Letter to the saver

Intel's gamble: more factories serving competitors' chips

Focus. The push to open up their plants to third-party production
also to boost business. Challenge on innovation and large investments

by Vittorio Carlini

INTEL CORPORATION DISPOSITIVI A SEMICONDUTTORE DISPOSITIVO MICROPROCESSORI MICROPROCESSORE COMPONENTI DI RETE COMPONENTE CHIPSET PER MOTHERBOARD CHIP PER SCHEDE VIDEO SCHEDA CIRCUITI INTEGRATI CIRCUITO INTEGRATO CORE

6' min read

6' min read

IDM 2.0. That is: Integrated design manufacturer 2.0. This is the strategy initiated a few years ago by Intel and in respect of which, at the end of last February, the group provided important updates. The project, on closer inspection, is at the heart of the re-launch of the Santa Clara microprocessor company. A business rethink that, in order to be better understood, requires a reminder of the group's corporate purpose.

The multinational company is an integrated chip manufacturer. That is, it carries out a large part of the three main phases of semiconductor construction: the design of the chip architecture; the production of the chip on the wafer (essentially by ultraviolet light photolithography); assembly (in plastic or ceramic containers); and testing. All the steps described were historically conducted on Intel's own solutions. Now, however, and this is the main point of IDM 2.0, the group's foundries (factories) are increasingly opening up to third-party production. The company has created an ad-hoc division - renamed Intel foundry a few days ago - to, on the one hand, support the project; and, on the other, underline its objectives. Which are several. First and foremost is growth in the foundry sector. This, according to Statista, is currently dominated by Tsmc (60.2% market share). Then, well behind it, are the likes of Global Foundries (9.9%) and Umc (6.6%). Intel, for its part, boasts a market share of less than 2%. Well: the aim, by 2030, is to become the second global player after the Taiwanese group.

Loading...

ESERCIZI A CONFRONTO

Loading...

The packaging

.

The goal is very - for some too - ambitious. From which it is logical to ask: what cards does Intel intend to play? One theme, emphasised by CEO Pat Gelsinger himself, is packaging (assembly) and advanced testing. This phase of chip production has become crucial. A few years ago it was worth about 15 per cent of the production costs in the most sophisticated processors (mainly used in AI). Now, however, it weighs 35-40% of the overall burden. It is clear, therefore, that on the one hand Intel's experience on this front is an advantage; and on the other, it is not surprising that, just last January, the company started up the Fab 9 in Rio Rancho, New Mexico. A plant - precisely - aimed at the production of advanced packaging technologies.

REDDITIVITÀ E SEGMENTI OPERATIVI

Loading...

High-Tech Manufacturing

.

But it is not just a question of assembly or testing. The technological challenge, essential to attract customers, is in the production itself. In general, the semiconductor industry is focused on continuous miniaturisation. A context in which the nanometre (one billionth of a metre) is used as a unit of measurement, typically referring to the gate size of the transistor. The smaller this becomes, the more transistors can be inserted into the chip, increasing its computing power. Well: Intel, at its meeting last February, first confirmed the approach - announced about three years ago - of the so-called 5N4y.i.e.: five new production processes in four years. Then he indicated that the programme will expand - by 2027 - with the Intel 14A process. In other words: it will lead to the production of transistors with a 14 Angstron gate, or rather 1.4 nanometres. In the shorter term instead - in 2024 - the 2 nanometre process is planned, and - between 2024 and 2025 - the 1.8 nanometre process. Beyond the individual processes and numbers, it is clear that miniaturisation - together with reliability - is a key element for the success of Intel's strategy. A difficult technological challenge that is played out on the same reduction of energy consumption. A front where the group has begun to exploit PowerVia. A technology which, by having the power supply arrive from the rear of the wafer - instead of the front -, makes it possible to make the chip's energy use more efficient.

That greater efficiency that - and herein lies another objective of the Intel Foundry strategy - is sought in the company's various intramural divisions. An example? The design of the chip architecture. "When,' explains Alessandro Piva, director of PoliMi's AI Observatory, 'the various internal areas, which no longer have exclusive rights to the foundry, are put in competition with external companies, they are forced to improve the quality and quantity of their output.

LA DINAMICA DEL GROSS MARGIN

Loading...

The competition

.

All as easy as drinking a glass of water, then? The reality is more complicated. Beyond technological competition, several experts point out one aspect. Intel, by entering the foundry sector, becomes the rival of the other players in the industry. A scenario in which the moves of TSMC on duty risk becoming a danger for Gelsinger's strategy. 'In general, the point raised,' says Carlo De Luca, head AM at Gamma Capital Markets, 'is valid. Intel's moves may lead 'other foundries to cut capacity availability in favour of Intel itself'. Having said that, however, the US company's idea is based on different levels of technologies in concert with each other: 'from the production process to the architecture of the systems up to the software to realise them'. Thus, 'this is a frontier foundry, designed also and above all for AI, which would allow Intel to get back into the chip manufacturing game, without necessarily competing directly with other players'.

"Not least because," echoes Bob O'Donnell, President and Chief Analyst of TECHnalysis Research, "Intel's direct competition with TSMC, on the one hand, will be modest for a few years"; and, on the other hand, "the work the company does with Taiwan is also relatively small". Consequently: in the short term, 'no major changes in the collaboration between the two are expected'. In the longer term, on the other hand, the Santa Clara company will 'move the advanced processes, which it carries out with TSMC, to its own factories', thus eliminating the problems. In short: the experts do not seem negative about Intel's bet. However, the do-it-yourselfer must still have one fact in mind: in frontier technology, such changes in strategy do not always bring benefits.

Especially, as in the present case, when they require billions in capital expenditure. An example? In order to realise the two factories in Germany, the group has put 30 billion on the plate (Berlin has planned 9.9 billion in subsidies). These are large sums which, should they not have the appropriate return, could impact on the financial structure. 'Undoubtedly,' De Luca resumes, 'disbursements can further weaken the financial situation. For the past three years the company has had a 'decreasing cash flow, going from almost 30 billion in 2021 to 11 billion in 2023'. And this, 'despite the 3 billion cost cut in the last financial year'. That said, 'the company,' O'Donnel resumes, 'takes advantage of the interest in more geographic articulation', also for geopolitical reasons, 'chip production and the government subsidies that come with it'. This is money - not least the possible $10 billion promised by Washington to Intel - which should act as 'a cushion as the company goes through the multi-year expansion process'. Again, however, the investor must be cautious. Not least because, as the company's CEO himself says, promises of subsidies often do not turn into cash in the short term.

IL CRUSCOTTO DEL TITOLO

Loading...

IL CRUSCOTTO DEL TITOLO

Loading...

The Profit and Loss Account

.

So far, some suggestions on foundry, geopolitics and technological development. The saver, however, also looks at the fundamentals. In the last quarter of 2023, Intel's revenue was 15.4 billion (+10% year-on-year). The gross margin, for its part, rose to 48.8% (43.8%). The market, even with the final figures above estimates, reacted negatively. Why? It disappointed the outlook on the first quarter of 2024: 12.2-13.2 billion turnover and gross margin at 44.5%. More generally, it must be said, the company has put weak years behind it. In 2021, turnover was 79 billion. Then, in 2022, it dropped to 63 and, in the last financial year, settled at 54.3 billion. The net profit itself followed a downward parabola: from USD 19.9 billion (2021) it dropped to USD 8 billion (two years ago) to USD 1,675 million in 2023. In short: the numbers describe a struggling business. True! Heavy investments weigh on profitability. In addition, revenues are suffering from the end-market trend (decline in PC sales). Having said that, however, the underlying dynamic remains. Just as it remains that, in the last 12 months, the stock has risen by more than 60%. 'Investors,' explains Piva, 'have evidently been riding the artificial intelligence narrative, especially the generative one. It is a trend that, according to the Bloomberg terminal, has brought the prospective P/e of the stock on 2024 to 34.7. 'It is a value,' De Luca comments in conclusion, 'which, although the situation can always change and it is advisable to also look at other indicators, is higher than the sector average.

To go deeper:

The stock trend

Copyright reserved ©
Loading...

Brand connect

Loading...

Newsletter

Notizie e approfondimenti sugli avvenimenti politici, economici e finanziari.

Iscriviti