to showcase AI use cases across finance, healthcare, and education.  

OpenAI has been expanding its presence in India by partnering with companies such as Pine Labs, Geo, Hotstar, Eternal Cause 24, HCL Tech, Phone Pay, Cred, and Make My Trip. The company intends to bring its models to consumer platforms, business systems, and digital payments in one of the world’s largest internet markets.  

The data center expansion, enterprise rollouts, and growing network of partners show that OpenAI is making its biggest effort to establish advanced AI infrastructure and applications in India.

Source: OpenAI taps Tata for 100MW AI data center capacity in India, eyes 1GW 

Remote work has encouraged Google and Microsoft to add more features to their video communication platforms.  

So how do you choose which one is a better fit for your team?  

In this blog, we will compare Google Meet vs Microsoft Teams to help you make an informed choice.  

Google Meet vs Microsoft Teams 

Let’s look at some key features and tools that set Google Hangouts and Microsoft Teams apart.  

Online Conferencing Capabilities 

Both Google Meet and Microsoft Teams offer a wide range of communication features to improve your meetings.  

Both platforms let you record meetings, use chat, set up breakout rooms, run polls, and share your screen. They also support accessibility via live transcriptions and automatic captions. Microsoft Teams offers captions in 50 languages, while Google Meet supports over 65 languages.  

Microsoft Teams lets you send direct messages to individuals, while Google Meet only has public chat. Teams also include a built-in whiteboard for brainstorming. Google Meet removed its Whiteboard tool, Jamboard, in 2024, so users now need third-party options.  

A key feature of Microsoft Teams is its strong webinar support. You can schedule webinars, manage registrations, give presentations, and analyze participant data all within Teams.  

Integrations 

Both Google Meet and Microsoft Teams have their own strengths for app integration.  

Google Meet video conferencing works well with other Google Workspace tools like Gmail, Calendar, Drive, Chat, Sheets, and Docs. It offers around 40 add-ons, including whiteboarding tools such as Miro and Lucidchart. You can also connect it to over 200 popular third-party apps through Zapier.  

Microsoft Teams, on the other hand, is closely connected with the Microsoft 365 suite.  

This makes it easier to share and collaborate on Word documents, Excel spreadsheets, and presentations in the chat. Teams also offers over 700 integrations with tools like Evernote, Zoom for Teams, Trello, ClickUp, and Asana.  

In the end, the best platform depends on your needs and the tools you already use.  

User Interface 

Most people prefer a simple, clear interface that is easy to use and doesn’t require much guidance.  

Google Meet’s simple, easy-to-use design makes setup and navigation quick. It also gives you good control over participant roles and works consistently across devices.  

Microsoft Teams, however, offers more features such as chat, video calls, and collaborative tools, which can take some time to learn if you are new to the platform.  

Meeting Length 

When choosing a video conferencing tool, meeting length matters. Meetings often run longer than planned, and you don’t want an important discussion to end abruptly due to a time limit.  

So, it’s smart to choose a platform with flexible meeting lanes. Luckily, both Google Meet and Microsoft Teams have you covered with generous options.  

Both platforms have free plans that allow group meetings for up to 60 minutes, which is enough for most teams.  

With paid plans, Google Meet allows video calls up to 24 hours, while Microsoft Teams offers sessions for up to 30 hours. This means you can have long discussions free of worrying about time limits.  

Maximum Participant Capacity 

Both Google Meet and Microsoft Teams offer similar free plans that let you host meetings with up to 100 participants. This is ideal for small businesses or teams that need basic conferencing features at no cost.  

With paid plans, Google Meet can support up to 1000 participants at its highest tier.  

Microsoft Teams allows up to 300 participants in its standard plan. The Enterprise Plan supports 500-1000 participants, and for live events, it can handle up to 100,000 participants.  

Storage 

Suppose your organization wants to record meetings. Cloud storage is an important factor to consider when choosing a conferencing tool.  

Google Meet gives each user 15 GB of free cloud storage, and paid plans offer up to 5 TB.  

Microsoft Teams offers 5 GB of storage in its free version, and paid plans provide between 10 GB and unlimited storage.  

Support 

Both Google Meet and Microsoft Teams know that fluid collaboration is important. They offer strong support to help with any questions or technological issues during meetings.  

Free Resources 

Both platforms offer a detailed help center with articles, FAQs, and troubleshooting guides, helping users find solutions to common problems.  

There are also short, helpful video tutorials. Both platforms have active user communities in online forums where people can share tips and find solutions together.  

Business Users 

Upgrading to a paid plan gives you 24/7 phone and web support, so you can get help whenever you need it. Enterprise plans offer even more, with priority support to make sure your issues are resolved quickly, and your meetings stay on schedule.  

Security Features 

Both Google Meet and Microsoft Teams offer strong security features to protect your privacy and data. You can control who joins your meetings, use encryption for data in transit and storage, and add protection with multi-factor authentication (MFA).  

Organizations can use enhanced security programs and single sign-on (SSO) to let users access the conferencing platform and other apps with a single login.  

There are some differences between the platforms. For example, Google Meet asks paid users to turn on MFA themselves, while Microsoft Teams automatically enables it for paid plans.  

Your decision depends on your company’s security needs. If you want an automatic MFA, Microsoft Teams might be a better option. If you prefer to control security settings yourself, Google Meet could be a better fit.  

Pricing 

Both Microsoft Teams and Google Meet offer free plans you can start using right away.  

Teams doesn’t have a free business plan, but you can use the personal plan if its limitations work for you.  

Google Meet’s free plan is a bit more flexible. You can host unlimited meetings with up to 100 people, but each meeting is limited to 60 minutes, which is similar to Microsoft Teams.  

To access Google Meet’s paid features, you’ll need a workspace subscription.  

Here are the costs for Google Workspace plans:  

• Business starter: $8.40 per user per month  

• Business Standard$ $16.80 per user per month  

• Business plus $26.40 per user per month  

• Enterprise: Custom Pricing  

In addition to the free plan, Microsoft Teams offers three paid plans for individuals and businesses.  

• Microsoft Teams Essentials: $4 per user per month  

• Microsoft 365 Business Basic: $6 per user per month  

• Microsoft 365 Business Standard: $12.50 per user per month  

The Verdict 

Choosing between Microsoft Teams and Google Meet depends on your organization’s needs and the platforms you already use.  

If you need a tool for personal or small business use, Google Meet might be a good choice. If your organization already uses Microsoft 365, Teams is likely the better option. The same applies to Google app users, who will find Google Hangouts more convenient. Zoom is strong for video calls, while Microsoft Teams makes it easier to start calls directly from chat.

Source:  Microsoft Teams vs. Google Meet: Which is Best for Small Teams in 2025?