The 2018 bet that paid off for Nvidia, and other tech and AI news this week

Welcome back to LinkedIn News Tech Stack, which brings you news, insights and trends involving the founders, investors and companies on the cutting edge of emerging technology.

First: This week Tom Chi from climate-tech VC firm At One Ventures joined us for VC Wednesdays, sharing his investment thesis and what he looks for in startups.?

Pitch me the investors, founders, ideas and companies powering cutting-edge technology like AI. Share your feedback by sending me an InMail. Follow me on LinkedIn for other tech updates.

Zooming In

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

If there’s one company that’s already reaped huge dividends from the rise of generative AI, it’s 英伟达 .?

That’s the result of a tactical bet the chipmaker made five years ago that has paid off, CEO Jensen Huang said during a keynote at the SIGGRAPH Conference in Los Angeles this week.

In 2018, the company realized that the predominant way of rendering 3D graphics through the technique of rasterization was reaching its limits, Huang said. Nvidia took a bet on AI as the future of computer graphics, rolling out its visual computing platform RTX.

Reinventing computer graphics through AI required that the company retool not only its existing hardware, software and algorithms, but also its graphics processing units, or GPUs — the tiny server chips that enable AI software to run. That led to the launch of Nvidia’s GPU architecture Turing, which has since evolved, said Huang.

“It was a ‘bet the company’ moment — we had to bring together computer graphics and artificial intelligence for the very first time,” he said, calling it “a giant risk” without which much of the current progress in AI would not be possible.

Since then, Nvidia has emerged as the sought-after supplier of the picks and shovels of the AI gold rush, with demand for its GPUs like the H100 going through the roof — propelling the Silicon Valley-based company to the trillion-dollar club and even leading to a mass shortage of GPUs.

Huang is well aware of this position, and he used the keynote to tout the company’s next-generation version of its GH200 Grace Hopper Superchip — which Nvidia says offers 1.7 times the memory capacity and 1.5 times the memory bandwidth of its current H100 chip.

The new offering could take “just about any large language model” and boost its “inference like crazy,” he said, meaning it would rapidly improve the ability of LLMs to generate text, make predictions or identify objects inside images.

Other highlights included Huang’s announcement that Nvidia had partnered with the startup Hugging Face to help developers train advanced AI models, and his introduction of AI Workbench, a new platform geared toward allowing developers to experiment with generative AI applications on desktop before scaling them to a data center and public cloud.

Here’s more chatter around Nvidia on LinkedIn this week.

This Week in AI

Here’s where we bring you up-to-speed with the latest advancements from the world of AI.

• Zoom faced considerable blowback this week for recent changes to its terms of service policy, in which it said it reserves the right to train AI on user data — with no mention of a way to opt out. The backlash was so severe that CEO Eric S. Yuan called the change "a process failure internally that we will fix." Yuan wrote in a LinkedIn post that the company "would absolutely never train AI models with customers' content without getting their explicit consent.”
• ChatGPT has gotten a slew of new updates. Highlights include Prompt Examples, where users will start seeing various ways in which they can test the chatbot, and Suggested Replies, where ChatGPT will start suggesting ways to continue the chat after a user’s initial prompt. In related news, OpenAI has also launched a web crawler to improve its AI models. Called GPTBot, it combs through the internet to train and enhance AI's capabilities — but users can opt out if they want.
• 华特迪士尼公司 is the latest company to jump on the AI bandwagon, forming a task force to determine how it can best use artificial intelligence in its work, Reuters reports, citing unnamed sources. That may involve developing AI applications in-house, as well as working with outside startups. The use of AI has been a driving issue behind the ongoing Hollywood strikes amid concerns that the technology could replace human jobs.
• AI-generated travel guides are apparently proliferating on 亚马逊 , The New York Times writes. These self-published travel guides — often less-expensive alternatives to well-known Fodor’s and Lonely Planet versions — are often of dubious quality and “bolstered by sham reviews.”
• 微软 (LinkedIn’s parent) is shutting down its digital assistant app Cortana this month. The move is not surprising given the company’s focus on new AI features like its ChatGPT-powered Bing Chat and other productivity features across Windows and the web browser Edge.
• 谷歌 and Universal Music Group are discussing a potential deal to license artists' melodies and voices for AI-generated songs, the Financial Times reports. The deal aims to tackle the growing issue of "deepfakes," or songs that mimic the voices of real artists without their permission. The deal would allow fans to create the songs legitimately, after paying the copyright owners for their use, and artists would be able to choose whether to opt in.
• As companies rush to build AI-based applications of their own, vector databases, which enable the storage and retrieval of data to power AI models, have become increasingly in-demand. Kyle Poyar , operating partner at VC firm OpenView , breaks down the growth of one startup in this space, Pinecone, in this post.

TechTok?

Catch up on the tech headlines you may have missed this week and what our members are saying about them on LinkedIn.

• President Joe Biden signed an executive order late Wednesday blocking and regulating some U.S. high-tech investments in China. The order — which comes after nearly two years of deliberations and anticipation from the business community — is targeted at advanced computer chips, micro electronics, quantum information technologies and artificial intelligence.
• Onetime co-working darling WeWork has issued a dire warning: that there is "substantial doubt" its business will be viable in the long term, with the next 12 months being key to its future. In second-quarter earnings details released Tuesday, the company laid out its financial strategy, saying it will focus on bringing rental costs down and negotiating better leases, while trying to increase revenue and raise capital over the next year. Its shares plunged 33% in extended trading after the warning, valuing it at $166 million. At its peak, the company was valued at $47 billion.

• A proposed $5 billion class-action privacy suit against 谷歌 will move forward after a federal judge spurned the company's request for summary judgment. The suit alleges that Google is tracking users' online activity in Chrome's incognito mode and other private browsing modes without consent, while Google disputes the claims.
• Meanwhile, Tesla is being sued for false advertising by three owners in California, who claim their cars fell well short of meeting range estimates provided by the company.
• And in a third legal battle, the U.S. Supreme Court has granted Apple a reprieve in its fight with Epic Games , which has argued that 苹果 ’s App Store constitutes a monopoly and unfairly taxes developers. A federal appeals court's ruling requires the iPhone maker to let App Store customers make purchases outside of Apple's payment system — a win for Epic — but Justice Elena Kagan has delayed that decision's implementation.
• Speaking of Apple, the company is testing a new M3 Max chip for its most powerful MacBook Pro to date, in a bid to revive its laptop sales, Bloomberg reports. The new chips, which represent the biggest upgrade since the company began replacing Intel chips in its laptops in 2020, are meant to provide both longer battery life and improved performance.
• Remember Meta ’s co-branded smart glasses with Ray-Ban ? Turns out, the company is working on a second-generation version as it looks to move on from the tepid consumer reaction to its first iteration. The tech giant launched Ray-Ban Stories in 2021, allowing users to take photos and listen to music through their glasses, but two years later, less than 10% of the 300,000 units it sold are still actively in use.

Movers and Shakers

Here’s keeping tabs on key executives on the move and other big pivots in the tech industry. Please send me personnel moves within emerging tech.

• TikTok ’s U.S. ecommerce general manager, Sandie Hawkins , is leaving the company, she wrote in a LinkedIn post. TikTok is replacing her with two new ecommerce executives: Nicolas Le Bourgeois , who earlier in his career worked at Amazon, and Marni Levine , a former eBay executive most recently with Meta (not to be confused with Meta's chief business officer Marne Levine ), according to The Information .
• Tesla chief financial officer Zach Kirkhorn has stepped down from his role and will be succeeded by its chief accounting officer, Vaibhav Taneja .
• Salesforce 's chief people officer, Brent Hyder , is leaving the firm, Business Insider reported.
• Bain Capital Ventures has promoted Tina Dimitrova to principal, focusing on fintech.

ICYMI

Here are other top stories of the week from beyond LinkedIn in the broader world of tech.?

• With the Def Con convention underway in Las Vegas this week, The Washington Post takes us inside how a group of ethical hackers are trying to unmask the ways LLMs and AI models work.
• Even with a market cap of $500 billion, very few people know about 台积公司 , or Taiwan Semiconductor Manufacturing Company. The New York Times writes a fascinating profile of its 92-year-old founder, Morris Chang.
• Elon Musk's brain-implant company Neuralink has raised $280 million in a round led by Peter Thiel-backed Founders Fund .

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.