Skip to content

Featured

Capex Malaise and ‘MisCapex’ Pains

Capex Malaise and ‘MisCapex’ Pains

By Bolaji Ojo

What’s at stake:

Semiconductor makers perform better when production capacity utilization rates and the number of plants closely match customer demands. Those periods are getting fewer, however. Despite the proliferation of advanced demand-supply management tools, the industry continues to experience forecast distortions, especially in the number of semiconductor fabs needed to meet future demand. We describe the result as “MisCapex,” a condition of overestimating or underestimating capital expenditure requirements, leading to either bloated production or undercapacity. In either case, the remedies used so far by the industry continue to yield the opposite of desired objectives.

The semiconductor industry goes into a fab building frenzy every few years, driven by undercapacity that eventually leaves chipmakers dried out financially or swimming in the murk of excess inventories.

The last time this happened was only a few years ago.

Electronics companies reacted differently to the last bout of shortages. Some made plans to build more fabs while many others quietly watched the situation unfold, preferring to maintain their conservative capex planning. Today, companies in the latter category look prescient while others are dialing back on their commitments. Intel Corp., for example, is reviewing its capex commitments under new CEO Lip-Bu Tan and sharply cutting back expenses related to capital equipment and operations, including R&D.

In the face of more pressing and troubling issues, the semiconductor industry is trying hard to ignore its recent financial past. Yet, factors related to capital expenditure cannot be swept under the carpet because they are so fundamental to the health of the sector. The industry is finding itself locked in the grips of capex promises made at the height of the last supply shortages. Promises made to customers, consumers and even politicians will soon come back to haunt the industry. The Ojo-Yoshida Report is getting ahead of that Day of Reckoning by examining statements about and commitments to fab construction a few years ago and asking: Are the fabs promised during the last shortages still needed; How many of the fabs will be built and; how will chipmakers finance the ones they cannot walk away from?

Read More »Capex Malaise and ‘MisCapex’ Pains
Intel: Will Lip-Bu Tan Build Gelsinger’s Promised Fabs?

Intel: Will Lip-Bu Tan Build Gelsinger’s Promised Fabs?

By Bolaji Ojo

What’s at stake:

Lip-Bu Tan’s predecessor as Intel CEO wanted to build multiple new fabs in Europe and US. Chip equipment makers, fab shell builders, political leaders and communities were counting on the planned $100 billion-plus capex splurge. Will Intel revive Gelsinger’s dreams or were they so implausible from the beginning that it’s a risk Intel must avoid?

Lip-Bu Tan’s comments at his first analysts’ conference call as CEO at Intel Corp. included an indictment of the tenure of Patrick Gelsinger, his predecessor, and even some prior leaders of the chipmaker.

Tan didn’t mention any of Intel’s five past substantive CEOs – Gelsinger, Bob Swan, Brian Krzanich, Paul Otellini and Craig Barrett – by name. He may not have meant to indict these people, but the scathing verdict delivered by Tan about what Intel has become represents a condemnation of the company’s previous leadership.

“One of my biggest learning so far is that we need to fundamentally transform our culture and the way in which we operate,” Tan said. “Organizational complexity and bureaucracies have been suffocating the innovation and agility we need to win.” Intel had developed a “siloed” operating system,” Tan said. “I’m here to fix this.”

There is a lot hanging on Tan delivering that “fix.” For this report, though, we will focus on fabrication and process development plans made by his predecessor Gelsinger, who promised to help fabless chipmakers and governments in America and Europe resolve their dependence on semiconductor fabs based in Asia. During Gelsinger’s tenure, Intel announced plans to spend a range of $100 billion to $200 billion on new fabs and back-end processing facilities, most of them to be based in the United States. What will be the fate of Gelsinger’s promised fabs, the sites selected in Ohio, Magdeburg, Germany, and plans to help fund semiconductor education at community colleges in the United States?

Read More »Intel: Will Lip-Bu Tan Build Gelsinger’s Promised Fabs?

Infineon Bulks Up with $2.5B Marvell Ethernet Deal. Expect More

By Bolaji Ojo

What’s at stake:

Defying market turmoil, Infineon is proceeding with its growth plan. It is buying Marvell’s automotive ethernet division to beef up its software defined vehicle (SDV) and robotics offerings, but Infineon may be angling for more as it morphs from one of Europe’s top chipmakers into an even bigger global leader in other market sectors.

Two months ago, Infineon Technologies AG quietly arranged a €2 billion ($2.7 billion) revolving line of credit without disclosing why, noting only that it was benefitting from the “the strong trust” it enjoys in the banking industry.

Industry observers assumed Infineon was filling up its cash reserves to pay down debts. However, as the Ojo-Yoshida Report observed in a LinkedIn post, the company had other plans for the funds. On Monday, Infineon cleared the air, saying it offered $2.5 billion for Marvell Technologies’ automotive ethernet business. The deal, Infineon said, will be “financed from existing liquidity and additional debt.”

“This [Marvel] business holds a number one position in automotive Ethernet and offers a complete portfolio that is fully complementary to our own offering,” said Jochen Hanebeck, CEO of Infineon, during a conference call with analysts. “This deal will enable us to accelerate the transition to software defined vehicles by advancing zonal architectures based on Ethernet networks.”

Read More »Infineon Bulks Up with $2.5B Marvell Ethernet Deal. Expect More