Pogbagram

Google announces Android 17's official transition to a "smart system".

Today, Google held a separate Android Show event ahead of the official I/O developer conference, officially announcing that Android will transform from an "operating system" to an "intelligent system," with Gemini Intelligence becoming the core identifier of the Android platform.

• The default keyboard, Gboard, now features multimodal autofill capabilities, supporting the extraction of information from sources such as photo albums, chat logs, and emails.
• Regarding desktop widgets, Google plans to launch the "Create my widget" feature, allowing users to generate custom widgets using natural language commands;
• The voice input function has also been upgraded, with the new Rambler feature automatically converting spoken language into structured text;
• Gemini's "automatic operation" capability has been expanded from single-step tasks to multi-step tasks, supporting the direct initiation of compound commands by shooting real objects;
• Android Auto will be updated to include Gemini Intelligence, improved 3D road views, and smarter streaming capabilities. Some models will also support remembering vehicle information.

On the hardware front, Google released a brand new laptop category, Googlebook, positioned as "the first hardware tailored for Gemini Intelligence," introducing the "shake cursor" operation mode to bring up Gemini, and also supporting direct running of Android applications.

At the system level, Android 17 also brings a collaboration with Meta, allowing apps like Facebook and Instagram to support accessing native camera functions; flat-style emojis are upgraded to 2.5D styles.

It's worth noting that at the launch event, Google also announced that QuickShare's compatibility with AirDrop will be extended to some models from brands such as OPPO, vivo, OnePlus, and Honor, with rollout expected in the second half of this year.

Reports suggest that Apple's iOS 27 camera will support custom layouts and include a Siri mode.

According to Bloomberg, Apple is planning a complete overhaul of the camera app in iOS 27, introducing a fully customizable control layout. Users can switch to "Advanced" for customization, with each shooting mode having its own independent widgets, selectable from a semi-transparent "Add Widget" tray that slides out from the bottom.

• The advanced options in photo mode include depth of field and exposure control, with controls grouped into three categories: "Basic," "Manual," and "Settings."
• The control overview entry, originally located in the upper right corner of the interface, has been moved to the right of the shutter button, and grid and level options have been added;
• The camera app will also include a Siri mode, integrating visual intelligence features to support operations such as plant recognition and text translation.

In addition, the report also revealed changes that several system apps will bring in iOS 27:

• Safari will get a new start page with four new tabs at the top, allowing you to quickly switch between favorites, bookmarks, reading list, and browsing history;
• The Weather App's main page will add a "Weather Conditions" panel, allowing users to view rainfall and wind speed data directly in addition to temperature without having to navigate to a different page.
• The image creation app "TuLeYuan" has been completely redesigned, with a simpler interface and an image editing entry that provides a description of the content to be modified.

Tencent's Zhang Jun: WeChat's read receipt and visitor features are "fixed" and will not be developed.

Recently, the hashtag "WeChat Status to Launch Visitor Records" trended on social media. Yesterday, a WeChat employee, "Kecun Xiaojiang," clarified that the trending topic was highly misleading. He explained that WeChat Status had recently tested two features on a "very limited scale," and both have since been discontinued.

• The status viewer count is displayed so that users can see the number of views of the status within its validity period after posting it. Only the number is displayed, and the specific users are not shown.
• The status page randomly displays pop-up icons of passing friends; clicking on them allows users to view the statuses posted by those friends, aiming to promote interaction between friends who post statuses.

Tencent's public relations director, Zhang Jun, subsequently reposted the Weibo post and added that the user misunderstood the function of testing the number of visitors on a small scale, "and this is our responsibility."

He further emphasized that the read receipt and visitor functions had received clear objections from users on multiple occasions, and that "these two functions are now fixed and will not be developed or provided."

Ilya revealed that she still holds approximately $7 billion in equity in OpenAI: she doesn't want it to be destroyed.

According to reports from the Los Angeles Times, CNBC, and Wired, the trial of Elon Musk v. OpenAI and CEO Sam Altman continued on the 11th local time. Microsoft CEO Satya Nadella and OpenAI co-founder Ilya Sutskever testified in court.

Nadella testified that Musk never communicated with him regarding whether Microsoft's investment in OpenAI violated any commitments. He stated that Microsoft was "very proud" of its early investment in OpenAI and emphasized that the partnership had a clear commercial purpose from the outset.

Nadella described OpenAI's board's firing of Altman in 2023 as an "amateurish move," saying it consistently failed to provide a convincing explanation.

Ilya Sutskever later testified in court that she still holds approximately $7 billion in equity in OpenAI's for-profit entities, making her one of the largest known individual shareholders.

Sutskever admitted to being involved in pushing for the brief dismissal of Elon Musk in 2023 and stated that he has a strong sense of belonging to OpenAI, "I just don't want to see it destroyed." Regarding issues involving Musk's core demands, Sutskever stated that OpenAI's transformation into a for-profit architecture is a consensus among all parties.

Xiaomi is giving away 100 trillion tokens for free.

Xiaomi announced yesterday the official launch of the "MiMo Orbit 100T Token Program," which will distribute token benefits to AI users worldwide for free, with a plan to distribute a total of 100 trillion tokens within 30 days.

Lu Weibing, partner and president of Xiaomi Group, forwarded the relevant content and said that as of yesterday morning, the plan had distributed nearly 80 trillion tokens.

Xiaomi Technology officially cited the latest data from OpenRouter, showing that Hermes Agent's daily token calls reached 291 billion, with over 1.75 trillion calls in the past week. In terms of model call volume, MiMo ranked first, contributing a cumulative 1.45 trillion token calls in the past month.

Li Xiang: Ideal's self-developed chips are not just following the trend; they aim to be "all-rounders" in the AI ​​era.

Li Xiang, CEO of Li Auto, posted on Weibo yesterday that Li Auto's chip development is not "burning money to follow the trend," but rather to enable AI to truly operate in the physical world and to use self-developed chips to solve problems that existing suppliers' technologies cannot overcome.

In his article, Li Xiang uses Apple as a reference, pointing out that Apple's ability to achieve the best user experience does not rely on the strongest technology in any one area, but rather on its independent design and responsibility for the entire chain of self-developed chips, operating systems, hardware, and cloud services.

He emphasized that the ideal goal is to "bring AI into the physical world and give everyone an Apple-like experience," but "not to make phones or imitate UI design, but to deliver an excellent experience through its integrated hardware and software design."

Accordingly, Li Auto is currently advancing in the areas of self-developed chips, operating systems, large models and hardware, which Li Xiang defines as "comprehensive joint design for the era of artificial intelligence".

Only by achieving "N all-around excellence" can one become the champion of user experience.

The first nationwide case involving AI-generated product recommendation notes has reached its final verdict: the defendant was awarded 100,000 yuan in damages, and the mass production of "digital swill" by AI constituted copyright infringement.

The Hangzhou Intermediate People's Court in Zhejiang Province recently disclosed the final judgment in the first case nationwide involving unfair competition related to AI-generated "product recommendation notes".

The court found that the AI ​​writing tool jointly operated by Company B and Company C, which used a well-known social media platform as its targeted application layer to generate product recommendation copy in bulk and induce users to post it, constituted unfair competition. The court ordered the two companies to cease providing the relevant services and pay RMB 100,000 in compensation.

The tool in question is available in three versions: web, app, and WeChat mini-program. Popular modules are named directly with titles such as "Xiao Xiaomoushu Seeding Copywriting". Membership costs 40 yuan per month or 168 yuan for a lifetime membership.

The two companies' "technology neutrality defense" was not supported. The court pointed out that technology itself is neutral, but commercial applications must comply with platform rules and legal bottom lines.

Steve Jobs commemorative coins are now officially on sale: a pack of 25 is priced at $61.

According to MacRumors, the U.S. Mint officially launched a $1 "American Innovation Commemorative Coin" featuring Steve Jobs yesterday, offering two purchase options: a roll of 25 coins for $61 (approximately $2.44 per coin) and a bag of 100 coins for $154.50.

The obverse of the coin depicts a young Steve Jobs, dressed in his signature turtleneck sweater, jeans, and sneakers, sitting in front of a natural landscape in Northern California.

JD.com's revenue reached 315.7 billion yuan in the first quarter, with annual active users exceeding 740 million, marking the 10th consecutive quarter of double-digit growth.

JD.com released its financial report for the first quarter of 2026 yesterday.

First-quarter revenue increased by 4.9% year-on-year to RMB 315.7 billion; net profit attributable to ordinary shareholders was RMB 5.1 billion, compared to RMB 10.9 billion in the same period last year. Service revenue reached RMB 70.9 billion, a year-on-year increase of 20.6%.

• JD Retail's revenue in the first quarter was 268.59 billion yuan, a year-on-year increase of 1.8%;
• As of the end of the first quarter, JD.com had over 740 million annual active users, marking the 10th consecutive quarter of double-digit growth since the fourth quarter of 2023.
• Revenue from new businesses, including JD Food Delivery, JD Industrial Development, Jingxi, and overseas operations, reached RMB 6.28 billion in the first quarter, a year-on-year increase of 9.1%.
• Marketing expenses in the first quarter increased by 45.8% to RMB 15.4 billion from RMB 10.5 billion in the same period last year, mainly for the promotion of new businesses;
• Research and development expenditures increased by 48.6% to 6.9 billion yuan from 4.6 billion yuan, and were used for technological capacity building and talent investment.

In terms of employee size, JD.com currently has over 900,000 employees. The company previously disclosed that it has provided 28,000 housing units for frontline employees and plans to invest 22 billion yuan over the next five years to provide 150,000 "JD Service Provider Homes".

The originator of the "vibe coding" concept believes that the ultimate output of AI is "interactive neural video," and for now, let's try HTML.

Andrej Karpathy, the originator of the "vibe coding" concept and a founding member of OpenAI, posted on X yesterday, publicly supporting Thariq, a member of the Claude Code team, for advocating "replacing Markdown with HTML".

Karpathy believes that the output format of large models is gradually evolving: plain text is difficult to read, Markdown is better, and HTML goes a step further, supporting richer typography, charts, and interactions. He predicts that there will be multiple iterations in the future, with the ultimate goal of creating "interactive neural videos" directly generated by AI.

Karpathy points out that about one-third of the human brain is dedicated to processing visual information, and vision is the most efficient channel for receiving information. Therefore, it is most natural for humans to "speak" to AI using voice, while AI should present results to humans as visually as possible.

He also acknowledged that existing input methods are insufficient—relying solely on voice or text lacks the spatial feel of "pointing to a spot on the screen." Until a better solution emerges, he suggested that users simply add "structure your reply as HTML" to the end of the prompt and then open it in a browser to view it .

Related reading: Is Markdown dead, and HTML the new king?

Starting at 3.9 million yuan, Unitree unveils the world's first mass-produced manned transforming mech, the GD01.

Yesterday, Unitree Robotics released the GD01 manned transforming mecha, priced from 3.9 million yuan.

Official test videos show that the GD01 can knock down a brick wall with a single punch. The GD01 is described as the world's first mass-produced manned mech, capable of transformation and also serving as a civilian vehicle, weighing approximately 500kg (when carrying a person).

Finally, Yu Shu reminded everyone: "Please use the robot safely and in a friendly manner."

Hermes Agent introduces background control functionality for Mac computers.

The AI ​​intelligent agent framework Hermes Agent recently launched the "Computer Use (macOS)" feature, which allows AI to control a Mac computer to complete complex tasks without the user's awareness—the mouse cursor does not move, windows do not pop up in the foreground, and keyboard focus does not switch.

Hermes' desktop control functionality is not tied to any specific AI model and can be used with Claude, GPT, Gemini, or open-source models running locally.

In terms of security mechanisms, Hermes has multiple layers of restrictions on high-risk operations. Operations such as emptying the trash can, forced deletion, and locking the screen to log out are hard-blocked, and clicks and inputs require user confirmation by default before they are executed.

Doubao Input Method Launches on macOS

Yesterday, the macOS version of Doubao Input Method was officially launched on the official website, version number 0.9.0, supporting macOS 10.15 and above. The Windows version displays "Coming Soon".

Claude Code adds Agent View: manage all parallel tasks from a single interface.

Claude Code introduces the "Agent View" feature, allowing users to manage all Claude Code sessions in a unified interface. It is currently available as a research preview to Pro, Max, Team, Enterprise, and Claude API program users.

Users can access Agent View by pressing the left arrow key in any session or by executing the `claude agents` command in the terminal. Each row in the list displays the session name, a summary of the latest reply, and the last interaction time.

After selecting a session, you can preview the content of the most recent conversation; for tasks that you want to run in the background, you can use the /bg command to put them in the background, or start a new session directly in background mode using the claude --bg [task] command.

NetEase Youdao launched ThinkFlow, a large model aggregation platform that allows users to access and call 20+ models at once.

Recently, NetEase Youdao Smart Cloud launched ThinkFlow, an enterprise-level large model aggregation platform. Through standardized API interfaces, it integrates more than 20 mainstream large models such as DeepSeek, Kimi, Qwen, and MiniMax into a unified schedule, allowing enterprises to flexibly switch between multiple models after completing the integration once.

ThinkFlow provides a full-chain token consumption visualization dashboard, pinpointing the cost of large model calls down to the level of a single request. Simultaneously, the platform centrally hosts all API keys and supports deploying model services on enterprise-owned servers or dedicated cloud environments, ensuring that all data calls are processed on the enterprise side.

According to official sources, ThinkFlow's underlying architecture originates from NetEase's large-scale internal business practices and has been validated in real-world scenarios such as NetEase Youdao, NetEase Games, NetEase Cloud Music, and NetEase Smart Enterprise, supporting a smooth transition from small-scale proof-of-concept to large-scale production environments.

Nintendo President: Switch 2 Price Increase Not Ruled Out

According to ComicBook and Nintendo Everything, Nintendo President Shuntaro Furukawa stated during an investor Q&A session following yesterday's earnings release that there is a possibility of another price increase for the Switch 2 in the future, and the company will maintain "flexibility" in its pricing strategy.

Previously, Nintendo had announced that it would raise the price of the Switch 2 later this year. The price in the US market will increase from the initial suggested retail price of $450 to $500, an increase of $50; the price in the Japanese market will increase by up to 400 yuan.

Shuntaro Furukawa stated that the price increase was not driven by a single factor, but rather by a combination of factors, including soaring costs of components such as memory, fluctuating exchange rates, and rising oil prices. He said that maintaining the original pricing would lead to a significant deterioration in hardware profitability, which would then have a medium- to long-term negative impact on the company's overall business.

The upward trend in component prices may continue not only into this year but also into next year. The company hopes to "maintain pricing flexibility" and prepare accordingly.

This statement has been widely interpreted as indicating that another price increase is not out of the question. He added that if market conditions return to normal, a price reduction would also be considered.

New regulations for short videos have been introduced: 6 mandatory tags are required, and content containing AI-generated content must be labeled.

Yesterday, the official WeChat account of "Cyberspace Administration of China" published an article disclosing new regulations regarding content labeling functions on short video platforms:

Each website platform must provide users with six types of "required tags" and make content tagging a mandatory step in posting short videos. Posters must select one of these tags to complete the posting process. These tags are:

"Contains fictional content," "Contains AI-generated content," "Contains marketing information," "Content is reprinted," "Content represents personal opinions," and "No labeling required." For short videos documenting real life, the "No labeling required" option can be selected; this label will not be displayed on the short video page.

For content that is not properly labeled, platforms must add or correct the labeling and educate and warn the relevant publishers. The Cyberspace Administration of China stated that it will further strengthen guidance, supervision, and inspection of website platforms, and will severely punish and publicly expose accounts that fail to label content as required and platforms that fail to fulfill their responsibilities.

Ink shortage leads to black and white packaging for Japanese potato chips.

According to Nikkei Asia, affected by the ongoing tensions in the Middle East, Calbee, a well-known Japanese food company, announced that it will gradually change the packaging of some of its products to a black and white version starting May 25.

Calbee stated that due to the tense situation in the Middle East, the supply of some raw materials has become unstable, and the company has made stabilizing the supply its top priority in the face of the unpredictable situation in Iran.

The first trailer for Matthew McConaughey's new film "The King's Rival" has been released.

According to blogger "Hollywood Watcher", the first trailer for Matthew McConaughey's new film "The Rivals of Amziah King" has been officially released.

Directed by Andrew Paterson, the film stars McConaughey as Amziah King, a music-loving beekeeper living in rural Oklahoma. The plot revolves around the sudden return of his adopted daughter, and the two run their beekeeping business together, only to have it sabotaged by villains, forcing them to fight back.

Reports indicate that a new Westworld movie is in the works, with script written by David Cope, screenwriter of Jurassic Park.

According to Deadline, screenwriter David Koepp will write the script for Warner Bros.' Westworld movie, and a heavyweight director is currently in talks for the project, though the specific candidate has not yet been disclosed.

The story of Westworld is based on the 1973 film of the same name written and directed by Michael Crichton. It tells the story of an adult theme park that allows wealthy tourists to experience the life of cowboys in the Wild West, which falls into panic when a robot gunman malfunctions and begins to actually shoot tourists.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

Every era has its own unique symbols—the first decade of the 21st century might be the "Olympics," the second decade might be the "Internet," and so on.

As a result, just halfway through this third decade, we saw a powerful competitor to a symbol of our time: memory.

Image | Wccftech

After all, memory is present in the vast majority of the world's billions of electronic devices, including home appliances, automobiles, infrastructure, and aerospace.

The booming development of artificial intelligence goes without saying—

It is both the instigator of the memory crisis and the biggest beneficiary of memory technology, only now it is in a more prominent position in the spotlight.

Figure | OpenAI

But the problem we face is neither AI nor memory, but this extensive development model of "relying on piling up memory and computing power to develop AI at all costs".

It is encroaching outwards from its original domain, gradually affecting all aspects that do not want to, and should not, be affected by the technological crisis.

1

As the most direct victim of rising memory chip prices, the mobile phone market in 2026 is destined to be anything but calm.

In the midst of the storm, small factories can neither band together for mutual support nor influence industry trends. They can only maintain their profit margins by constantly cutting low-end lines, subtly lowering prices for flagship lines, and directly raising prices.

Even more shockingly, although Samsung phones and Samsung memory share the same name, they show no mercy to each other. The DS division raised prices and blocked supplies as it should, and the MX division did the same.

In such situations, Apple and Huawei, which have reserved ample profit margins, can grit their teeth and persist in price wars to capture more market share.

But given time, even a ship of this size will not be able to withstand it.

Whether it's the recently released Pura 90 series or the rumored iPhone 18 series, we've seen trade-offs made due to runaway memory prices.

What upsets ordinary consumers is that the rise in memory prices not only affected the initial suggested retail price, but also turned the secondary market and secondhand trading prices upside down.

Image | Xianyu

Take OPPO, which has the latest mid-term product line, for example. Not long after the first half of the year's release cycle ended, there were already multiple rounds of "price increases upon release, discounts for first sales, and price drops on the secondhand market".

As for the original entry-level models and even lower-priced contract phones, they have almost disappeared in this spring that should have been theirs, just as iFanr summarized last year:

The era of abundant and readily available mobile phone hardware is over; what lies ahead is a comprehensive price increase and reduced specifications.

2

The computers that followed closely behind suffered an impact that was no less severe in comparison.

While the iPhone can maintain a "basically unchanged" price by relying on profit margins and supply chain bargaining power, the Mac, which is more constrained by costs, is not so lucky.

Since last year, although Apple has rarely made official statements about encountering difficulties in memory procurement, the reduction in its product line is a real issue.

Image | AppleInsider

For example, the Mac mini, which we all love, quietly removed the 32GB and 64GB memory options from the M4/M4 Pro models, leaving only 24GB and 48GB options.

The M3 Ultra fared even worse—it originally offered 96GB, 256GB, and 512GB options, but now only the basic 96GB version is available.

Not to mention the MacBook Neo, which is currently experiencing unprecedented popularity. Originally, the A18 Pro inventory was already running low, but now the memory chips are also becoming insufficient.

Foreign media reports suggest that Apple is considering removing the 256GB option from the entry-level Neo model, almost entirely deviating from the original design.

Windows isn't having an easy time either, since RAM and SSDs have always been a pair of star-crossed lovers.

Coupled with the disappointing performance of Windows 11 and the relentless demand for local AI, even disregarding graphics cards, the DIY PC market has effectively "died" over the past six months.

Image | Reddit

The laptop market below $500 (RMB 3,500) is in complete disarray.

The traditional laptop's profit model was originally very simple and limited, and the huge price fluctuations of essential components such as memory have even led to a renaissance of 8GB memory.

The video game industry also suffered.

Nintendo recently announced that the price of the Switch 2 is expected to increase from 49,980 yen to 59,980 yen, an increase of approximately 434 yuan.

Image | Notebookcheck CN

Meanwhile, Valve's highly anticipated Steam Machine has been in development for a long time since last year, and its fate remains uncertain .

3

Unfortunately, the impact of memory extends far beyond the technology industry—the challenges they face present a rare opportunity for some long-established "traditional crafts."

Recently, South Korean media reported that theft gangs have shifted their primary targets from gold, silver, and valuables to memory modules.

The offices of some tech companies and startups in Seoul have been targeted by organized crime, with thieves making quick, deliberate moves like those in Escape from Tarkov, grabbing only computer cases and RAM sticks before fleeing .

Image | XDA Developers

After all, memory modules are not only lightweight, but also do not require remelting and cutting like gold and jewelry. Once they enter the market, tracing their origin is extremely difficult, and it is only a matter of time before price-related crimes occur.

Besides "traditional handicrafts," the automotive industry has also been affected this year. According to The Paper:

Starting May 1st, BYD adjusted the price of its assisted driving option for some models, including the Dynasty, Ocean, and Formula Leopard, from 9,900 yuan to 12,000 yuan, citing "a significant increase in global storage hardware costs."

Image | BYD

Even Elon Musk, known for his outspokenness, couldn't escape this trend. During the earnings call in late January, he expressed concern about the impact of memory prices on Tesla's Dojo supercomputer and robots, and made a bold statement:

If supply from Samsung and Micron continues to tighten, Tesla may consider building its own semiconductor factory to produce memory chips in the future.

4

The most outrageous impact of the memory price hike is on the dating market in South Korea.

With the development of HBM (High Bandwidth Memory) for AI servers to its current state, only Samsung Electronics and SK Hynix remain in the supply chain, giving South Korea a significant advantage in the global AI technology market.

According to South Korean media reports, Hynix abolished the bonus cap system last September, and instead directly distributed 10% of its annual operating profit as performance bonuses to all employees .

Image | Seoul Economics Daily

Hynix's projected profit for fiscal year 2026 will reach 250 trillion won (approximately US$169 billion), and the bonuses for its nearly 35,000 employees are expected to reach 14 million won (approximately US$95,000 or RMB646,000).

As a result, overnight, the prestige of "I work at Hynix" quickly surpassed that of "I work at Samsung".

Some Hynix employees have even anonymously stated online that they have been receiving so many blind date invitations that they are "overwhelmed" with them.

A clip from a South Korean variety show

While Hynix is ​​celebrating, Samsung Electronics is busy with a strike.

On April 17, the Samsung Group union negotiated with Samsung, demanding a 7% pay rise, the removal of the bonus cap, and the distribution of 15% of operating profit as bonuses. Otherwise, a general strike would be held from May 21 to June 7.

Photo | Korea JoongAng Daily

Scholars have calculated that such a large-scale strike would cause Samsung direct losses of approximately $6.9 billion to $17.7 billion and damage its reputation as an HBM4 chip supplier. It is understood that the two sides have reached an agreement on a 13% bonus.

The bubble continues

Now that things have come to this point, the "memory price increase" has moved beyond the supply chain fluctuations originally caused by AI capacity expansion. Instead, it has spread to all aspects of daily life through electronic products that are now ubiquitous.

Contrary to what many readers who boldly declared in the iFanr comment section, "I'll refuse to buy! Let the manufacturers feel the pain!" might have imagined—

If ordinary people don't buy it, it won't really make a difference to the price increase of memory.

After all, whether it's HBM for AI or DDR/LPDDR chips, the vast majority of sales come from downstream manufacturers.

Samsung and SK Hynix would love for ordinary consumers to stop buying anything , so they can dedicate all their production capacity to HBM and high-performance VRAM chips. AI giants and tech companies will still be lining up to buy them, and they will actually make more money.

Unfortunately, no one knows how long this chaotic wind will blow.

No one knows what kind of era AI will usher in, but the current AI bubble has undoubtedly brought an end to the golden age of microcomputer science from the 1980s to the present.

Memory is ruining everything.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

As with last year, Google held a small separate event for Android before the official Google I/O developer conference.

At this year's Android Show, Google practically unveiled its grand vision for Android and the entire AI product ecosystem for the coming year.

In addition to Android 17 as the foundation, we also saw some unexpected platform updates and the release of several hardware products.

The key point is that although there aren't many products and it's still some time before Android 17 is widely rolled out, we can already see enough of Google's plans for the AI ​​ecosystem over the next five or even ten years.

What's even more crucial is that tonight's Google is exactly what Apple envisions itself as after entering the AI ​​era.

From operating system to intelligent system

Right at the start of the event, Google announced a significant initiative:

Android will no longer be a simple operating system, but an intelligence system.

Image | Google

In other words, the software and hardware usage method that was once the absolute mainstream of "manual user operation" is, in Google's view, a symbol of a previous era.

Starting today, Android, as a cluster of intelligent functions, will more proactively intervene at the forefront of the user's operation process, anticipating your needs and doing what you do.

Image | Google

This is why the specific system version Android 17 was mentioned very few times at this event, while Gemini Intelligence became the system's synonym.

First, Gemini Intelligence, as the latest form of Google Gemini on mobile hardware, greatly enhances its multimodal, cross-environment, and highly integrated operating mode.

For example, the default keyboard Gboard, which serves as the primary input point for the system, received a functional enhancement.

Leveraging Gemini Intelligence's multimodal capabilities, the "autofill" function, which was previously less prominent in keyboard functions, has greatly expanded its information sources:

Image | Google

In addition to manually saved passwords, Gboard will also support automatically filling in ID information from the photo library, address information mentioned in chat, schedule information from email composition, and more.

To put it more bluntly: Gemini Intelligence has gone far beyond simply "helping you remember passwords," and truly acts like an assistant, helping you remember and suggesting information from various sources and of various types.

Another effective way to utilize this multimodal information is through Android's desktop widgets.

In Android 17, Gemini Intelligence will support a feature called "Create my widget," but it won't be available immediately; the goal is to release it later this year.

Image | Google

The main function of this new feature is to create new widgets on the desktop based on your commands, similar to Vibe Coding, breaking the limitation that widgets can only be those pre-installed in the app.

For example, instead of a simple calorie tracking app, I could say to Gemini: "Make a widget that recommends two high-protein meals to me every weekday."

Image | Google

In this way, desktop widgets truly become the entry point for a complex task, essentially the same as a person coding an app.

In addition, users who prefer voice input will also benefit – the new version of Gboard will support a highly intelligent voice input function similar to Typeless, called Rambler.

Compared to having to say "comma… period…" aloud before, Rambler can translate, clean up, and organize an entire speech full of "uh-huhs" into a neat text:

Image | Google

On the other hand, Gemini Intelligence's automated execution capabilities have also been further enhanced.

At last year's Google I/O and product launch event, Google demonstrated how to give Gemini commands to automatically order takeout, hail a ride, book tickets, and perform other similar tasks. The response was positive after its official launch.

In Android 17, Gemini was upgraded to Gemini Intelligence, and this "proxy operation" also supports multi-step tasks.

For example, previously only a simple "Book me a ticket" was supported, but now you can directly take a picture of a travel brochure in the Gemini chat box and tell Gemini, "Find me a similar group tour for two on Ctrip."

Image | Google

The key point is not what tasks Gemini can perform, but that it has a more powerful ability to "do one more step." Sometimes it is this extra step that makes AI go from "usable" to "useful."

Of course, the Android 17 update wasn't just a one-man show for Gemini; Google also optimized many aspects of the "Android core experience."

In Android 17, Google and Meta partnered to enable native camera features in apps like Facebook and Instagram, such as Ultra HDR, Super SteadyShot, and Night Video.

Google's flat-style emojis, which they've used for many years, have also been updated—changing from pure 2D to 2.5D, making them more stylistically similar to the emojis used on iOS.

Image | Google

The QuickShare feature that we previously introduced, which is compatible with AirDrop, will also support devices from more manufacturers on Android 17.

Besides Samsung and Pixel, newer models from (international versions) OPPO, Vivo, OnePlus, and Honor will also receive firmware updates to be compatible with AirDrop in the second half of this year.

Image | Google

Google's AI PC

Beyond Gemini Intelligence, Google has also not forgotten to create a set of tailor-made hardware for these more complex and powerful AI functions.

The form of this new hardware is neither a pendant, nor headphones, nor a watch or bracelet—it's the old Chromebook.

That's right, in the AI ​​era, Google has renamed its laptops yet again.

From the Pixelbook to the Chromebook, and now the latest Googlebook:

Image | Google

Like Chromebook, Googlebook is not a specific laptop model, but rather a name that can be used by any third-party manufacturer that meets the standards.

The Googlebook is the "first hardware product specifically designed for Gemini Intelligence".

In addition to the new Gemini Intelligence features mentioned above, Googlebook has made revolutionary innovations in the most frequently used basic human-computer interaction aspects—

To access the Gemini feature on a Googlebook, you don't need to speak, press any keys, or use a right-click menu; you simply need to "shake the cursor."

Image | Google

Based on Gemini Intelligence's multimodal capabilities, the recommended AI commands can even be automatically adjusted based on factors such as the content under the cursor, the selected content, and the operations that can be performed on the screen.

With AI features becoming increasingly prevalent on computers and laptop keyboards lacking sufficient space, Googlebook's "Magic Pointer" is undoubtedly one of the most intuitive and elegant solutions.

In addition, the Googlebook solved a long-standing problem in the history of ChromeOS: it was a Google product, but it couldn't run Android apps.

In other words, all apps on Android phones can run directly on Googlebook, basically matching the current level of macOS running iOS software.

The foundation of all this is the GKI (Generic Kernel Image) project, which Google has been vigorously promoting, as mentioned in a previous article by iFanr. It is enabling Android to break free from the constraints of mobile phones and seamlessly connect to more types of devices.

Although Google has not yet mentioned the compatibility of this feature, we speculate that, based on processor specifications and network conditions, Googlebooks should support both local operation and screen mirroring via mobile apps.

Image | Google

According to the event information, the manufacturers of the first batch of Googlebooks are still the same familiar names: Acer, ASUS, Dell, HP, Lenovo, etc., and the first batch of products are expected to be launched this year.

Image | Google

In addition to laptops, Android 17 also updated some features of the Android Auto in-vehicle infotainment system.

For example, Gemini Intelligence, which is more like OpenClaw, provides automated operation, optimized 3D road visuals, and more intelligent streaming media playback capabilities.

It also supports displaying custom widgets created on your phone on the car's infotainment system.

Image | Google

Meanwhile, the range of brands that natively support Android Auto is also increasing, with some models even supporting the memorization of current vehicle information , such as trunk size and dashboard specifications.

In this way, when users use Gemini's Q&A feature, the car's system can provide specific answers, such as "Can I put two 27-inch suitcases in at the same time?" or "What does that assassin-like warning light mean?"

Unfortunately, this feature will not be available immediately and is expected to be available in products "later this year".

Overall, this event was just an appetizer for this year's Google I/O, but the conceptual changes it involved were very fundamental.

In fact, at the beginning of the event, Google pointed out that good artificial intelligence technology should be imperceptible to people, and it should be integrated into every layer of software and hardware experience.

Image | Google

And that's exactly what Gemini Intelligence is doing.

Whether it's a Pixel phone, an Android Auto car infotainment system, or a Googlebook, these hardware devices are ultimately just one way of embodying Gemini's intelligence.

What's interesting is that what Google did tonight is exactly the AI ​​ecosystem that Apple has been desperately trying to achieve.

The idea is to allow iPhones, watches, and Macs to share a common intelligent system, so that users will experience highly similar functionality and experience regardless of where they use them. The hardware will only differentiate the interaction methods, without affecting the level of intelligence.

Image | Apple

Unfortunately, despite Apple's long struggle, it has failed to master the "model" aspect, and its hardware has instead become a stepping stone for other companies' models.

Future intelligence systems will take on more diverse forms than they do now, but their core will be more unified.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

If the initial impact of large-scale models was that they made users accustomed to engaging in natural language conversations with machines, then the next step for agents is to move AI from "answering questions" to "completing tasks." It is no longer just a chat window, but a system-level capability that can understand scenarios, break down tasks, invoke applications, coordinate devices, and remember user preferences over a longer period.

For many, AI is not just a chatbox, but also an AI agent.

In the past, devices such as mobile phones, tablets, PCs, cars, headphones, and watches primarily provided functions within their respective hardware forms and operating systems. However, with the widespread adoption of agents, the boundaries between terminals will be further blurred. What users truly need is not necessarily to open a specific app and complete an isolated operation, but rather for mobile phones, cars, IoT devices, and cloud services to collaborate within a continuous life scenario, proactively providing more suitable services.

Over the past two or three decades, chips in smart devices have evolved from determining the primary computing tasks to becoming the underlying hardware that determines benchmark scores, power consumption, gaming, and imaging capabilities. Now, chips also need to become the entry point for AI experiences, the foundation for computing power, and the interface for the ecosystem.

This is the core message MediaTek wanted to convey at the Dimensity Developer Conference MDDC 2026: on the one hand, it provides a computing power foundation covering multiple scenarios through a full-stack product portfolio including mobile phones, automobiles, IoT and AI infrastructure; on the other hand, it opens up more capabilities to developers through the Dimensity AI Intelligent All-in-One Engine, AI Development Kit, automotive platform and gaming technology.

Dimensity AI: From edge computing power to system-level Agent OS

In the process of implementing AI Agents, mobile phones remain one of the most critical terminals.

The reason is not complicated. Smartphones possess the densest user data, the most frequent usage scenarios, and the most mature application ecosystem. They are both the entry point for personal information and the hub for cross-device collaboration. Therefore, as the agent moves from the application layer to the system layer, the smartphone naturally becomes the primary venue for the intelligent agent experience.

Over the past three years, the Dimensity AI ecosystem has achieved significant growth: the number of ecosystem partners has increased by 240%, and the download volume of the Dimensity AI development kit has increased by 440%. MediaTek also mentioned that the number of autonomous tasks performed by intelligent agents has increased from 120 million per day in 2025 to 870 million per day in 2026, a sevenfold increase in one year. These figures at least demonstrate that agents are no longer just a conceptual future narrative, but have begun to enter the growth path for developers and users.

In response to this trend, MediaTek launched the Dimensity AI Intelligent Engine 2.0 at the conference.

Compared to version 1.0, which was more user-command driven and executed single tasks independently by apps, the key change in version 2.0 lies in proactive sensing. Leveraging Dimensity SensingClaw technology, the Dimensity platform can provide low-power, always-on sensing capabilities, enabling device manufacturers to create Agent OSs with proactive sensing and cross-application driving capabilities.

In other words, future AI assistants will not just be "you say a sentence and it does a step", but will be able to understand the user's situation in advance based on information such as vision, hearing, location, and environmental status, and mobilize different applications and devices to complete more complex tasks.

At the conference, MediaTek announced its collaborations with OPPO, Xiaomi, and Transsion.

Dimensity technology will empower OPPO's Xiaobu Assistant, integrating system-level native application data and building a user-specific memory database with Xiaobu Memory. It can perform tasks such as interpreting medical reports, automatically planning fitness programs, and importing them into the calendar. Xiaomi focuses on cross-device smart experiences, allowing users to execute tasks across all devices with a single command, enabling task transfer between multiple devices. Transsion emphasizes Always On proactive sensing capabilities; for example, its AI assistant can complete complex tasks such as checking logistics and comparing prices in a wake-free, self-running state.

The cases from these three leading mobile phone brands all point to one thing: AI agents are no longer just an application within a mobile phone, but are becoming part of the system-level capabilities. This requires not only sufficiently high AI computing power from the chip, but also continuous on-device sensing capabilities, and collaboration between privacy, security, and the application ecosystem.

Development Kit 3.0: Making Edge AI Easier to Deploy

If the AI ​​intelligent engine is geared towards system-level experience, then the Dimensity AI Development Kit 3.0 is a toolbox for developers. Chip platform companies have consistently emphasized the value of edge AI: faster response times, better privacy protection, stronger offline capabilities, and reduced reliance on cloud resources.

However, actually deploying the model to terminals such as mobile phones, tablets, and in-vehicle systems is not a simple matter of "transferring" it. Developers often have to face a series of engineering problems such as model compression, operator compatibility, power consumption control, memory usage, and deployment efficiency. The Dimensity AI Development Kit 3.0 is designed for this purpose.

Version 3.0 is the first to support visual deployment of LVM models, upgrading from command line to a modular GUI. Parameter settings can take effect in real time, improving model deployment and tuning efficiency by 50%. For many application developers, this lowers the barrier between the model and the terminal.

The newly added Low Bit compression toolkit can reduce device memory usage during generative AI model compression, improving model compression efficiency by up to 58% while maintaining the same quality. This is especially crucial for large edge models, as terminal devices have more limited memory, power consumption, and heat dissipation space. The eNPU development toolkit helps developers fully leverage the advantages of the Dimensity chip's highly efficient NPU, saving 42% on power consumption for resident, lightly loaded AI models. For always-on perception, voice wake-up, and environmental recognition scenarios, low power consumption is more important than peak computing power. For AI to truly be always online, the energy consumption problem caused by "always running" must first be solved.

Tianji AI Partner serves as a one-stop model-to-device conversion assistant. It automates processes such as model analysis, adjustment, and verification, helping developers reduce a model conversion task that might have taken 5 days to about half a day, and saving up to 90% of the time required for deploying edge-side LLM models.

These upgrades demonstrate that MediaTek is not merely showcasing AI computing power as a chip parameter, but rather attempting to fill key gaps in the development process. For developers, the real hurdles of edge AI extend beyond "how powerful the chip is," encompassing "how smooth the toolchain is," "how fast the model adaptation is," and "how low the optimization costs can be."

Automotive Platforms: From Smart Cockpits to AI-Defined Cars

Over the past few years, "software-defined vehicles" has become an industry consensus. Smart cockpits, autonomous driving, over-the-air (OTA) updates for vehicles, and vehicle cloud services have transformed cars from transportation tools with mechanical structures at their core into continuously updated intelligent terminals. With the introduction of large-scale models and agent technologies into vehicles, the industry is entering a new phase: "AI-defined vehicles."

MediaTek believes that automobiles are evolving from simple transportation tools into intelligent third spaces that understand users, anticipate their needs, and seamlessly integrate into daily life. MediaTek's automotive platform has already established deep partnerships with over 20 leading global automakers, with more than 190 projects in hand and a cumulative shipment volume of 35 million units, representing a nearly fourfold increase in shipments over the past five years. This demonstrates that MediaTek's entry into the automotive field is not starting from scratch. It has transferred its accumulated experience in performance, energy efficiency, connectivity, imaging, AI, and ecosystem development from its mobile chip platform to the longer lifecycle and higher safety requirements of the automotive environment.

In its Dimensity intelligent cockpit solution, MediaTek divides the capabilities of the future cockpit into several key directions: multimodal interaction, proactive services, concurrent command execution, and edge-cloud collaboration.

This is significantly different from traditional in-vehicle voice assistants. In the past, voice assistants were often passive, performing a fixed task when the user spoke a sentence; while the intelligent cockpit of the future needs to be able to recognize the people in the car, understand the environmental status, judge the user's intentions, and connect services such as navigation, social networking, maps, dining, and payment.

For example, in the case studies presented at the conference, on a weekday morning, after a user gets into the car with their child, the system can automatically identify the passengers and proactively ask if they need to take their child to school before going to the company, and even recommend buying coffee along the way based on the time. The value of this type of scenario lies not in how new a single function is, but in how the in-car system transforms from a "tool" into a "contextual understander".

To achieve this kind of experience, the underlying platform must solve three types of problems.

The first layer is the platform. Vehicles need to efficiently run large-parameter AI models and support multiple models and tasks running in parallel. MediaTek mentioned that the Dimensity flagship cockpit platform can provide up to 400 TOPS of AI computing power and, through a hardware-software co-engineering architecture, compresses AI platform requirements by 90%. Even in a five-screen, heavily rendered scenario, it can still smoothly run two large models at speeds exceeding 50 tokens/s.

The second layer is the model layer. A car's lifespan is often 6 to 10 years, but an AI model's iteration cycle may only be a few months. How to ensure vehicles continuously use new models throughout their entire lifespan is a problem that intelligent cockpits must solve. The Dimensity Cockpit 7 series directly integrates the NVIDIA GPU resource library, supporting the migration of new models and algorithms developed based on CUDA to the cockpit platform; the Dimensity Cockpit S series supports the Dimensity AI Development Kit, helping mainstream models and their derivative models adapt more quickly.

Third is the application layer. The Tianji software platform provides tools such as scene mapping, intelligent model loading, edge-cloud collaboration, and system optimization. For example, edge-cloud collaboration allows the edge to preprocess and filter requirements when complex tasks need to be processed in the cloud, uploading only key tokens, thus protecting privacy while reducing cloud costs.

Automotive AI is not simply about bringing your smartphone assistant into the car's infotainment system. The in-car environment is multi-user, multi-modal, multi-screen, and requires high safety standards. It needs to understand both the driver and passengers; provide entertainment while avoiding interference with driving; rely on cloud capabilities while ensuring local responsiveness and privacy. Therefore, the challenges of automotive AI are more complex than those of smartphones and place greater demands on platform-level capabilities.

Gaming Technology: Mobile Devices Are Approaching Console-Level Experiences

Beyond AI, gaming remains an important scenario for the Dimensity platform to showcase its performance and ecosystem capabilities.

In the past, the upgrade of mobile game experience mainly revolved around high frame rates, high image quality, and low power consumption. Now, with the improvement of mobile GPU capabilities and the development of game content towards AAA levels, mobile devices are beginning to pursue more realistic lighting and shadows, more detailed models, lower latency audio, and longer periods of stable operation.

At this conference, MediaTek highlighted Ray Tracing Pipeline, also known as RTP (Real-Time Ray Tracing) technology for mobile devices. Compared to traditional ray tracing solutions, RTP aims to adapt the rendering pipeline across PC and mobile platforms, delivering complex game lighting and shadow effects in real time, including dynamic objects, skeletal animations, and external environment and object reflections.

MediaTek has partnered with Tencent's "Delta Force" project team to develop a new RTP technology solution. Its significance lies in the fact that if the PC rendering pipeline can be migrated more smoothly to mobile platforms, the development cycle and adaptation costs of cross-platform AAA games could potentially be reduced.

Another key focus is virtual geometry technology. MediaTek Dimensity and Unity Engine are deeply integrated with Virtual Geometry, leveraging the GPU rendering capabilities of the Dimensity mobile platform to achieve over 1 billion triangles in a mobile environment, and continuously outputting a full hour of frame rate at 1.5K high resolution. This means that the limitations of mobile games in terms of model detail have the potential to be further overcome.

In terms of audio, Dimensity LE Audio low-latency technology, through end-to-end optimization, delivers a low-latency Bluetooth stereo performance of 32ms on the Dimensity flagship mobile platform. This technology has already been implemented in the Peacekeeper Elite test server to reduce audio latency. For competitive games, audio latency is not a minor difference at the perceptual level, but can directly affect a player's judgment of enemy positions and operational rhythm.

In addition, the Dimensity platform also showcased GPU Dynamic Cache, Dimensity Frame Multiplication Technology 3.0, Adaptive Adjustment Technology 5.0, and Dimensity Profiler 2.0, a one-stop analysis and optimization tool for Android game developers.

The GPU Dynamic Cache architecture allows the GPU to simultaneously schedule system cache and memory, enabling developers to transfer critical game data through the system cache, saving bandwidth and reducing power consumption. Currently, this technology has been integrated into games such as *Crossfire: Future* and *Dark Zone*.

The Dimensity Frame Multiplier Technology 3.0 adds options such as Depth, which can better predict and generate high-quality virtual frames, supporting 165 frames per second and 144 frames per second. It also supports integration with engine plugins such as UE and Unity, and covers mobile, tablet and cockpit platforms. "Honor of Kings" can unlock a 144-frame low-power experience with this technology, while "Arknights: The End" also achieves higher smoothness and lower power consumption.

The Adaptive Control Technology 5.0 adds intelligent frame control and scene prediction functions, which can make the information flow between the chip, game, and screen more refined. Taking "Mingchao" as an example, after cooperation, there are significant optimizations in the 1% low frame rate and power consumption indicators; games such as "Peacekeeper Elite" can also reduce CPU and GPU load in heavy-load scenarios.

These technologies collectively demonstrate that mobile game optimization is no longer simply a matter of chip manufacturers unilaterally pushing for higher performance, nor is it just game developers unilaterally compressing image quality. Instead, it involves increasingly deeper collaboration between software and hardware. Chips, engines, game content, optimization tools, and development processes are all being re-integrated.

MediaTek is also introducing edge AI into gaming scenarios. At the conference, they announced the results of their collaboration between Dimensity AI Play and the game "Delta Force." Leveraging the edge AI of the Dimensity mobile platform, the in-game CC voice assistant responded much faster, with latency reduced by 56.7% compared to the cloud. This means that AI in games can play more than just NPCs or generated storylines; it can also provide more immediate experiences such as voice interaction, real-time companionship, and tactical assistance.

In the past, discussions about mobile chips primarily revolved around CPUs, GPUs, NPUs, image ISPs, manufacturing processes, and power consumption. However, with the trends of AI agents and multi-terminal collaboration, competition among chip platforms is becoming more complex. A platform needs not only sufficiently strong underlying computing power but also development kits, model tools, cross-platform capabilities, system interfaces, and partner networks.

This is why MediaTek repeatedly emphasized developers and the ecosystem at the conference.

The implementation of AI agents will not be solely the responsibility of chip manufacturers. System manufacturers need to turn agents into native capabilities, application developers need to re-intelligentize functions, terminal manufacturers need to solve cross-device flow issues, car manufacturers need to integrate AI with in-vehicle sensors, cockpit systems, and cloud services, and game developers need to find a new balance between high image quality, high frame rate, and low power consumption.

In the future, when users evaluate a mobile phone, a car, or even an IoT device, they may no longer only look at the hardware parameters, but also at whether it can understand itself, can proactively complete tasks, and can seamlessly transition between different scenarios.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

Lotus is going to make a V8 sports car.

This new car is scheduled to debut in 2028, with the internal codename Type 135. In fact, this codename was originally reserved for the all-electric successor to the Emira, but with the cancellation of the all-electric sports car project, Type 135 changed direction, abandoning pure electric and replacing it with a V8 hybrid system.

Many people may not be familiar with the "Lotus + V8" combination. The last Lotus model equipped with a V8 engine dates back to the Esprit, which was discontinued in 2004. Over the past 22 years, this British sports car manufacturer has experienced several ups and downs, but the number of cylinders has remained at four or six.

The reintroduction of a large-displacement V8 engine into the engine bay has sparked considerable discussion within the sports car community.

The new car's design is inspired by the Theory 1 electric concept car released recently. The official release so far has only included a short teaser video and a rear-end teaser image, the most striking feature being the two large exhaust pipes at the rear. This traditional design element of gasoline-powered vehicles stands out in the current wave of new energy vehicles.

Based on recent power figures circulating from various sources, the new car's combined power output is expected to exceed 1,000 horsepower.

The only question is, where did Lotus get this V8?

A rumor circulating online suggests that this sports car might use an engine from Horse Powertrain. This engine manufacturer, a joint venture between Geely and Renault, happens to have the technological reserves to provide a powertrain solution that shares the same lineage as the Lynk & Co V6 engine.

However, there is currently no concrete evidence to support this speculation, and in comparison, the possibility of Lotus continuing its partnership with Mercedes-AMG is clearly greater.

After all, the current Emira already uses a 2.0T engine provided by AMG. Deepening cooperation along this line of relationships and directly adopting the M177 4.0-liter twin-turbocharged V8 engine that AMG supplies to brands such as Aston Martin would be a more secure business path.

As luck would have it, Mercedes-Benz recently underwent a technical upgrade to this machine.

They replaced it with a flat-plane crankshaft derived from racing technology. In the new Mercedes S-Class sedan, this engine already produces 530 horsepower and 750 Nm of torque. British media outlet Autocar previously reported that AMG engineers are optimizing the M177 for higher-performance models, potentially boosting its base power to over 650 horsepower.

With such an internal combustion engine as its foundation, coupled with a hybrid system, pushing the combined power to 1000 horsepower is not difficult.

After discussing the engine, netizens quickly turned their attention to the battery.

Regarding the new car's powertrain, some Chinese media outlets have exhibited confusion regarding technical concepts during translation. One report confidently states that, in order to reduce weight, the new sports car "will adopt a hybrid rather than a plug-in hybrid system."

To control weight, this supercar won't use a large 70 kWh battery like the hybrid SUV FOR ME. It's more likely to follow the route of the Ferrari 296 or Lamborghini Temerario, using a small 3 to 7 kWh battery pack to provide instantaneous performance.

However, this small battery solution is still technically classified as a plug-in hybrid.

More sensitive than 1000 horsepower is weight.

When news of the hybrid V8 supercar broke, dissenting voices quickly emerged overseas, with Lotus fans, accustomed to extreme lightweight designs, expressing strong dissatisfaction with the battery.

Dude, you're Lotus, can't you just stick to building lightweight sports cars? Please, just remove the battery.

Similar comments are common on major automotive websites. Enthusiasts worry that the increased vehicle weight will dilute Lotus's pure chassis quality and driving experience, repeating the mistakes of the brand's pure electric models.

Of course, some people also expressed their expectations.

For a long time, Lotus's lack of horsepower has been a point of ridicule in the supercar world. While its initial acceleration is quick, it often lacks power in higher speeds and overall acceleration. A 1000-horsepower V8 sports car fills this gap in Lotus's power range, giving it a competitive edge against Ferrari and Lamborghini.

This V8 supercar is actually based on Lotus's current strategic adjustment called Focus 2030.

Current demand in the pure electric vehicle market has not met industry expectations. High-net-worth consumers, accustomed to the roar of mechanical engines, have consistently lacked the desire to purchase the quiet, all-electric Lotus. Meanwhile, the Euro 7 emission standards, set to be implemented in 2027, are further restricting large-displacement pure gasoline vehicles.

Besides the large V8 engine, Lotus's small-displacement all-gasoline models have also been retained—the Emira will continue to use a pure gasoline powertrain, and a mid-cycle refresh will be launched soon. The company states that it will be the most powerful and lightest Emira ever.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.