Is the generative AI bubble about to burst? Plus, Web Summit’s backlash and other tech and AI news this week

Welcome to 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 .

Pitch me the interesting investors, founders, ideas and companies powering emerging technologies like AI. Share your feedback and follow me on LinkedIn for other tech updates.

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

AI has been a glimmer of hope in an otherwise lackluster tech landscape in 2023, with a public market downturn and a tough fundraising environment in the private markets. But could the tide be turning?

“You’ve started to see cracks in the whole generative AI hype train in the past few months,” FirstMark investor Matt Turck told me in the latest edition of VC Wednesdays this week. “It’s a clear case of Amara’s Law, where people have overestimated what a specific technology can do in the short term and underestimated what it can do in the long term.”

He’s not the only one suddenly casting doubts on generative AI’s promise. A recent report by research firm CCS Insight predicted that generative AI will get a “cold shower” in 2024 as the costs, risks and complexity associated with the technology reach a tipping point. And not long before that, Gartner declared that generative AI is in the "Peak of Inflated Expectations" phase of its Hype Cycle.

These predictions aren’t unfounded, when you consider a range of other factors that are converging to create a perfect storm that could derail – or at least temporarily stall – the generative AI hype train:

• Investors are starting to have second thoughts about the technology , with both deals and deal value slowing in the third quarter of 2023, according to PitchBook data. The research firm counted 101 investments in the months of July, August and September – a marked slowdown from 142 investments in April, May and June.
• A number of AI startups are facing growing pains as the technology becomes more commoditized and they go head-to-head with larger incumbents with far deeper pockets. AI speech recognition startup Deepgram and AI marketing startup Jasper have recently cut staff and slashed revenue projections, respectively. Even the big players aren’t immune, with Google DeepMind slashing employee costs by 39% as revenue continues to fall .
• Running AI models is also getting increasingly hard and more expensive, as soaring demand for computing infrastructure leads to an acute shortage of GPUs – the specialized server chips required to run AI software – which is prompting players including OpenAI and 亚马逊 to invest in developing their own alternatives . These computing resources also raise sustainability questions.
• Plus, there is a growing chorus of risks around the potential regulation and other societal implications of the technology, including copyright violation concerns , inherent bias in training data, data privacy and burgeoning AI-generated spam content and misinformation.

Amid these concerns, generative AI startups that mushroomed after ChatGPT’s overnight success – what investors have termed "shiny wrappers'' – are learning that building successful, sustainable businesses is not an easy task. Such companies will churn out of the system in the months ahead, wrote Glasswing Ventures investor Rudina Seseri .

“Does this company/project have a genuinely unique application of generative AI or will OpenAI be able to create another extension/plugin and completely upend your business model?” wrote Mish Adelanwa , a market intelligence analyst at Otravez Capital .

A related issue is how hard it is to prove the return on investment, or the value that these applications deliver. Businesses that have jumped on the generative AI bandwagon without pin-pointing use cases that are profitable for all stakeholders won't succeed, wrote Mike Bryant , co-managing partner at KNOX Capital .

And while things may change over time, at this point the current costs of investment – both human and equipment – to train new LLM models isn’t offset by a measurable productivity gain, wrote Shawn Rosemarin , VP of customer engineering R&D at Pure Storage .

Scaling is another beast, as deploying AI technology brings added operational costs of APIs, compute, storage and data engineering, in addition to the standard SaaS software costs, wrote Wayne Boulais , co-founder and managing director at Tensility Venture Partners .

“What we see in the market is the difficulty in commercializing this valuable intelligence at scale,” wrote Ryan Johnson, chief product officer at lead intelligence platform CallRail .

Given these speed bumps, tech giants like 微软 (LinkedIn’s parent), Meta and 谷歌 will continue to have an edge over startups that don't prioritize profitability, wrote Vin Vashishta , founder of AI consulting firm V Squared. That’s because they “have a clear path to monetization and profitability,” and can afford to acquire “market share at a loss today, because they have an optimization strategy to reduce training and inference costs in the next 12 months.”

That said, startups that can crack delivering generative AI products for specific domains and use cases, as well as a way to employ smaller and lower-cost LLMs, could still come out on the other side, Vashishta said.

Other experts, including Hari Narayanaswamy Krishnan , engagement director at 塔塔咨询服务公司 , agreed.

“We shouldn't be surprised to see enterprise or industry specific not-so-large models come into play to support value-driven use cases (higher accuracy; less bias; limited use case horizon),” he wrote . “I am thinking of all that data sitting in enterprise data lakes, created when that was the hype for some time, but never got tapped.”

Ultimately, the consensus was that AI is here to stay – perhaps in a different shape or form than the applications seen thus far.

“Bursting of this AI bubble will occur, just as the internet bubble in 2000; however, in both cases the technology will ‘change everything’ and will generate outsized returns in the long run for thoughtful investors and companies,” wrote Scott Kosch , managing partner at Kosch Capital .

“We are entering the next phase of the cycle where companies and investors must focus on real world applications of AI to reduce costs and accelerate growth opportunities," Kosch continued. “It is possible to hold the beliefs that this is a ‘hype bubble’ and ‘a generational opportunity’ at the same time.”

What are your biggest concerns around generative AI, and how are you preparing to mitigate these risks? Join the discussion here .

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

• The U.S. is restricting Nvidia chip exports. The Biden administration is limiting exports to China of advanced artificial intelligence chips that could have military applications , like 英伟达 ’s A800 and H800 semiconductors. Semiconductors has been a hot sector this year, with Nvidia the top performer in the S&P 500. The company said it doesn't see any meaningful impact on its immediate financial results.
• Speaking of chips, global chipmakers are facing a new threat. The Israel-Hamas war is threatening to add more instability to the sector, with 英特尔 , Nvidia and 苹果 among major firms with a chipmaking footprint in Israel. The nation has also regularly produced "semiconductor startups the big companies often want to acquire." One of the most immediate concerns is the industry's workforce: Israel has called up 360,000 reservists for the war. It could also threaten tech fundraising that has already been hit by a global slowdown.
• Blowback from the Israel-Hamas war has also spread to Europe’s biggest tech conference, scheduled for next month. A growing roster of tech companies and entrepreneurs are withdrawing from Web Summit after its founder, Paddy Cosgrave , suggested in a post on X that Israel was guilty of war crimes in its response to Hamas' terrorist attacks. The comments were slammed by prominent tech figures including Y?Combinator president and CEO Garry Tan , former Meta and PayPal executive David Marcus and adtech company Taboola CEO Adam Singolda , some of whom said they would never work with Web Summit again.
• Senate Majority Leader Chuck Schumer's next forum on AI will take place Oct. 24 and focus on innovation, Axios reported . The guest list spans AI startups like SeedAI and Cohere , and VCs like Andreessen Horowitz and Kleiner Perkins , including heavy hitters like Marc Andreessen, Steve Case and Stripe CEO Patrick Collison . Meanwhile, New York City Mayor ERIC ADAMS and Chief Technology Officer Matthew Fraser released a plan for responsible AI use among city agencies.
• Speaking of Marc Andreessen, the billionaire venture capitalist posted a follow-up to his June essay "Why AI Will Save the World '' titled "The Techno-Optimist Manifesto " this week. In the write-up, Andreessen denounced efforts to regulate AI and argued that because AI might be used to save lives, any pause on development that limits preventable deaths "is a form of murder." Incidentally, he makes no mention of the social and economic harms linked to various technologies, including AI.
• ChatGPT maker OpenAI pulled the plug on the launch of a new AI model called “Arrakis” that it had hoped would be on par with GPT-4 and could run more cheaply and efficiently, The Information reported . Since the failure of Arrakis, the company has pivoted to developing a version of GPT-4 designed specifically to generate responses more quickly.
• Turns out, AI large language models like GPT-4 or PaLM 2 are not very transparent. A new study based on publicly available information ranked the LLMs of 10 companies including OpenAI and Google and found that their developers have released virtually no information about the real-world impact of their systems, the labor used to build datasets and the computational resources needed to train models.
• Google said last week that it will take on the intellectual property legal risk associated with using many of its generative AI tools, following similar moves by Microsoft and Adobe – as copyright issues around AI become an increasing concern .
• Here’s a roundup of a slew of new AI tools and updates over the past week: Facebook parent Meta just announced Image Decoder, a new application that translates brain activity into images of what the subject is looking at or thinking of nearly in real time. | Amazon is revamping its warehouses with new artificial intelligence and robotics technology. The system, dubbed Sequoia, can identify and store inventory up to 75% faster and cut fulfillment time up to 25%, according to the company . | CapCut, TikTok parent 字节跳动 ’s video editor, is rolling out AI-generated ad scripts and product presenters. | Google’s AI search engine, or "Search Generative Experience" (SGE), now also comes with image generation for those who are opted in. Here’s how to use it , via Allie K. Miller . | Chinese tech giant 百度 rolled out a new version of its AI chatbot, Ernie, which founder Robin Li declared can give OpenAI’s GPT-4 a run for its money. | Swedish buy-now-pay-later company Klarna has rolled out new features, including an AI-powered image-search tool .

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

• Elon Musk’s X announced this week that it will begin testing a subscription model charging $1 in annual fees for new users, Fortune first reported . New users in New Zealand and the Philippines will have to pay the fee in order to post content, like posts, reply, repost or quote other accounts' posts. On the other hand, Tesla saw its profit sink after price cuts, even as it notched a rise in revenue compared with last year thanks to more vehicle deliveries.
• Hundreds of thousands of taxpayers may be able to e-file this year without paying for tax-prep software or services. The Internal Revenue Service said Tuesday it will send select taxpayers in 13 states an invitation to test its new Direct File platform as early as mid-February.
• Netflix 's crackdown on password-sharing has paid dividends again — and in an unexpectedly robust way in its third quarter earnings . The streaming giant is also expanding its push into video games , which includes leveraging video game studios it has acquired over the past two years. It will launch mobile games based on hits like "Squid Game" and "Wednesday" within several months, The Wall Street Journal reported .
• ICYMI: LinkedIn parent Microsoft closed its $69 billion deal for video game maker Activision Blizzard after it was approved by the UK competition watchdog Competition and Markets Authority. The U.S. Federal Trade Commission , meanwhile, is moving forward with an administrative proceeding on the deal, with a spokesperson calling it "a threat to competition."
• Mass layoffs have continued to take their toll across industries in 2023, and hit home this week as LinkedIn eliminated approximately 668 roles across the company . Roles were impacted in the engineering, product, talent and finance arms and are equal to roughly 3% of the company's global workforce. Other companies with layoffs this week included Stack Overflow and 高通 .

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.

• New York Life has hired Northwestern Mutual ’s former chief data officer, Don Vu , to lead a newly formed AI and data team.
• Canada’s 加拿大西安大略大学 has appointed Mark Daley as its Chief AI Officer – the first role of its kind – tasked in part with advocating on ethical AI use.
• VC firm Primary Venture Partners has hired Etan Grosinger as a go-to-market operator in residence focused on building relationships with experts and portfolio companies.
• General Catalyst is joining forces with European seed fund La Famiglia , the VC firm’s CEO and managing director Hemant Taneja posted on LinkedIn.
• New York-based VC firm Greycroft has cut five investors after missing its fundraising target, The Information reported .

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 .