It’s time to explore visionary minds leading the artificial intelligence (AI) revolution, as three pioneering CEOs unveil their strategies for navigating the future of advanced technology.
Straight from the helm of innovation, these leaders share their unique insights and bold predictions, offering a glimpse into how their companies are not just participating in, but actively shaping, the AI landscape. Read on to get the latest, greatest AI insights from the C-suite leaders of Alphabet (GOOG 2.04%) (GOOGL 2.12%), Kulicke & Soffa Industries (KLIC 5.43%), and Cloudflare (NET 19.50%).
This electronics packaging equipment provider remains one of the cheaper ways to play AI growth
Billy Duberstein (Kulicke & Soffa): Investors may think every tech company exposed to powerful artificial intelligence trends has already taken off, but Kulicke & Soffa may be an exception.
For sure, Kulicke & Soffa’s core ball bonding and wedge bonding businesses for high-volume electronics markets are still in the dumps, but they may be reaching a bottom. These legacy core segments make up a majority of the business and are still feeling the downturn in smartphones and PCs, following the boom of the pandemic. In addition, K&S stock fell a bit after recent earnings, as the auto and industrial segment is also now in a downturn, keeping revenue down 3% year over year last quarter, even as other parts of the semiconductor sector begin to recover.
But in addition to the extreme cyclicality in its core legacy businesses, Kulicke and Soffa is also cultivating new secular high-growth businesses in advanced packaging and mini-LED lighting. These segments are small now but could see strong hypergrowth in the future.
First, K&S’s thermocompression bonding (TCB) solutions are now used in leading-edge AI chips, especially “chiplet” architectures as used in Advanced Micro Devices‘ (AMD 1.85%) new MI300 chips. While the business is currently much smaller than the core ball bonding business, the company noted in a November press release that its TCB sales had grown at a 149% compound annual growth rate since 2020 and that it expects “aggressive” follow-on orders through 2024 and 2025 on strong AI demand.
On its recent fiscal Q1 conference call, K&S management had more positive things to say about the TCB business, indicating that it’s taking market share in this important new product line:
We continue to actively support several customer engagements in parallel and anticipate additional orders from OSAT, IDM, and Foundry customers as we complete key evaluations and qualifications. In addition to our incumbent position in high-volume semiconductor markets, our advanced packaging efforts allowed us to take share in new high-growth markets including leading-edge logic, mobility and co-packaged optics. We are increasingly confident on the longevity and market potential of our TCB portfolio and expect to extend the technology well beyond the previously targeted 10 micron pitch. Reaching below this 10 micron threshold will further unlock the flexibility and long-term potential of our TCB solutions. We increasingly anticipate TCB will be an essential component to leading-edge assembly for many years to come.
In addition to new technological breakthroughs and growth in TCB, management also had positive things to say about Project W, a partnership with a very large and high-profile but unnamed customer in advanced display. K&S has recently developed solutions for miniLED and microLED lighting solutions, which offer better resolution than standard screens today, and which are found only in very high-end TV screens and tablets. However, there is the possibility they eventually make it to more mass markets, such as smartphones.
While we don’t know the identity of the large customer, management noted on the call: “I am happy to report that we are reaching new milestones, expect to ship additional capacity during the March quarter, and recognize revenue during the upcoming June quarter. As we work to successfully qualify a previously shipped system, we are beginning to ramp production in support of our customer’s long-term capacity plan.”
While details will emerge about both Kulicke’s TCB growth as well as Project W in the coming months and quarters, K&S remains a pretty cheap stock. It hasn’t taken off like other AI names, and based on the average of the $7 it earned at the top of the cycle and the $1 it earned at the bottom last year, KLIC’s current $47 stock price makes it a cheap way to play the growth of advanced electronics.
Google’s new secret weapon: AI subscriptions?
Nicholas Rossolillo (Alphabet): Google’s parent would really like to be known as the AI company. And perhaps for good reason. OpenAI’s ChatGPT was once thought to be a disruptor of the internet search titan. Not only has that not transpired, but Google is still thriving as well. It’s tough to dislodge a business with computing prowess of such epic proportions.
Rather than hamper Google Search, AI tools could actually yield Google new growth outlets. For example, during the last earnings call, Alphabet CEO Sundar Pichai mentioned that subscription services just reached $15 billion in annual revenue, a fivefold increase from 2019. YouTube is the marquee subscription product for Alphabet. But Pichai said on the earnings call:
Let me also talk about our subscription service, Google One. It’s doing incredibly well, with strong user growth. It provides expanded storage, unlocks exclusive features in Google products and allows us to build a strong relationship with our most engaged users. Google One is growing very well, and we’re just about to cross 100 million subscribers. From here, we are going to bring in more AI features, and look forward to more strong growth to come.
That’s a lot of subscribers.
Granted, a Google One subscription starts at just $1.99 a month. But on Feb. 8, Google is rolling out a new Google One AI Premium Plan for $19.99 a month, powered by Gemini (formerly Bard). Even if just a few million of those 100 million subs convert to the new premium plan — say 3 million, which would equate to $60 million in monthly sales — it could equate to yet another sizable jump in Google’s subscription revenue this year. And that’s just from one new service.
Google has been shelling out billions of dollars in new data center infrastructure to support its new AI aspirations — $32.3 billion in 2023, to be specific. New subscription services could be one of the ways Google starts to pay for all of that spending. So could Google Cloud business subscribers, which just crossed $9 billion in quarterly revenue at the end of 2023. Gemini is being embedded in Google Cloud, too, and a host of other Google products.
Alphabet stock trades for 25 times trailing-12-month earnings, or 19 times one-year-forward expected earnings. I’m happy to keep holding.
Inference at the edge of the cloud: Cloudflare’s pivot to AI infrastructure
Anders Bylund (Cloudflare): You might not think of network performance and security expert Cloudflare as an AI company. Truth be told, it really isn’t yet. But Cloudflare is setting up the tools and infrastructure to become a major AI name as the industry evolves.
In Thursday evening’s fourth-quarter earnings call, CEO Matthew Prince noted that Cloudflare isn’t collecting “material” revenues from AI yet, but that should change over time. For now, the company is covering its global data center network in graphics processing units (GPUs) specialized in inference, which is the end-user phase of accessing advanced AI systems.
“By the end of 2024, we plan to have inference-tuned GPUs deployed in nearly every city that makes up Cloudflare’s global network and within milliseconds of nearly every device connected to the Internet worldwide,” Prince said. “I think a lot of the money being spent on AI right now, especially with some of the hyperscale public cloud, is for training of models.”
That’s not what Cloudflare’s worldwide installation of inference GPUs will do. These modest batches of Nvidia (NVDA 3.58%) can’t compete with the monster supercomputers used to train the large language models behind ChatGPT or Google Gemini, but they are available at the drop of a digital hat for lighter workloads. The company is all about network performance and edge computing, not heavy-duty number-crunching. Hence, Cloudflare’s GPUs will tap into the AI engines built and trained by big-iron hardware, delivering chatbot conversations and data analysis results to the end user.
“We are not the right place to actually do model training,” Prince conceded. “But as that transitions over time and people start to figure out how can you take the models that you’ve built and turn them into real products, that’s where you’ll start to see revenue that is meaningful to us in terms of delivering the value in the AI space. But we’re still so early.”
So Cloudflare expects to build a robust AI business in the long run. For now, the company is just getting ready for future stages of the AI industry’s evolution. Even so, Prince noted that the inference GPUs (also known as Workers AI systems) helped the company land a large number of new or expanded network service deals in the fourth quarter, obliterating Wall Street’s consensus estimates across the board.
“Workers is a big piece of a lot of the deals that we see. So somewhere around 20% of the large deals that we closed have some workers component to it,” Prince said.
That’s not too shabby for a brand-new product in the early days of a technology revolution. I can’t wait to see Cloudflare finding its long-term niche in the AI market. This stock isn’t cheap, but it may very well be worth every penny of that premium price in the long haul.
