Shulou(Shulou.com)11/24 Report--

The original title: "Intel, I lost hemp!" "

In February last year, Intel announced the acquisition of Israeli company Gaota Semiconductor at a 60 per cent premium of $53 per share, or about $5.4 billion (36.9 billion yuan). The core goal of Intel's acquisition of High Tower Semiconductor is to promote the chip foundry business at the core of the IDM 2.0 strategy.

At that time, Randhir Thakur, president of Intel contract manufacturing service IFS, said, "We are very pleased that the tower team has joined Intel." Their senior work experience and technical products will accelerate the development of IFS. Gaota Semiconductor and IFS will provide comprehensive contract manufacturing solutions around the world to meet customer needs. "

A year and a half later, in August 2023, the acquisition fell through. Intel disclosed on its website that both Intel and Gaota agreed to terminate the previously disclosed merger agreement for Gaota. Under the terms of the previous acquisition agreement, Intel will pay a reverse termination fee of $353 million to Gaota Semiconductor.

But in this acquisition, Intel "lost" much more than a $353 million "break-up fee" and a year and a half of waiting time. Thakur, then president of contract manufacturing service IFS, left Intel at the end of March and was replaced by Stuart Pann, the former head of corporate planning, rather than the rumored high-tower semiconductor CEO Russell Ellwange.

Today, Intel is still a long way from "surpassing Samsung to become the second largest foundry in the world", and the two contract manufacturing giants TSMC and Samsung have not even paid too much attention to it. After ChatGPT led the outbreak of the entire AI chip market, Nvidia and TSMC have become popular in everyone's eyes, while Intel is still the lonely blue giant.

The departure of Deere Thakur, the man who leads Intel's contract manufacturing business, is not strictly speaking this year's news.

As early as November last year, Thakur said in an interview with Nikkei Asia that Intel would "overtake Samsung to become the world's second largest fab" by 2030, shortly after it was first disclosed by The Register that it would leave in the first quarter of 2023. Intel CEO Pat Kissinger (Pat Gelsinger) said in an email:

"(Thakur) has been a key member of the executive leadership team for the past two and a half years and has held a number of senior leadership positions since joining us in 2017. he has contributed a lot to our IDM 2.0 transformation, but most notably his leadership role in supporting IFS's business. "

Pat Kissinger, Tu / Intel in March 2021, the newly appointed Pat Kissinger delivered an hour-long live speech around the world, focusing on his blueprint for Intel's transformation-IDM 2.0. IDM 2.0 consists of three main parts, namely:

-most Intel products are manufactured in-house (IDM 1.0)

-expand the capacity of third-party contract manufacturers (such as TSMC)

-to build the world's leading chip contract manufacturing business

The day after Kissinger announced IDM 2. 0, Intel immediately set up the IFS contract manufacturing business, headed by Diel Thakur. During his tenure, Thakur led the leadership team, which is made up of senior employees from leading contract manufacturers such as TSMC and Samsung, and successively won chip customers such as Qualcomm, Amazon and MediaTek, as well as some major customers in the automotive industry.

Diel Thakur, Tu / Intel in addition, Thakur also led Intel's acquisition of Israeli chipmaker Gaota Semiconductor, which specializes in making high-value semiconductor components for the automotive, industrial, consumer, aerospace and defense markets. The acquisition is expected to add at least $1.5 billion a year to Intel's contract manufacturing revenue. But for Kissinger, the core reason for the acquisition is:

Intel needs more Foundry (foundry) DNA.

In other words, Intel needs more team members who know more about the contract manufacturing business, so it's not surprising that after the news of Thakur's departure, there were reports that Intel would seek to let Gaota Semiconductor CEO Russell Ellwange, which has more experience in contract manufacturing, take over Intel's contract services division.

But the reality is, in the end, one person leaves, the other person also "has no chance" to come together.

Surpassing Samsung becomes more difficult as early as the end of last year, in the article "Intel OEM business wants to catch up with TSMC Samsung, the biggest resistance comes from within", Lei Technology pointed out that the problem with IDM 2.0 is that this is an extremely complex and difficult business model.

Kissinger hopes that IDM 2.0 can achieve the result of "IDM (vertically integrated manufacturing) makes IFS better and IFS makes IDM better", but the actual process needs to balance the rights and interests of internal chip design departments and external chip companies all the time.

At the time, we also cited the core reasons why technology commentator Ben Thompson called for Intel to be split:

The design and manufacture of integrated chips has been Intel's moat for decades, but this kind of integration has become a constraint to each other. The design department has been held back by manufacturing factors such as backward process, and the manufacturing department has no pressure or motivation to take the lead in technology. "

In June of this year, Intel finally announced that it would spin off its contract manufacturing business and plan to operate independently in the first quarter of 2024. The earnings report will list the profit and loss separately, and it is expected to start making a profit by then. The goal remains the same, still surpassing Samsung to become the second largest fab in the world.

After the figure / Intel split, the most direct benefit on the one hand is that Intel's design department also has to pay corresponding fees to IFS, on the other hand, third-party customers, especially those who directly compete with Intel chip products, will also reduce their worries about "unfair treatment" of IFS contract manufacturing services.

Soon after, the spin-off also aims to free IFS's contract manufacturing business and become the main body of a more contract manufacturer, DNA, so as to achieve IFS's goal of surpassing Samsung and the vision set by IDM 2.0.

In contrast, the acquisition of Gaota Semiconductor could have been a big boost to IFS's contract manufacturing business, whether it was the expansion of contract revenue and customers, or members of the contract manufacturer DNA. But in the end, the termination of the acquisition also means that Intel needs to face greater challenges and more investment if it wants to achieve its original goals.

In the same month that the split was announced, Intel announced a series of foundry investments around the world: investing $4.6 billion in a new semiconductor assembly and testing plant in Poland; significantly increasing the overall investment in German fab (originally scheduled to be 17 billion euros); and spending $25 billion on a new plant in Israel.

For Intel, this is a necessary investment that has to be made, but perhaps it has foreseen the outcome of the acquisition of High Tower Semiconductor, or the huge possibility and change of the entire chip foundry around AI, so that Intel needs to put more chips on.

Intel wants to seize AI since the beginning of last year, the consumer electronics market has been in a cold winter, and the chill has spread to the upper reaches of the industry, including chip makers and contract manufacturers, with the exception of AI.

Although the recovery in the PC chip market helped Intel turn a profit in the second quarter of this year, revenue from IFS contract manufacturing still accounts for only 1% of revenue, and revenue from the most profitable data centers is expected to continue to shrink. Kissinger admitted that the data center business performance was expected to be weaker in the third quarter as major customers' spending was concentrated on AI.

To some extent, Kissinger acknowledged Intel's weakness in AI chips and contract manufacturing.

In the past, Intel focused on CPU business, and it was only last year that GPU officially launched an independent graphics card. There are problems in the emphasis and direction of AI. For example, Meteor Lake's approach to AI applications is to integrate CPU, GPU and VPU (for AI acceleration) to participate in AI accelerated computing. There are great doubts about efficiency, but it is still because Intel does not see the importance of AI computing power in advance.

Also in AI chip manufacturing, TSMC CoWoS (a 2.5D packaging technology) has become the core competitiveness. Intel and Samsung both have similar packaging technology, but both have obvious backwardness, coupled with the gap in the most advanced process. All these actually constitute the biggest challenge for IFS contract manufacturing business and even Intel at present.

As the top priority of Intel IDM 2.0 strategy, the success of contract manufacturing business essentially determines the success of IDM 2.0 and the result of Intel's transformation. In other words, Intel might have lost everything if it hadn't ridden the wave of AI chips.

This article comes from the official account of Wechat: Lei Technology (ID:leitech), author: Lei Tech Digital 3C Group

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.