Generative AI training infrastructure remains white hot as an investment. Numerous start-ups have appeared to address the need. Given the end-to-end solution Nvidia created that spans chip design, algorithms, and software, this AI infrastructure consisting of data centers filled with accelerated computing systems (primary using Nvidia technology) is arguably easier to build than “traditional” cloud computing infrastructure. That has been rocket fuel for these specialized data center builder-operators in the last year.
One of those specialists is Applied Digital (APLD -1.42%), formerly Applied Blockchain. Applied Digital raced higher in 2023, fell hard, then started to rally again at the end of the year. But to kick off 2024, shares are falling again as the market comes to terms with the cadence of future profitability for the business. Here’s why most individual investors should steer clear.
Growth starts to moderate for AI infrastructure
We’re a long way from the “chip shortage” of 2021 when pandemic-disrupted supply chains paired with skyrocketing consumer demand for all things electronic wreaked havoc on the semiconductor industry. But new supply shortages struck last year, affecting the electrical components used to network GPUs (graphics processing units, chip systems used to accelerate computing work) and turn data centers into massive computing units that can train AI algorithms.
In its latest update, Q2 fiscal 2024 (the three months ended in November 2023), Applied said some of these component shortages have begun to improve. However, the shortage is still bad enough that it led management to significantly downgrade its full fiscal-year guidance from the outlook provided three months ago.
Applied Digital Financial Metric |
New Fiscal 2024 Guidance (as of Jan. 16, 2024) |
Old Fiscal 2024 Guidance (as of Oct. 9, 2023) |
---|---|---|
Revenue |
Exit fiscal 2024 with annualized revenue run rate of approximately $500 million |
$385 million to $405 million |
Adjusted EBITDA |
Exit fiscal 2024 with annualized adjusted EBITDA run rate of approximately $250 million |
$195 million to $205 million |
Data source: Applied Digital.
Component shortages happen, it’s something outside of Applied’s control. But there’s one annoying issue here: a switch from providing explicit guidance on what to expect for the current full-year period, to providing an annualized rate of revenue and adjusted EBITDA (earnings before interest, tax, depreciation, and amortization) at the end of the year. What does that mean?
Basically, in Q4 fiscal 2024, Applied is expecting revenue of $125 million ($500 million in “annualized revenue” divided by four). Adjusted EBITDA should now be about $62.5 million (the $250 million annualized figure divided by four). With Applied’s first-half fiscal 2024 revenue and adjusted EBITDA at a respective $78.5 million and $20.2 million after the latest update, suffice to say the original full-year guidance (again, see chart above) was an overestimate.
Focus on profitability, and how losses are paid for
I’ve been skeptical of this company since it first landed on investors’ radars as a “top AI start-up,” and the downward revision in guidance certainly doesn’t help.
But besides my issues with the way the business itself is structured, some investors may still be looking at the high rate of “adjusted EBITDA” as a reason for optimism. I think it best to look past this metric. On all other metrics — GAAP net income and loss, unadjusted EBITDA, free cash flow, etc. — Applied is still losing money.
- First-half fiscal 2024 GAAP net loss was $22 million.
- First-half fiscal 2024 EBITDA loss was $22.4 million.
- First-half fiscal 2024 free cash flow was negative $36.8 million.
Suffice to say this is still a high-risk start-up, with only a potential high reward.
Adding fuzziness to the long-term thesis is the fact that Applied is also borrowing money from one of its key shareholders, B. Riley Financial, and it’s also raising additional cash by selling new shares (which dilutes ownership of existing shareholders).
All of these issues added up to a big 20%-plus decline in share price immediately following the earnings release. The AI hype is moderating again, and for the average investor, there are far safer options to invest in AI right now.