Part Three: Generative AI, Investing in the 21st Century Megatrend

This story features MACQUARIE GROUP LIMITED, and other companies. For more info SHARE ANALYSIS: MQG

How are global Generative AI tailwinds impacting on Australian companies, and which stocks are in the sweet-spot to benefit?

-Tailwinds from BigTech earnings
-NextDC and Goodman Group are just two of the winners
-Key takeaways from Global Portfolio Manager at Alphinity Investment Management

By Danielle Ecuyer

The story below is the third installment in a series on Generative AI. The first installment was published on 2 May 2024 https://fnarena.com/index.php/2024/05/02/generative-ai-investing-in-the-21st-century-megatrend-part-one/

The second installment was published on 9 May 2024 https://fnarena.com/index.php/2024/05/09/part-two-generative-ai-investing-in-the-21st-century-megatrend/

Stock prices only tell part of the macro story

Differentiating between the stock price moves of companies associated with artificial intelligence versus the growth and potential size of the total addressable markets will be an ongoing challenge for investors.

Comparisons can be drawn between previous secular trends such as electrification, mobility, telephony, the internet, and industrialisation, but like most historical precedents, the present circumstances are similar but different. 

The clean energy transition is a case in point. Despite the billions of dollars being committed to the transition, including major incentives such as the US Inflation Reduction Act, the top-down secular growth in clean energy/decarbonisation has been far from a straightforward investment proposition for investors. 

FY24 results for Macquarie Group ((MQG)) highlight the short term versus longer term challenges of investing in a megatrend. Demand for the clean energy assets owned by Macquarie Asset Management is proving to be more cyclically impacted than forecasts due to the higher interest rate environment.

Tesla is also case in point. As the electric vehicle market grew, competition, particularly from the subsidised Chinese manufacturers commoditised the product.

Secular megatrends do not grow in a straight line. Electric vehicles will ultimately replace internal combustion engines, however, the path to transition is likely to exhibit some of the normal cyclical demand and supplier drivers.

Top-down analysis of a total addressable market (TAM) to company specifics makes for a strong narrative, but ultimately the driver of stock prices will be revenue and earnings growth, alongside strong cashflow generation to support new investment.

The question of whether earnings are pulled forward or front loaded into the present should also be at the forefront of investors’ minds. The market will set earnings expectations and compare against actual results. Strong growth can be either rewarded or punished depending on expectations and recent performance.

“We encourage investors to do thorough due diligence on the technology underpinnings of these companies and differentiate real AI winners from the short-term AI beneficiaries likely to end up short in the long term.” The quote is from RBC ImagineTM: The Software Investor’s Handbook to Generative AI

Big Tech’s latest earnings offer insights into artificial intelligence investment trends

The latest quarterly results for the Magnificent 7 revealed impressive 43% growth in earnings compared to 14% a year ago, which includes the not so hot Tesla and META earnings, and also includes forecast earnings from Nvidia (to be released on May 22).

RBC Capital Markets analysis of the recent US big tech companies’ quarterly results extrapolates the trend to Australia’s data centre and ancillary service companies, noting hyperscalers are emerging from a period of consolidation and re-entering a growth phase.

Amazon Web Services (AWS), owned by Amazon, is the world’s largest hyperscaler with revenues split 60% from America, 23% internationally and 17% for AWS. 

AWS reported better than expected 1Q24 results including revenue growth of 17% on the previous corresponding period, with a 400-basis point acceleration in growth via its AI expansion. 

Management guided to ongoing growth between 7% to 11% in net sales growth for 2Q, or US$144bn to US$149bn.

Amazon’s commentary confirmed the company’s ongoing commitment to artificial intelligence through multiple levers, including Amazon Q, an AI powered assistant for software development; the flywheel effect from AWS growth and demand, and over 100,000 of Amazon’s selling partners are using one or more of their GenAI tools’.

Microsoft also delivered 1Q standout results. Revenue was split 43% from the “Intelligent Cloud”, 32% from Productivity and Business Processes, and 25% Personal Computing.

Microsoft’s Azure is the second largest cloud operator/hyperscaler after AWS and reported 31% growth versus consensus expectations of 28%, against the previous corresponding period.

Revenue growth for Intelligent AI’ rose 21% to US$26.7bn. Looking ahead management expects revenue to expand between 19% and 20% in constant currency.

Like its competitors, capital expenditure will continue to be driven by strong demand. Azure has continued to take market share and recently announced a US$1bn plus multi-year deal with Coca-Cola, with 88% year-on-year (yoy) growth in US$100m-plus sized Azure contracts, plus a doubling in the US$10m-plus segment.

Github reported 45% growth yoy and over 90% of the Fortune 500 is using the product.

The third largest hyperscaler after AWS and Azure is the Google Cloud Platform from Alphabet.

Alphabet generates 78% of revenue from Google Advertising, 11% from Google Cloud, and 11% from other sources.

Google Cloud reported 1Q revenue growth of 28% on the previous corresponding quarter to US$9.6bn, 2.4% above market consensus, and up 26% on the December quarter.

Management noted a doubling of capital expenditure to -US$12bn from -US$6bn compared to the previous corresponding quarter, driven by data centre demand.

Google is working across multiple artificial intelligence platforms including Nvidia GPUs with Google’s own TPU or application-specific integrated circuit (ASIC) in its neural networks; it added AI features to pixel, photos, Chrome, messages as well as Gemini (formerly Bard, the LLM) via a Gemini App on Android and the Google app on IOS.

 “We have developed new AI models and algorithms that are more than 100 times more efficient than they were 18 months ago.” 

In reference to Meta, this company reported a 12% rise in capital expenditure to -US$37.5bn for FY24 from -US$33.5bn previously, due to rising investment in AI related infrastructure.

RBC research concludes that even a snapshot from the 1Q big tech results are all positive tailwinds for Australia’s data centre exposed companies, as the trickle-down impact from higher investment feeds through. 

Catching the Big Tech cloud investment tailwinds, which Australian companies are in the sweet-spot?

Morgan Stanley takes a more in-depth and nuanced approach to the domestic data centre industry in its research.

With a base case of an 8% compound annual growth rate (CAGR) out to 2030 for the Australian data centre market, and a bull case of a 20% CAGR, the outlook looks structurally robust.

Australia is already the fifth largest data centre operator globally, in line with London (after US, Europe, Northern Virginia and Beijing/Shanghai). The market is forecast to expand from 1050MW currently to 2500MW by 2030, at a 13% annual rate.

Citi observes Australia has circa 1GW of live data centre capacity. Sydney is the largest market locally having expanded 2.5 times in the last 5-years and is in the top 20 globally.

Melbourne capacity has doubled in the same period and the research indicates Melbourne has little available capacity.

The demand from government for the digital and cloud first strategy, as well as data sovereignty requirements, are driving the forecast 14% compound average growth to 2030 of some 2.5GWs.

Morgan Stanley estimates there could be revenue uplift of between $5.6bn to $8.4bn p.a. on the back of an estimated total spend between $21bn to $28bn in data centre development over the next 8 years.

Against such a favourable backdrop, how does the earnings outlook shape up for NextDC ((NXT)), Megaport ((MP1)), and Goodman Group ((GMG))?

Some highlights from Morgan Stanley’s analysis:

–NextDC has a domestic and growing international presence as an owner and operator of data centres and is one of the largest domestic players, alongside of US operator Equinix, AirTrunk (unlisted), Digital Reality, Global Switch (not listed), Canberra Data Centres (not listed), Macquarie Technology ((MAQ)) and hyperscalers Microsoft and Amazon.

-Higher development and land costs are expected to push down the internal rate of return (IRR) on its development pipeline from over 20% to between 10%-12%. The IRR is forecast to exceed the expected average cost of capital (estimated at 7.7%). 

-Longer term increased competition has the potential to push down pricing, but for now demand remains robust.

-Traditionally, NextDC has traded at a premium valuation to international peers, like Equinix and Digital Realty Trust, which the analysts attribute to the higher growth rates and the very regulated domestic markets. Post an adjustment for growth, the valuation is relatively not as high.

-NextDC has a total build capacity of 960MW of which 420MW are incorporated into the broker’s core earnings forecasts and 560MW from the land bank.

-The target price is lifted to $20 from $17, accounting for a $16 per share valuation for the existing data centres and those under construction, plus $4 per share for the future land bank.

Citi is also a fan of NextDC and considers it as the optimal direct investment play in the artificial intelligence in APAC, due to its strength and leadership in the market. 

Although the recent equity raising was sooner than expected, the cash raised frees up development opportunities and removes the overhang of further equity requirements.

The broker estimates contracted utilisation (MW) to expand almost twice in the next three years. Citi has a Buy rating with a $20.10 target price on the stock.

–Goodman Group revealed in August last year a 3.6GW portfolio of industrial sites with power either secured or close to secured, referred to as a power bank, with another 1-2GW in different planning stages.

-Morgan Stanley estimates the 3.6GW power bank, depending on how Goodman decides to advance the assets, could realise another $20bn in valuation for the group versus the current market capitalisation of $64bn.

-The flexibility of the business model allows for three forms of earnings accretion including development profits, rental income on the completed assets, and management fees for the Goodman funds.

-Currently the group has a $4bn exposure to data centres versus the $79bn portfolio. Morgan Stanley forecasts this could grow to $28bn by 2030 as a base case.

-Importantly, capital expenditure can be funded by cashflow or, if necessary, some smaller asset sales.

-Management recently upgraded guidance, noting the upside potential for the data centre business pipeline which has increased to 3.9GW from 3.6GW.

Goodman Group is rated Overweight with a $36.65 target price by Morgan Stanley.

Powering the growth in domestic data centres

The lastest update from Morgan Stanley research reveals Australia has sufficient capacity in its existing power system for the data centre developments up until 2030. Data centre power demand is forecast to expand to 8% of the grid’s generation capacity by 2030 from 5% currently as a base case.

Water usage is also important.

The growth in data centres is viewed as macro tailwind for both AGL Energy ((AGL)) and Origin Energy ((ORG)) via power prices on the back of higher energy demand and the development of renewable energy projects.

Other companies set to benefit

The artificial intelligence tailwinds are not exclusively for the above-mentioned cloud/data centre beneficiaries, Pro Medicus ((PME)) has been highlighted by both RBC and Goldman Sachs as a potential AI winner.

Goldman Sachs initiated coverage on Pro Medicus with a Buy rating and $134 target price on the growing significance of Gen AI enhanced software services from the cloud-based radiography diagnostics specialist.

The broker sees multiple opportunities for Pro Medicus. AI is expected to reach 9% of the company’s revenue by FY30 from less than 1% in FY25. Revenue is already being produced from the Visage breast density AI algorithm. The analyst also envisages the company has scope to diversify into Cardiology.

Megaport ((MP1)) is also a slated beneficiary by RBC from the growth in data centres, with its software solutions offering business connections and access points for data transferal in and within data centres.

Citi is also positive on Megaport as a global leading data centre and cloud-on-ramp service provider. The broker views the growth in inference workload as offering more upside potential than artificial intelligence connectivity. There is a Buy rating and a $16.05 target price.

Let’s not leave Macquarie Technology ((MAQ)) out of the mix with the hyperscalers as anchor tenants of its five data centres. Management is expanding the capacity of the Macquarie Park Data Centre Campus to reach 56MW from 50MW with power availability up to 63MW.

Also: NZ-headquartered Infratil ((IFT)) is a major shareholder in Canberra Data Centres ((CDC)). Macquarie Group and Global Data Centre Group ((GDC)) are equity owners of AirTrunk.

Addendum: Key takeaways on Generative AI from Global Investment Manager, Trent Masters at Alphinity Investment Management

“In terms of Ai, as the saying goes “history doesn’t repeat itself, but it often rhymes”. I think we are seeing this in Ai with the parallel being the Internet era starting in 2010. There we saw stock outperformance begin with the enablers like semiconductors, then saw it shift into infrastructure/devices before the final move into software and services. In terms of stocks, this saw the outperformance start with Qualcomm/ARM before shifting to device makers such as Samsung and Apple before the enormous value creation in software enabled businesses such as Google and Amazon.

“Comparing that to today, the leading enablers this time are Nvidia, AMD and memory players such as SK Hynix and Micron. We have seen significant outperformance already wash through these businesses, and I think there is still some room to run. In terms of the infrastructure level, we are seeing this emerge particularly in cloud players (the recent results for Azure, AWS and GCP were all strong) while the devices element around Ai on device for PC’s and smartphones is just starting to emerge. In software and services, it is still very early and trying to pick winners in this area given the speed of evolution in Ai can be fraught. We’ve seen companies such as Adobe and SNOW swing from potential winners to losers in quick time as the landscape rapidly shifts.

“What I think is different this time at the software and services level is that the Internet era created fertile ground for disrupters to create completely new business models. The foundational tech of the internet was largely free, and an avenue for companies like Google and Amazon to create new businesses and disrupt incumbents. But with Ai, given the inherent cost in compute requirements this will at least in the initial stages see the benefits flow to incumbents until some real breakthrough use cases emerge.

“In terms of stocks, where this leaves us is more comfortable where we can touch and feel the earnings inflection. This is in the enablers such as Nvidia where the size of the accelerator market continues to be revised up as evidenced by the robust capex spend from cloud players in recent results. Companies such as memory business SK Hynix also benefit from High Bandwidth Memory requirements for Ai.

“In the Infrastructure layer, it really is all about the cloud players at this stage, with Microsoft as the standout. Not only is Microsoft Azure taking share due to their Ai capabilities, but Microsoft also has the potential software and services inflection from being able to distribute Ai services into a 450m commercial customer base.

“Also benefiting in the infrastructure layer is network business Arista networks, along with some signs of inflection in what are considered “old world” businesses such as Dell.

“In software and services, as discussed it is still early (true software and services company outperformance only emerged around 3yrs into the Internet era) and fraught to try and pick early winners. As such the preference remains for those companies that can push Ai product extensions into a broad existing customer base. It isn’t so much about having the “best model” but about the application created from it and the channel through which to distribute it in the initial stages. However this software and services area is where potentially the next Google or Amazon for the AI era could emerge”

The author owns shares in Goodman Group, NextDC and Pro Medicus.

For more in depth analysis please refer to FNArena’s Stock Analysis and the associated stories.

Find out why FNArena subscribers like the service so much: “Your Feedback (Thank You)” – Warning this story contains unashamedly positive feedback on the service provided.

FNArena is proud about its track record and past achievements: Ten Years On

Click to view our Glossary of Financial Terms