NVDA 凭借大规模回购以及先进的 AI 安全能力，处于非常有利的地位。

We now turn to Nvidia, which is doing a lot of hard work today with the return of Max, a technical analyst at Creative Strategies.

there are multiple parts to this story. Obviously, you have stock buyback news, AI software security news , and then there's the China angle to all of this. However, many people argue that the first two parts are closely related .

They represent almost the same story with regard to share buybacks and AI security.

I believe that Nvidia's buybacks are a sign of confidence more than anything else. They are making so much money that they can afford to buy back now, with a $25 billion mandate for buybacks, as well as continuing to invest in research and development and data centers at other companies to help grow the AI ecosystem and other major cloud service providers as well.

Overall, I think this shows that they are in a very strong position , which is a good sign for the market. And "Max," to what extent does this concern the employees? Because employee quotas are one of the many ways Nvidia attracts top talent, especially engineering talent for a company like Nvidia.

How important is this in terms of diversifying stocks? I mean, that's the only part of the conversation that's somewhat obvious, but nobody's really talking about making up for that here.

I think that as they start compensating their employees more, these bonuses become bigger as part of the company's overall growth.

As you know, these are not insignificant amounts of bonuses and shares that are awarded as compensation. This does indeed help to alleviate some of the stress related to relief and the like .

It's a relatively small percentage , don't get me wrong , but I think overall, it's important to say, you know, that this isn't just , I know some other companies might call it boosting EPS if you keep revenue the same and you now have this in EPS, it's a 4% difference.

It doesn't amount to much in the overall context, but I think all these little bits combine to form the whole story here, which is that Nvidia is so strong now, it no longer has to worry about diluting its stock, and it can buy back and help its employees more, and use that as a more powerful means of attracting top talent as they seek to grow this open ecosystem.

This has certainly set a minimum price for the stock today, at least . I mean, we know that the valuation has declined in recent quarters. I'm just wondering how much the buyback plan versus what we're hearing about AI security software is influencing the move we're seeing today at Nvidia, especially in light of the broader sell-off in the semiconductor sector .

I don't think much of it comes down to buybacks, but I consider this a strong indicator of strength more than anything else, because ultimately it's a large number overall.

I mean, this would have represented the market value of quite a few of these huge companies 10 years ago, and now it's just a big buyback for them. Overall, I think that this morning they made their announcement about "sandboxing" and some of their AI security solutions , which is a very strong signal for them .

But on the other hand, if AI security is a problem and we don't have solutions for it, Nvidia might have the solution. Will this, as Sam and Daria have repeatedly stated, bring development to a complete halt ?

In general, I don't necessarily think that will happen , but these are questions that people will ask, and they will obviously affect the overall share price and everything related to it.

Max, what do you think of this advertisement? I mean, it wasn't just the launch of a new security program, the " Open Proxy Security Platform". They came out and said things like they were able to stop a "Hugging Face" attack in fractions of a millisecond.

I mean, they definitely have a point of reference here since they own "Hugging Face," so they probably have very detailed information about the attack itself.

But also, as I mentioned, with Dario Amodie and Sam Altman pushing for this slowing down and this deceleration, this presents a completely opposite viewpoint. The safety of artificial intelligence here is an engineering and infrastructure problem, not something to be suppressed.

So, how do you view its place in this broader context?

Alex Coffey noted that he was surprised by the reaction, or lack of reaction, from some cybersecurity companies to this issue. I think that in general, if you look at how the "sandboxing" protection technique actually works, in the " ugging Face" incident for example, OpenAI said that they disabled the protection because part of the test was intended to measure the cyber model's capabilities , and to test it, it had to have full access, which required removing some of those safeguards.

This is something they mentioned as part of the process , and the model did things it shouldn't have done, didn't it ? He escaped the protective environments that were supposed to keep him contained, and that was simply a security flaw, a design flaw in the protective environment itself that allowed this to happen when it should not have been allowed.

These protection environments and some of Nvidia's new "Open Shell" and "Open Agent" platforms are different solutions to this problem.

None of them are 100% guaranteed. I mean, OpenAI assumed that its environment did not contain flaws that might allow this, yet something came up to prove the opposite.

And now Nvidia claims to have a better solution, with more protection and security, which is great . We must have that. Furthermore, "Open Agent" is a good sign in terms of computing as well, because part of it is that while the model is thinking, you have to monitor what it plans to do before it executes, and feed that to another model.

You need to run the inference and make sure that the model's intent is sound before allowing it to take action, which is what will happen inside the "Open Shell" environment.

So , all of this reinforces the idea that we have very strong computing capabilities here. And yes, I could clearly have stopped the "Hugging Face" incident and other incidents after they happened.

But now we also need to add more computing alongside inference to prevent these models from doing things before they happen, correcting faulty behavior and adjusting as part of the inference process and model training in general.

So, I think we'll build on what Molly just said, Max. I mean, what impact do you think this will have on the sector as a whole? Because today's reaction was interesting. I mean, you see all the analysts coming out into the open and saying that they support this news.

This is an industry leader coming out to say that they are facing it head-on . They will deal with it. They will defend him. Things won't slow down. It's about putting in place safeguards.

This is not just for Nvidia. Rather, it is intended for the entire industry.

However, the rest of the semiconductor sector is still in decline. As we mentioned earlier, cybersecurity stocks, which were low before the opening, perhaps as a reaction to this, are now seeing an increase.

Is there something the market is still unconvinced of and waiting to see?

I think it's when the government gets involved in a lot of this, which they don't , or when it goes beyond just saying "Okay, we can promise you that cybersecurity here is good."

The question is: Do we have the confidence of regulatory bodies that this will actually happen? Does everyone believe this is true ? Do we need an external party? I know they talked about the existence of an organization like FINRA for artificial intelligence. Do we need this?

And if you start adding these things, these remain unanswered questions regardless of open systems, better platforms, isolated testing environments, and security for all.

I think for cybersecurity companies , this doesn't really change anything. Let's say you prevented a model from doing something accidental; all of this is accidental. Let's say you prevented a model from accidentally hacking someone in the first place because you put it in an isolated environment.

Okay, let's assume that a malicious party did not do that and directed it directly to hacking. Cybersecurity is still needed . So, I think much of this is just a knee-jerk reaction to the news, but it doesn't change the fundamentals; Any malicious party without an isolated environment will still be able to penetrate something .

Cybersecurity is still needed, and these cybersecurity companies can scale up their operations, and in an isolated environment, let the system be hacked in a similar environment.

You can scale massive hacking attacks against a single company, predict them, protect against them, build cybersecurity controls , and improve overall security before anything happens.

I think overall this all comes down to the basics, which confirms that we still need very strong cybersecurity. We still need really strong protection, but we don't know if all of this has been resolved and promised, and if the government is confident it's true.

It may only take one small incident and it could happen. I know that OpenAI suffered one last weekend. Yes, it's just one thing that sets everything back, and I don't know if we've reached the point where we can truly say it's over and nothing else will happen.

Nobody knows, and I think the market is kind of , you know, holding on to that .