Intel's Startup Chief on Trends in the AI Landscape

Tzahi Weisfeld, head of Intel's startup accelerator, talks to AI Business about what is hot in the land of AI startups

Deborah Yao, Editor

February 5, 2024

6 Min Read
Getty Images

Intel Ignite is the chip giant’s startup accelerator program that focuses on deep tech. It takes in and mentors startup cohorts in the U.S., Europe, U.K., and Israel throughout the year.

Out of 250 to 300 applications per cohort, Intel Ignite said it accepts about 10. The incubator has different deadlines for its cohorts. The Europe cohort is accepting applications with a deadline of Feb. 23 for a start date of April 22 for a 12-week run. The U.K. cohort has a deadline of March 1 for a start date of April 30 and Tel-Aviv is taking in applications until March 20, for a start date of May 7. The U.S. cohort just closed its application window.

Intel said it does not take an equity stake in these startups.

AI Business recently caught up with Tzahi (Zack) Weisfeld, who is the general manager and vice president of Intel Ignite, to find out what he is seeing in the AI startup landscape.

What follows is an edited version of this conversation.

Tell us about Intel Ignite. You have been there since the beginning?

I will take a step back for a second. I have been an entrepreneur most of my life – for about 30 years, eight years in Silicon Valley, the rest here in Israel – starting multiple sizes of companies from bootstrapping, to raising $120 million, selling to Google close to nothing, but also a few successful ones.

Related:Trends to Watch: The Impact of AI on Startups Across Industries in 2024

I founded and started the Microsoft accelerator in Bangalore, Beijing, Shanghai, Berlin, London, Paris, Seattle, Sydney and Tel Aviv. I managed Microsoft for Startups in 110 countries. Then the former CEO of Intel asked me in 2019 to join Intel and build something similar to what I built there, (saying they) have Ignite.

I found that it is actually a model I have been perfecting for the past 14 years or so − how do you reinvent large corporations through the work with startups? Not an easy task. … That has been my life for the past 14 years, how to reinvent big companies by working with startups. Startups matter and they are critical to the success of these large companies, and you need to build the right engagement model for that to be successful. So that is what I do.

Ignite has existed since mid-2018. It started out in Tel Aviv and then expanded to a program in Munich for Europe, we have a program in London for the U.K., and in Boston for the U.S.

What kinds of startups are you looking for in your Spring cohort?

We differentiate ourselves by taking on what we call deep tech startups. What is deep tech? Deep tech is pretty broad, but clearly, hardware silicon is deep tech, computational biology is interesting, quantum computing is interesting, but also when we look at Gen AI, LLM, we are looking at, first, infrastructure companies − companies that accelerate models, that enable AutoML, do stuff that is infrastructure for generative AI.

Related:7 Top AI Companies to Watch in 2024

Then there are companies that are doing something very unique: Very unique IP in the space or very unique data in the space that no one else has, or a different approach. (However, what) we are not looking too much into are all the applications built on top of them.

There is an interesting move from computer vision to LLM. Two years ago, everyone did AI for computer vision for automotive, robotics or whatever. Now everyone has shifted to using their knowledge and talent to deal with LLMs. There are interesting companies we are looking at in security for AI and AI for security. We call it at Intel, Responsible AI – it is such a challenging space. You have no idea if what you are getting back is real or not and how things are being used, what can you use and how many holes are there and backdoors, etc.

There is a space we call Nimble AI. Not everyone needs 170 billion parameters like OpenAI. Even 5% of that is more than enough for most applications. If I need to drive an application for a specific space, I do not need (data about) world history, world geography, every literature; I just do not need it. It is not relevant. What if I am trying something in the health care space?

Related:How Comcast NBCUniversal Lift Labs Leaders Accelerate AI Startups

There are multimodality companies. How do you integrate many sensors? Are you crunching all the data and then multiple outputs? Interesting space for us, Intel has a really strong stance in confidential computing and trusted computing. So anything in regulated markets: health care, fintech, government, Department of Defense − these kinds of applications are super interesting. We were very uniquely positioned in that space.

Computational biology: The world of Pharma is changing in front of our eyes. The ability to get highly targeted drugs, no longer ‘you take one antibiotic for everything.’ The world is heading to ‘you can actually get stuff much more specific’ based on the ability to do these things.

Interesting move to the edge. For us it is super interesting, as I talked about this nimble and small LLM space. Intel has been pushing that AI Everywhere and how you can do a lot of that stuff even on the laptop using the computing power of smaller machines.

Half of the code today has been AI-generated, so how do you deal with that as well? Security, quality, etc. Look, I love my job. It is fascinating. I am in this candy store of so many interesting companies that are writing the future, and we want to be the platform for their future.

Security for AI and AI for security is a big deal. That’s one. The other interesting one I would say is we have seen a lot of demand in the regulated market that they're struggling with leveraging what is available today and they need an environment that is confidential, that data cannot be transferred to anyone else.

We are seeing companies that want to use cloud LLM, not local, but cloud LLM, but in a secure way in fintech, health care, government, etc. I think we are going to see a big group of companies going there. And more edge than what we have seen before because of the size of models that we can now fit on a small CPU, or CPU GPU combination. I think we will see more companies doing stuff at the edge.

Any words of advice to our readers on how do to choose a startup partner?

Generally I try to look at startups (with) a good set of founders. … If they have a really good set of founders, they are going to do something that is interesting and relevant that we need to know. … Stick to the basics and find the right group of people that you believe could drive relevant IP and do not get too married to what they do specifically, but (rather) the way they do it.

Read more about:

ChatGPT / Generative AI

About the Author(s)

Deborah Yao


Deborah Yao runs the day-to-day operations of AI Business. She is a Stanford grad who has worked at Amazon, Wharton School and Associated Press.

Keep up with the ever-evolving AI landscape
Unlock exclusive AI content by subscribing to our newsletter!!

You May Also Like