Top execs from Nvidia, Google, Palo Alto Networks and more share their biggest learnings on AI — and predict where it’s headed

Welcome to this special edition of LinkedIn News Tech Stack, which brings you news, insights and trends involving the founders, investors and companies on the cutting edge of technology, by Tech Editor Tanya Dua . You can check out our previous editions here .

Norwest Venture Partners ’ Scott Beechuk joined us for last week’s edition of VC Wednesdays. He shared where the firm plans to deploy its latest $3 billion fund, what he looks for in early-stage startups and why AI agents are overhyped. Catch the full interview here.

LinkedIn is offering a short series of exclusive programming to help tech leaders excel on LinkedIn.

• Expect insider tips, real-time coaching from LinkedIn’s Sr. Tech Community Manager, Iris Mansour , who regularly advises Fortune 500 tech CEOs on how to excel on the platform.
• Register here for LinkedIn’s Series for Tech Leaders — Masterclass 1 on May 16 at 2pm ET.

A deep dive into one big theme or news story every week.

AI is starting to become ubiquitous, with companies across the board stepping on the gas when it comes to the technology. This was also the main theme at #TieCon2024 late last week, a conference hosted by the tech nonprofit TiE Silicon Valley , which is dedicated to fostering entrepreneurship.

Amid this transformational shift, tech heavy-hitters from the likes of 英伟达 CEO Jensen Huang to Palo Alto Networks CEO Nikesh Arora took to the stage for keynotes that centered — unsurprisingly — on AI, sharing their perspectives on the rise of the technology while also reflecting on where they see it going next.

Here are the top takeaways — share your own in the comments below if you attended the conference last week.?

The secret sauce of AI pioneers like Nvidia? An open ecosystem

Nvidia has perhaps emerged as the biggest winner of the AI gold rush, with an ever-increasing demand for its GPUs going through the roof, propelling the Silicon Valley-based company to the trillion-dollar club .

But this was no accident. Rather than merely building new graphics processors like its competitors, Nvidia took a page out of 苹果 and Salesforce ’s playbook — building a full-stack accelerated computing platform and creating an ecosystem around its products. This is what set it apart from its competitors, said CEO Jensen Huang in a fireside chat with Mayfield Fund ’s Navin Chaddha .

“AI is a full-stack problem… It is the reason why we were able to invent new computer graphics algorithms and see it adopted by the industry, because we always were working with game developers, we were always working with scientists, we were working with the ecosystem,” Huang said. “Our perspective wasn't to build a great graphics chip — our perspective was to enable the applications to somehow be better. We had to shape, shift and inspire an entire industry to follow.”

It is this collaborative "co-design" approach that has ensured that Nvidia has now become synonymous with the generative AI boom, or what Huang called “the single biggest technology transition” of our lifetimes. Nvidia’s GPUs, or the specialized server chips that help power compute-intensive AI tasks, can be integrated across different hyperscalers in the market.

“We had the good wisdom of designing our data centers in such a way that they could be disaggregated, taken apart and then integrated into 微软 Azure, Google Cloud and Amazon Web Services (AWS) — so we're in every cloud, we're in PCs, we're everywhere,” he said. “The great achievement of our company is one, of course, this approach of accelerated computing, but two, making it available everywhere.”

Nvidia is not the only company adopting an ecosystem approach. An open-stack infrastructure is what has also enabled Google Cloud to nab over 90% of generative AI unicorns as clients, said Amit Zavery , VP and general manager, Business Application Platform at Google Cloud.

“I've been working on AI for eight to 10 years, and for longer with our research teams inside the company,” he said. “We are working with the community to make sure that we can absorb and bring third-party technologies into our ecosystem. The goal is to really have an ecosystem versus only Google-led [products].”

Move over AGI, we need to start thinking about ASI

Many AI frontrunners including OpenAI and Google DeepMind believe that we will achieve artificial general intelligence, or AGI, in the near future — where AI systems become so advanced that they begin to display some kind of consciousness or sentience like the human brain.

That’s just the tip of the iceberg, according to Sumit Sadana , EVP and chief business officer at Micron Technologies . Instead of AGI, where AI systems display broad human-like intelligence, we should be thinking about ASI, or artificial super intelligence, which he envisions as the next stage after AGI, where AI actually attains self-awareness.

“ASI is that point in time where the collective wisdom of humanity — not just all of the data, knowledge and insights — but all the wisdom gets collected in a single data center with a level of capability that is superhuman in nature,” he said. “We are going to get there — it is only a matter of time, investment and a few breakthroughs that most researchers have a line of sight to, and of course a lot of data.”

Getting to that point requires significantly scaling AI large language models — which require not just processors like the chips Nvidia produces, but also memory capacity and bandwidth in order to run. That’s where Micron comes in.

The company is making considerable investments to boost memory (known as DRAM) and storage (known as NAND Flash), and recently started mass producing its high-bandwidth memory (HBM) semiconductors, which will be integrated within Nvidia's next-generation H200 GPUs, for example. While the latest Nvidia H200 GPUs contain six stacks of HBMs, each with eight die of memory chips, in the future, GPUs will contain 12 die, according to Sadana.

“The DRAM portion of hardware is becoming much bigger and more complex — high-bandwidth memory is the critical resource for the processors,” he said. “Already, the area of memory that is being used is more than seven times the amount of no no? silicon that is there in the GPU.”

AI will require a whole new user experience

When mobile phones emerged on the block, Google was primarily a web-based business. It was hard for anyone at the company to imagine that web-based interfaces would be all but replaced by mobile applications, said Palo Alto Networks CEO Nikesh Arora , who worked at the company back then.?

According to Arora, we’re at a similar crossroads now, and must reimagine what the user experience in the AI era should look like.

“The question I have for people now is…can you imagine the world five years down?” he asked the audience during his keynote. “There is no app itself. It is basically an AI agent.”

AI is quickly becoming part of our daily lives and workflows. Yet, the user experiences around AI applications remain largely unchanged. Even the many AI copilots being rolled out are “all adjunct to your mobile app,” Arora pointed out.

Arora urged designers and product managers to imagine a future when, instead of using a travel aggregator like Expedia Group to browse through and book travel, a person might simply ask an AI agent to book them a flight with a simple instruction.?

“If the AI is smart enough to customize the answer for you, it eliminates 30-40% of the work,” he said. “In five years to a decade from now, the whole concept of product management in tech will change. The whole concept of UI will change. The whole concept of how we get answers will change.”

Thanks for reading. Please share Tech Stack and forward it around if you like it! And if you have any news tips, find me on InMail .