octubre 11, 2023
1380
Recently, Intel publicly announced its plan to spin off its Programmable Solutions Group (PSG). Starting from January 1 next year, PSG will operate as an independent entity. PSG is the division responsible for developing products like Agilex, Stratix, and other FPGA products within Intel. Following this spin-off, Intel hopes that PSG can go public independently through an IPO within the next two to three years.
Interestingly, PSG's predecessor was Altera, which Intel acquired for a whopping $16.7 billion in 2015, marking the largest acquisition in Intel's history. The acquisition of Altera was aimed at strengthening Intel's FPGA business, and that's the origin of PSG.
Altera itself held a significant position in the FPGA market. Established in 1983, the American company was an advocate for System-on-a-Chip (SoC) solutions, offering FPGA solutions by combining programmable logic technology, IP integrated cores, and technical services for customers.
FPGAs find wide applications in fields like prototyping, communications, automotive electronics, industrial control, aerospace, data centers, and more. In the market, there are four major players, and Altera was one of them. Notably, Altera was acquired by Intel, while Xilinx, another key player, was acquired by Intel's competitor, AMD.
In addition to acquiring Altera, Intel also acquired eASIC in 2018. From Intel's perspective, the acquisition of eASIC complemented its FPGA business to better adapt to next-generation computing demands.
This acquisition significantly bolstered Intel's FPGA capabilities. According to Intel's Q2 2023 earnings call, the PSG business unit's revenue grew by 35% year-on-year, marking three consecutive quarters of record-high revenue.
Moreover, in 2022, Intel's FPGA market size reached $9 billion, with an estimated $10 billion for 2023 based on current shipment levels. These achievements are not to be underestimated, but Intel seems to be aiming for more.
Intel now intends to separate PSG and prepare it for an independent IPO. According to Intel, this move will grant PSG the autonomy and flexibility it needs, fostering its growth and enhancing its competitiveness in the FPGA industry. A successful IPO, if realized in the future, would also bring financial returns to Intel.
According to Intel's plan, PSG is expected to officially operate as an independent entity starting on January 1, 2024. Sandra Rivera, Executive Vice President and General Manager of Intel's Data Center and AI Group (DCAI), will assume the role of CEO for PSG. During this period, Rivera will continue to fulfill her role at DCAI until a replacement is found.
As per the plan, Intel will report PSG as a separate business unit in its Q1 2024 earnings report and is set to conduct the first public offering of shares for PSG over the next two to three years while retaining the majority of shares.
Intel has stated that, following PSG's independence, both companies will maintain strategic alignment, including their relationship with Intel Foundry Services (IFS), addressing critical areas of the FPGA market. Rivera believes that PSG's independence will enable it to play a leadership role in the demanding and crucial semiconductor industry, fully unlocking PSG's potential.
Clearly, PSG's independence is only a matter of time. Interestingly, earlier this year, Intel decided to sell four office buildings in San Jose, California, with a total area of 47,000 square meters, valued at $193 million. San Jose was the former headquarters of Altera. This may signify that while Altera remains in history, PSG, inheriting its legacy and bolstered by eASIC, can make a significant impact on the FPGA market.
Intel's financial reports reveal that PSG is a profitable unit within the company. In fact, earlier this year, Intel announced plans to release 15 new FPGAs, exceeding the total number of FPGA products released by Intel in previous years.
In September, Intel also announced the launch of six new products and platforms, including the expansion of the Agilex FPGA series with Agilex 3, Agilex 5, and Agilex 7. So, why spin it off now?
To understand this, let's go back to why Intel acquired Altera in the first place. Intel was willing to spend such a substantial amount to acquire Altera because it recognized the critical importance of FPGAs in the future of data center computing. Intel predicted that hyperscale and cloud builders would offload a significant amount of network, storage, and security functions to FPGA devices below the powerful CPU cores. Intel was willing to sacrifice some profit to seize this future market.
While some of these predictions have come true, with large data center customers starting to use FPGA cards for heterogeneous acceleration, it doesn't seem to have benefited Intel as much as expected. Instead, companies like NVIDIA and AMD have profited more. In July this year, AMD's market value surpassed Intel's, and it still hasn't been overtaken by Intel. The acquisition of Xilinx, now owned by AMD, played a significant role in this.
In recent years, Intel acquired several companies, including Nervana Systems, Movidius, Habana Labs, in the AI field. However, these acquisitions haven't translated into competitive products for Intel. In the AI business, Intel still relies on x86 CPU software ecosystems and libraries like MKL and MKL-DNN to maintain market share.
But these markets are under threat with products like NVIDIA's Grace Hopper superchip and the introduction of domestic CPUs. These acquisitions have not translated into substantial competitive advantages. Altera has faced a similar situation; before the acquisition, its FPGA market share was close to 40%, but it has since decreased. Meanwhile, independent Lattice has seen significant growth.
Rather than holding onto assets that cannot be effectively leveraged, Intel seems to be refocusing on its core chip design and manufacturing capabilities. This reflects the strategic goals set by Pat Gelsinger, Intel's CEO, who took office in 2021.
A notable example is the spin-off of Mobileye, the autonomous driving chip manufacturing company that Intel acquired for $15.3 billion in 2017. Mobileye went public, with its stock surging nearly 40% on the first day of trading, and its market value exceeding $23 billion. Intel later announced plans to sell approximately $1.5 billion worth of Mobileye shares to fund its chip manufacturing initiatives.
The spin-off of PSG and its eventual IPO appears to be a similar strategic move. If successful, it would provide Intel with financial resources to subsidize its core chip design and manufacturing business.
Clearly, this decision is part of Intel's broader restructuring efforts under its new CEO. In addition to spinning off PSG and Mobileye, Intel closed its RealSense computer vision technology division, sold its Datacenter Solutions Group (DSG) to Taiwan's ShenZhen Group, divested from direct investments in the Next Unit of Computing (NUC) business, transferring it to ASUS, and recently sold a 10% stake in Austrian semiconductor equipment company IMS to TSMC. Intel's determination to focus on chip design and manufacturing is evident.
For PSG, with the accumulated technologies of Altera and eASIC, a resurgence in the FPGA field is likely only a matter of time. However, Lattice may face some challenges. PSG's future development is worth watching. For now, Intel's decision to spin off PSG and prepare it
