Reborn in 2010, the AI ​​in my mind is too powerful.

On May 15, the annual Google I/O developer conference began at the Moscone Center in San Francisco.

Although Google is not in a particularly good situation right now, it is still a giant, and the scale of this conference is still very large.

There weren't any particularly exciting highlights in the first two days. The first day's content was about Android, and all the senior executives at Google admitted that Google was facing a complete defeat in the mobile field.

However, they wanted developers to trust Google's commitment to mobile, so they showcased Google's own phone hardware, the Pixel 1 and the Moto X.

Unlike the many branded phones that Google has previously collaborated with HTC on, these two phones can be considered Google's own creations. The Pixel is a phone that Google designed and developed entirely on its own, while the Moto X was designed by Motorola, a subsidiary of Google.

The Pixel has relatively basic specifications, targets a younger demographic, and offers a wide range of colors for its casing.

The Moto X, on the other hand, is clearly a more targeted design.

Although they can't compete with Apple and Samsung in terms of hardware specifications, they've put all their effort into the user interface and designed a brand-new smart assistant, hoping to achieve a breakthrough in hardware.

These things had basically been leaked before the meeting, so no one found them particularly amazing.

The main focus of the second day was on Chrome, the absolute leader in the Google camp and the strongest competitive advantage it has against Apple.

This time, there weren't many surprises, but several new Chromebooks were released, all with built-in ChromeOS, emphasizing affordability, ease of management, and suitability for school purchases.

Because the underlying logic of the system is designed based on the Chrome browser, it is difficult to call a Chromebook a computer; it is more like a large mobile phone.

If playing games on an Apple computer is already disastrous, then playing games on a Chromebook is downright hellish. Forget about AAA titles, you can't even play casual games, and you have to rely entirely on immature cloud gaming to keep things going.

Various large-scale professional engineering software programs are completely unusable, and even Photoshop cannot be used.

Therefore, their advertising is basically aimed at people who do light office work, such as making documents, sending emails, and holding video conferences.

This group is actually quite large, and they are now mostly Apple customers. Google clearly hopes to poach some of them.

The announcements over the past two days have been rather uninteresting. The most exciting thing for everyone was the $9.99 music subscription service. This kind of monthly music subscription product is still quite trendy.

However, the reporters were not too disappointed; they knew the surprise was at the very end.

On May 17, the last day of the conference, Google's arrangements were quite peculiar, with no mini-presentations or other networking events scheduled for the entire day.

There will be only one very long keynote address throughout the day, from 9 a.m. to 12 p.m. Google officials said that this was deliberately scheduled because important projects deserve a super presentation.

On the day of the press conference, Google CEO Larry Page took the stage and, without saying much, immediately began to play the victim.

This is a brand-new strategy that Google's strategy team hastily set up just two weeks ago, which has hurt the pride of Google's somewhat arrogant senior management.

There was nothing they could do about it. A few months ago, they planned to directly accuse Chen Yiran of monopolistic practices on stage, especially targeting the EDA sector. After Aurora Group monopolized chip EDA, it also owned wafer fabs, which was indeed very unfair to other companies.

But now Google doesn't dare to. They feel that if they speak ill of Chen Yiran on stage now, they might be drowned in spittle by more than ten million people in six states.

In the short term, they dared not make any accusations against Chen Yiran on a human or moral level, so they could only change their strategy and try to portray themselves as a poor student who was striving to improve himself and fighting against genius through his own efforts.

Larry Page had recently been experiencing some health problems, with damaged vocal cords and a very hoarse voice, which made him look pitiful and garnered sympathy from many reporters at the scene.

After an hour of dithering, Larry Page, having laid the groundwork, finally revealed Google's next steps.

"I know that our technology is no longer leading the way, and the emergence of geniuses leaves us ordinary people at a loss. But we don't want to give up. We hope to prove with our efforts that ordinary people are also qualified to reach the pinnacle of technology."

Therefore, we will not abandon our own deep learning framework, and we will learn from the brilliant thinking of those geniuses.

I am now announcing the official launch of Google's Maze EDA. We are now officially entering the EDA industry, and we are already using our Transformer to revolutionize EDA.

With the help of AI with deep learning, we've made EDA easier to use and more aligned with the needs of chip designers. And I'm announcing that we've reached strategic partnerships with NVIDIA and TSMC.

In the coming year, with the help of Maze EDA, Nvidia will complete the design of a brand-new AI-dedicated chip, which will eventually be manufactured by TSMC.

Next, let's invite Nvidia's CEO, Mr. Huang, to come up on stage and talk about Nvidia's ambitious goals for the future.

The reporters in the audience began to applaud enthusiastically. Regardless of whether it succeeds or not, Google's current actions are quite forceful; it's still struggling and not waiting to die.

Larry Page hugged the yellow-clad man who walked onto the stage with a smile, then slipped backstage, leaving the stage entirely to the yellow-clad man.

Instead of directly stating the plan, the man in the yellow jacket first showed several slides, explaining the role of tensor processing units in future AI chips.

"Nvidia was the pioneer of general-purpose graphics cards. Now we will add a separate tensor processing unit to the graphics card. Our graphics card will reach the state-of-the-art level in both training and inference."

We believe in Google's capabilities. Our company is already experimenting with Maze EDA and is actively communicating with TSMC. All three parties will work together to strive to launch the next generation of graphics cards in September.

In addition, CUDA will be comprehensively optimized based on Transformer, and I hope that all developers will continue to support us.

For the next hour, Google seemed to be going along with its guests, not saying anything new and giving Nvidia the time.

It was mostly about promoting CUDA and demonstrating some optimizations to Transformer.

Essentially, this is a public announcement that the partnership with Chen Yiran has broken down, and DeepFlow is no longer a deep partner of NVIDIA. NVIDIA will prioritize supporting Transformers, and hopes everyone will choose them.

On the day the Google conference ended, major tech media outlets weren't very impressed. Everyone's not stupid; Transformer isn't a new technology, and everyone knows that, from a fundamental logic perspective, it's not on the same level as DeepFlow.

To put it simply, the two sides are using computing power at different levels. For example, Deepflow can learn that one plus one equals two using only one kilowatt-hour of electricity, while Transformer may need two kilowatt-hours.

Moreover, this effect becomes increasingly larger as the data becomes larger and the level of deep learning becomes higher. To achieve the same effect as DeepFlow, Transformer requires several times or even ten times the computing power.

Most media outlets viewed Google's actions as a desperate gamble. Everyone knows that while Oracle's all-in-one PCs are excellent, delivery dates are now two months in advance, and the world knows that Mocha chips are good but hard to find.

Google wants to use this to gain some basic market share, survive in this market, build a small ecosystem, and not be forgotten by developers.

But things didn't go as planned, and Google's presentation greatly spurred several companies on.

AMD, GlobalFoundries, and UMC from Taiwan contacted Aurora Group that same day, all expressing their willingness to undertake larger tasks and their desire for more cooperation with Aurora Group.