Broadcom: How AI Will Influence Next Week’s Q2 Report (NASDAQ:AVGO)

The market certainly appears to have figured out some of the winning companies in what I call the “picks & shovels” area of AI: Those companies that provide basic infrastructure critical to running AI/ML algorithms on mega-data and LLMs (large language models). In a previous Seeking Alpha article, I suggested that three winners in AI infrastructure would be DataDog (DDOG), Nvidia (NVDA), and Broadcom (NASDAQ:AVGO) (see DataDog And 2 Other ‘Sure-Fire’ Winners In The AI Arms Race). Indeed, all three of those stocks have rocketed higher year-to-date (as have the Nasdaq-100 Trust (QQQ) and SPDR Technology Sector ETF (XLK)) and have more than doubled the returns of the Vanguard S&P 500 ETF (VOO) with Nvidia leaving the field in the dust after a severe bear-market correction last year (see graphic below). Today, I’ll take a closer look at Broadcom, which has lagged relative to my other picks, but which should release another strong quarterly report next week. Despite AVGO’s strong fundamentals, high margins, strong free cash flow generation, and a $18.40/share annual dividend (2.7%), Broadcom is still – somewhat inexplicably – trading at a big discount to peers (forward P/E=16.4x).

Investment Thesis

Despite the amount of press Broadcom gets over its relationship with Apple (AAPL) (see Broadcom: CEO Shakedown – Hock Tan Vs. Tim Cook) and the pending acquisition of VMware (VMW), in my opinion the heart of the company remains its high-speed networking development platform (both hardware and software) and the products the platform produces that always seems to keep Broadcom at least one-step (if not two …) ahead of the competition.

High-speed networking, of course, will be critical to providing the mega data to/from the cloud for the AI/ML algorithms and Large Language Models (“LLMs”) to feed high-performance computers (“HPC’s) – such as those provided by Nvidia.

LLMs refers to AI models that can generate natural language texts from large amounts of data. LLMs typically use deep neural networks to “learn” from billions or trillions of words in order to produce texts on any topic. But LLMs are just one aspect of the AI/ML revolution that could benefit Broadcom: IoT, enterprise cloud, data centers, and business optimization (i.e. DataDog) are other sub-sectors that will also require fast access to mega-data. Clearly then, there are multiple drivers for access to high-bandwidth low-latency data – and Broadcom has the leading-edge high-speed networking products to satisfy that demand.

Products

Broadcom has three primary switching platforms that will benefit from AI: Tomahawk, Trident, and Jericho. For example, in March, Broadcom announced the world’s first 51.2 Tbps (terabits per second) switch in production volume. The leading-edge Tomahawk 5 Family of Ethernet chips were engineered specifically to accelerate AI/ML deployments. The company says the Tomahawk 5’s 51.2 Tbps of switching capacity is twice that of any other switch on the market and, as a result, enables the fastest data transfer between AI/ML endpoints. The Tomahawk 5:

… enables single-hop connectivity between 256 high-performance AI/ML accelerators, each having 200Gbps of network bandwidth. This results in the fastest completion time for AI training and inference jobs, including for today’s increasingly complex and prevalent generative AI models.

On the Q1 conference call back in March, Broadcom CEO Hock Tan reported on a companion networking product that will dovetail nicely with the Tomahawk 5 switch:

So just this week, we announced the industry’s first integrated silicon photonics networking solution code name Bailey, which integrates the active optical interconnects with our next-generation Tomahawk 5 switch at 51.2 terabit per second. Bailey doubles switching performance but it will reduce total system power.

Tan went on to describe the impact AI is having on its hyperscaler customers, which already deploy the company’s Jericho 2 switches in their AI networks. He estimated that, in 2022, the company’s Ethernet switch shipments deployed in AI was over $200 million but:

With the expected exponential demand from our hyperscale customers, we forecast that this could grow to well over $800 million in 2023. We anticipate this trend will continue to accelerate and mindful that we need even more higher performance networks in the future.

That’s an expected quadrupling of its switch business in one year, and, therefore, I expect the impact to begin to hit the top and bottom lines in the Q2 report next week. Indeed, Broadcom’s networking segment could easily grow ~15% this year.

In addition, Broadcom also has a thriving custom application specific integrated circuit (“ASIC”) business (~$2 billion in FY22), which should also have strong AI related tailwinds going forward and could potentially grow by an estimated 40% this year.

Q2 Earnings Preview

As can be seen in the chart below (per Yahoo Finance), Broadcom is widely followed (20+ analysts) and the current Q2 consensus estimate is for EPS of $10.08:

That would be +11.1% yoy. Q2 revenue is expected to come in at $8.7 billion, which would be up 7.3% yoy.

Valuation

Given the full-year EPS estimate above ($41.29), Broadcom currently is trading at a forward P/E of 16.5x. The chart below compares Broadcom’s current valuation with my other two AI picks and I added AMD (AMD) as well:

Click to enlarge

Clearly Broadcom is the “value” of the four chosen stock, with risk/reward rising from AMD to Nvidia to DataDog. That said, don’t let DataDog fool you: Despite posting a Q1 GAAP net loss, DDOG grew revenue by 32.8% yoy, generated $116.3 million in free-cash-flow during the quarter, and raised forward guidance.

In my book, Broadcom is significantly undervalued. I say that based on Broadcom’s strong performance during the tech-sector bear-market last year. Here is what Broadcom did in FY22:

• Grew revenue by 21%.
• Grew earnings by 76.9%.
• Grew the dividend by 12%.
• Generated $16.3 billion in free cash flow (49% of revenue).

Is it just me, or does a company that accomplished all that during a bear market in the technology sector in which it is a part of deserve a higher forward P/E than 16.5x?

As can be seen in the chart above, I’m not the only one who seems to think so as Broadcom has rallied sharply from the 2022 bear-market low and is currently trading at or near an all-time high. Indeed, the stock is up over $50/share just over the past five trading days.

Risks

I give the VMware deal a 50/50 chance of going through. As most of you know, the UK recently blocked Microsoft’s (MSFT) bid for Activision Blizzard (ATVI) while the EU and China both gave antitrust approval. That may be a read-across to Broadcom’s bid for VMware, with CEO Hock Tan recently meeting with EU regulators to reportedly offer remedies in order to move the deal forward. Honestly, your guess is as good as mine in terms of the deal’s ultimate approval by all the required global regulators.

Regardless, at the end of Q1 AVGO had $38.2 billion in debt and $12.6 billion in cash – an arguably and relatively large net-debt position. However, note that in Q1 Broadcom generated $3.9 billion in free cash flow (+16% yoy), so investors should view AVGO’s debt in the context of its high-margin business that generates strong FCF quarter after quarter. That said, if the VMware deal goes through, it may be a short-term minor negative in that the debt-load will obviously increase. Remember, the agreement was for a 50/50 cash/stock transaction, with Broadcom also assuming ~$8 billion of VMware’s net debt.

For Q2 earnings, downside risk is a potentially softer quarter due to weaker than expected business with Apple during an already seasonally slow quarter.

Upside risks include a faster than expected ramp-up of high-speed networking sales and orders.

Summary and Conclusion

Broadcom is likely on target to deliver another strong quarterly report characterized by moderate-but-steady revenue growth, strong margins, and strong free cash flow. The stock has had a huge rally this week into the earnings report next Thursday. For current Broadcom holders, I rate the stock a hold. For those who don’t already own Broadcom and would like to establish a position, I suggest you buy a small starter position (say 10% of your ultimate allocation) and wait to add shares on pullbacks due to market volatility. Regardless, I’m relatively sure that a year from now, Broadcom will be significantly higher than it is today.

I’ll end with a 10-year total returns comparison of Broadcom vs. the QQQs, the S&P 500 as represented by the VOO ETF, and the DJIA as represented by the SPDR DJIA ETF (DIA):

As you can see, Broadcom has dominated the major broad stock market indexes over the past decade. In addition, note that Broadcom made multiple large acquisitions over the timeframe shown, and that CEO Hock Tan proved himself again-and-again by buying those businesses and ringing out high-margins and free cash flow. I see no reason to doubt a deal (if consummated) with VMware would be any different. Or that Broadcom will continue to outperform the broad major market indexes over the coming decade.