The hottest investment area in AI right now, the race to make AI more mainstream and other tech 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. It’s created by Tech Editor Tanya Dua and you can check out our previous editions here. 

First: Catch up on the latest edition of our weekly Q&A series, VC Wednesdays, featuring solo investor Rex Salisbury of Cambrian. And check out the bonus video, where he shares when’s the right time to quit something, here.

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

Last week, NVIDIA and Capital One backed cloud data analytics startup Databricks in a new funding round in which the company raised more than $500 million at a $43 billion valuation. It wasn’t Nvidia’s first rodeo, but a notable addition to the chip maker's growing roster of AI infrastructure startups.

Nvidia isn’t alone. As AI adoption matures and enterprises go all-in on the technology, AI infrastructure startups – the tools and platforms that actually power AI models – have become hot areas of investment, amassing 70% ($11.3 billion) of the overall funding in the generative space between the third quarter of 2022 and the second quarter of 2023, per CB Insights.

“The easiest way to think about it is all the picks and shovels that are helping the AI ecosystem develop such that we can take these applications into the enterprise,” said Elliott Robinson, partner at VC firm Bessemer Venture Partners. “I’m seeing more and more competing term sheets for these kinds of startups.”

A major driver is that as companies of all stripes scramble to figure out their AI strategies, they are facing infrastructural challenges like what data to use and how to organize it, how to build and deploy AI models safely, and how to scale efficiently, said George Mathew, managing director at Insight Partners.

“There’s a tremendous amount of pressure in the market for companies to figure out how to enable themselves from an AI standpoint – and that's coming totally from the CEO and the board level,” Mathew said. “To get AI applications and foundation models into production, you have to have a number of key capabilities figured out.”

Investors who flocked to AI startups after the overnight success of ChatGPT are also learning that converting buzzy startups into successful businesses is not easy, said Rudina Seseri, founder and managing partner of Glasswing Ventures.

For one, it’s hard to compete with the “oligopoly” of foundational or large language model (LLM) providers like OpenAI and Cohere, which are backed by deep-pocketed incumbents. And second, investors are moving away from AI applications like chatbots because of the realization that they are neither differentiated nor have a defensible moat. If all you’re doing is building a generative AI application on top of an LLM, bigger players could easily mimic it, said Seseri.

“The middle layer is where the opportunity is,” she said.

That’s not all. AI infrastructure is an umbrella term that’s broad in scope. AI models are trained on reams of data that enable them to get increasingly efficient in making predictions, or what’s called inference. Everything that aids this process – from hardware like chips and GPUs, software platforms that store data and secure data, tools that help fine-tune models and even startups that evaluate the best LLMs to use – fits into the bucket of AI infrastructure, investors said.

“It’s ever-evolving, but it’s essentially a compilation of tooling systems and platforms that enable engineers and data scientists to build and deploy AI,” said Madison Hawkinson, an investor at Costanoa Ventures.

To be sure, so-called MLOps within AI infrastructure, which focuses on streamlining the process of taking machine learning models to production, has existed for a while, said investors including Kanu Gulati, partner at Khosla Ventures. But LLMs are a different beast given their size, and they bring a new set of challenges when it comes to training and evaluating them, she said.

One area that Hawkinson said she’s paying attention to is startups that help companies convert their messy, unstructured data into structured data that can be used for AI models, like unstructured.io. Another set is those startups that enable enterprises to easily collect, transform and analyze their data in real time – without which they wouldn’t be able to build any AI applications. Feldera, which Costanoa just invested $6 million into last week, is an example.

Like was the case with the cloud revolution, another buzzy subset of startups include those that remove adoption barriers by focusing on security, compliance and data privacy, said Glasswing’s Seseri. An example is Allure Security, which aims to tackle bad actors who use generative AI to spoof genuine websites and carry out scams.

Perhaps the most visible group of startups are those on the hardware side that are trying to tackle the ongoing GPU shortage, including CoreWeave, Lambda Labs, Foundry Technologies and Together AI, said Bessemer’s Robinson.

“Right now there's a huge imbalance between the supply and demand of AI in terms of compute and chips – and a huge opportunity for startups,” he said. “Because if you are Microsoft Azure or Amazon AWS, how likely are you to totally restructure your data centers for AI-specific workflows when you have a P&L to manage?”

What areas of AI infrastructure are you paying attention to? Let us know in the comments.

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

• The race to make AI even more mainstream heated up this week:

  ChatGPT now has the ability to produce custom images, not just text. Parent company OpenAI unveiled a new version of its DALL-E image generator Wednesday that will be incorporated into ChatGPT for paid users starting next month.

  Meanwhile, LinkedIn parent company Microsoft is introducing a unified AI assistant across its suite of popular products. While CoPilot AI already exists in various iterations across the Edge browser, Microsoft 365 platform and Windows, it will now be a single entity across Microsoft's products and devices, noted Engadget — from Powerpoint to Teams. It also comes with a slew of new features.

  Plus, Alexa, Amazon's omnipresent voice assistant, is becoming more conversational. The company is leaning into generative AI and tapping a new large language model for "near human-like interactions," said David Limp, senior vice president of devices and services.

  And Google rolled out some major updates to its AI assistant Bard, hooking it up with services including Gmail and YouTube. Bard can now act more like a personal assistant to users by answering questions about their email and other documents, and its answers on any given topic can also be double-checked through Google’s search function.

• Red teaming, where companies get ethical hackers to infiltrate their systems in order to learn how to better safeguard them, has become all the rage in AI. OpenAI has launched an open call for its Red Teaming Network as it looks to more robustly evaluate its generative AI tech's risks, TechCrunch reported.

• Morgan Stanley is the latest example of companies embracing AI. The financial services giant is incorporating ChatGPT into its financial advisors' workflows with a tool called AI@Morgan Stanley Assistant, which accepts full-sentence prompts and searches roughly 100,000 reports and documents. Meanwhile, Airbnb is using AI to crack down on potential partiers.

• AI startups continue to raise funding at a time when the tech startup landscape has dimmed overall. Here’s a rundown of some of the key ones that caught our eye this week:

  Stealth startup Essential AI, founded by two ex-Googlers behind the seminal  2017 paper “Attention Is All You Need,” has raised $40 million in fresh funding, according to Bloomberg.

  Writer, which helps businesses like L’Oréal and Uber write and summarize content, has raised $100 million in a deal led by ICONIQ Growth, with participation from investors including WndrCo, Balderton Capital and Insight Partners.

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

• Instacart, Klaviyo and Arm have all IPOed in recent days, in what could suggest that the IPO market is finally thawing. While Arm got off to a good start and Klaviyo also popped in its debut, the hype around Instacart's IPO may already be fizzling out. The grocery-delivery app kicked off its second day on the Nasdaq trading at just over $32 a share — only a little above its IPO price and well below its first-day highs. Relatedly, check out the 14 Instacart alumni who have founded their own startups, per Insider.

• Elon Musk wants to charge users of X, formerly Twitter, to use the app. He said that introducing a subscription service would drastically slow the “vast armies of bots,” as operators would need to individually pay for each account. X is currently free for most users, though it has a subscription service that some users and brands pay for.

• A new smartphone from Huawei Technologies could threaten Apple’s dominance in China, according to CNBC. The phone, called the Mate 60 Pro, has a semiconductor chip that’s apparently 5G capable — which, if true, represents either a breakthrough achievement or a violation of U.S. trade restrictions.

• Many new MBA grads want to buy a business rather than join or start one. That's thanks to an increasingly popular strategy known as entrepreneurship through acquisition, which argues that would-be business owners are better off purchasing an existing company than trying to raise venture capital for a startup.

• Cisco on Thursday announced an agreement to buy cybersecurity and observability company Splunk for about $28 billion — the networking giant's largest-ever acquisition.

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.

• Microsoft product chief Panos Panay, who headed up the hardware and Windows divisions, announced that he was exiting the company this week. He will be joining Amazon, where he’ll replace retiring devices chief Dave Limp, per Bloomberg.

• The U.S. Department of Homeland Security has appointed Eric Hysen as its first Chief AI Officer and Matthew F. Ferraro, formerly counsel at WilmerHale, as senior counselor for cybersecurity and emerging technology.

• Kate Goodall, co-founder and CEO of D.C. accelerator Halcyon, is stepping down from her role to head up a new investment fund.

• Mayfield Fund has hired Sri Pangulur, who formerly led the enterprise software investing practice at Tribe Capital, as a partner.

• Spark Capital has hired Natalie Vais of Amplify Partners as a general partner. The VC firm has also promoted Clay Fisher and Fraser Kelton to general partners.

Thanks for reading. 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.