Gaudi 3 unveiled! Intel's secret weapon against Nvidia, when will the stock price rebound?

After Intel's results for the first quarter of 2024 were released, it had to lower its expectations — once again making many people think that Intel is a “laggard.”

Despite falling short of future guidance expectations, overall, we can't say that the first quarter results were terrible.

I think Gaudi 3 may have a competitive advantage in the long term, especially in enterprise AI applications where efficiency and cost effectiveness are critical.

Intel's goal is to become the second-largest external foundry by 2030, driven by AI opportunities and a strong product line based on 20A and 18A.

I'm not giving up on Intel too soon. My DCF model showed that it was undervalued by about 18% in different situations, which is quite impressive and worth upgrading to a “buy”.

I started watching Intel Corporation (NASDAQ: INTC) stock at the end of January 2024, when the stock price was $43.71. My point is simple: investors should continue to avoid this obvious laggard because at the time, it seemed to me that Intel was just empty talking about its plans, while Nvidia (NVDA) and Advanced Microdevices (AMD) were working hard and continuing to seize market share.

Exactly after the announcement of the results for the first quarter of 2024, Intel's management had to lower expectations — which of course also made many market participants think Intel was a “laggard,” and Intel's stock price dropped more than 26% from what it was when I posted the “neutral” rating.

But when Intel's stock price is down more than 36% year to date, while NVDA and AMD have risen 91% and 13%, respectively, over the same period, is the market really fair to Intel? I suggest exploring this topic in depth.

In the first quarter of fiscal year 2024, Intel reported revenue of $12.7 billion (up 9% year over year), but this was a 17% month-on-month decrease compared to the fourth quarter of fiscal year 2023. This figure reached the midpoint of management guidance, but fell short of the $127.8 billion consensus estimate. Non-GAAP net income was $0.18 per share, compared to a loss of $0.04 per share for the same period last year, exceeding management and consensus expectations of $0.13 per share:

Overall, we can't say that the first quarter results were terrible. But what the market clearly doesn't like is that Intel failed to meet quarterly revenue expectations for the second consecutive quarter (mainly due to losses in the foundry business and questions about its AI efforts). Furthermore, the company's expectations for the second quarter fell short of Wall Street's expectations, leading to significant negative revisions to earnings per share expectations for the second quarter:

During the earnings call, CEO Pat Gelsinger highlighted “successful expense management and steady progress on long-term priorities.” However, there is a worrying fact: Intel has yet to reflect a significant impact based on booming AI trends in its financial data. Despite many efforts, Intel's gross margin still lags behind its direct competitors — AMD and Nvidia — the latter, in my opinion, still operate more efficiently:

Although Intel's EBITDA margin is currently slightly higher than AMD's, it is more cyclical, as shown in the chart above. This introduces additional risk, makes forecasting more challenging, and results in discounts on stock valuations.

However, past performance doesn't always accurately predict the future. Especially for cyclical industries such as semiconductors for PCs or other end markets. During the earnings call, Intel said it expects the first quarter of 2024 to be the bottom of the cycle, and revenue is expected to gradually increase in 2025. Key growth drivers include the corporate refresh cycle, the recovery of AI PCs, data centers, and the cyclical recovery of Altera, Mobileye, and NEX businesses.

Among recent major corporate events, we can note that Intel's first “Intel Foundry Direct Connect” event attracted 300 partners and customers, and Microsoft Corporation (MSFT) became the 5th customer of its 18A process node (expected in 2025). Intel also introduced 14A process nodes (15% higher performance per watt than 18A), high-NA EUV technology, and introduced the next generation Gaudi 3 accelerator.

In mid-April, Forbes wrote that Intel's internal testing showed that Gaudi 3 Llama 2 (13B parameters) was 1.7 times faster than Nvidia's H100 and 1.4 times faster than the GPT-3 (175B parameter).

The speed of reasoning is also impressive: the Llama2 7B is 1.1 times faster than Nvidia's product, and the Llama2 70B is 1.7 times faster. Of particular note is the Falcon 180B model, which is up to 4 times faster. Of course, these results are information of “internal company testing” — no company would publicly criticize its own development. But even if only half of these results are true, I believe Gaudi 3 may have a competitive advantage over the long term, particularly in enterprise AI applications where efficiency and cost effectiveness are critical.

As a result, Intel may not be financially efficient enough, but it is still a huge and important player in the market, and its potential to influence the market at this stage cannot be ignored.

Intel's goal is to become the second-largest external foundry by 2030, driven by AI opportunities and a strong product line based on 20A and 18A. The company has also redivided its business divisions, now disclosing foundry revenue and profit and loss separately, and separating Altera from the data center division. The company is also actively expanding its global manufacturing facilities in Israel, Ohio, and Arizona, partly funded by government grants and the Chip Act. In my opinion, all of this could have a pretty positive impact on profit margins in the foreseeable future.

According to David Zinsner, this year will initially be a year of high startup costs for the company (the second quarter was more pronounced than the first quarter). This has put some pressure on gross margins. But as revenue increases, they are likely to perform well in the third and fourth quarters, increasing gross margins.

Summarizing comments from the CFO and management during the earnings call, investors should expect profit margin data for the second half of 2024 to be more positive than in the first half of the year. However, after lowering expectations for the second quarter and continuing to sharply lower earnings per share expectations for the next few quarters, the market is still in some kind of aftershock:

I think these expectations are an overreaction. why?

According to Seek Alpha's consensus data, Intel will earn $0.81 per share in the last two quarters of fiscal year 2024, which is only 15.7% higher than in the second half of 2023. The same consensus data also shows that Intel's sales should exceed $30 billion in the second half of fiscal year 2024, an increase of nearly 4.55% over last year. Since gross margin should be higher, I think Intel's operating leverage should theoretically provide a stronger increase than currently anticipated.

Therefore, I think the recent pessimistic correction on earnings per share is too pessimistic in the medium term. The current pressure on the stock has far outpaced its direct competitors—of course, for a reason, but in my opinion, abandoning one of the semiconductor industry's flagships would be unwise.

In my previous article, I used the multiple method for valuation. Given the recent sharp drop in stock prices, I now recommend using the DCF model to assess the company's fair value — an approach designed to determine whether the extent of this sell-off was reasonable. Intel's valuation should take into account its cyclical nature. If we look at long-term indicators of EBIT profit margins, we can see the direction and approximate upper limit of Intel's profit margin expansion over the next few years. My expectations are shown in the image below:

I haven't mentioned income before. My recommendation is to rely on consensus predictions, as they come from the extensive experience of Wall Street analysts who closely follow the industry, and they are more likely than me to accurately predict trends in earnings and earnings per share. Note: I'm not using the 22% sales growth forecast for fiscal year 2028, as this consensus comes from one analyst's forecast alone and may not be representative. Therefore, I'm going to take the 2027 fiscal year forecast (8.48%) and reduce the growth rate to 7%.

Assuming depreciation remains almost constant over the next few years, I still think Intel is likely to reduce capital expenditure as a percentage of sales (at least 0.25 in fiscal year 2028) — also due to the cyclical nature of the business. As the investment begins to pay off, the new investment cycle no longer requires the same amount of capital injection into the project, and Intel's revenue growth should make this forecast very likely to come true.

Here's my current forecast for the company's free cash flow (FCFF):

I estimate Intel's debt cost to be around 5.54% — this is the yield to maturity on its liquid company bonds. Assuming a risk-free interest rate of 4.35% and a market risk premium of 5%, this results in a weighted average cost of capital (WACC) of 7.9%. This number seems low, but this is due to Intel's capital structure.

If FCFF only continues to grow at a rate of 3%, this would result in a fair value of around $37.5 per share, which is approximately 18% higher than the current share price.

So I think my DCF model clearly shows that Intel is not only oversold today, it's also undervalued.

I think Intel's main risk is the competitiveness of its competitor AMD, which has gained significant CPU market share in both PC and data centers. If Intel fails to maintain its dominant position in the CPU market, it will be difficult to regain hope in the market. Although Intel has developed several product strategy plans since the beginning of 2023, investors should remain cautious in facing these announcements, although these plans provide clear milestones for evaluating the company's future performance. As I wrote in my previous post, a company must actually prove that it should be an industry leader and is really ready to evolve with the times (not just talk about it).

Additionally, Intel, like other chipmakers, faces the risk of a potential economic downturn, which could lead to falling demand for semiconductors, inventory backlogs, and falling cyclical orders. Today, structural difficulties in terminal markets such as PCs are putting a lot of pressure on Intel — if the cycle doesn't reverse as expected, we are likely to see a further decline in stock prices.

Another risk is the valuation of the company. I mentioned earlier that Intel's intrinsic value is undervalued, but if you look at how EV/EBITDA has changed in recent weeks, this conclusion might be challenged:

Based on EV/EBITDA dynamics, we can see a significant increase in multiples in recent weeks despite falling stock prices. This means that in reality, the actual EBITDA based on TTM has become disproportionately smaller than the value of the enterprise — as a result, the multiples have increased despite falling stock prices.

Therefore, this approach could explain Intel's decline more fully. However, we're talking TTM data, not forward-looking data. In my opinion, the future plays a more critical role in determining today's fair value. Comparing Intel's expected EV/EBITDA with its successful peers reveals a significant 2—3 times gap, which actually makes Intel seem really cheap.

Also, watch out for the PEG ratio — it's trading at a much lower discount than NVDA and AMD:

Another risk that cannot be ignored is an increase in debt on a company's balance sheet. Although the debt-to-equity ratio is still quite low, in my opinion, the rise in debt itself is a matter of concern.

After Intel announced its results for the first quarter of 2024, its stock fell sharply again, confirming what many people (myself included) suspected: Intel is falling behind. Although Intel reached the midpoint of sales forecasts, it failed to meet sales targets for two consecutive quarters, and the outlook was not positive, increasing negative sentiment.

However, I think the sharp drop in Intel's stock price is an overreaction. Actually, the situation isn't that bad. Yes, the company does lag behind its peers, but it still has projects that can boost sales and profit margins in the future. For example, Intel's new products, such as the Gaudi 3 accelerator, hint at its possible return in the field of AI. So I'm not giving up on Intel too soon.

My DCF model showed that it was undervalued by about 18% in different scenarios, which is quite impressive and worth upgrading its evaluation to “buy.”