Chartbook 306 Nodes, rebar and private equity- How Intel, the weak link in the chip strategy of Bidenomics, is resorting to financial engineering to raise billions for fabs.
AUG 7
Amidst the market focus on AI, Nvidia and the fab 7, and the political cycle which is directing attention towards VP-picks and rallies, I worry that a big story of last week may get swept under the rug: the catastrophic news out of Intel.
Last week Intel, announced 15,000 job cuts, and suspended a dividend that has been in place since 1992. As Bloomberg reported, Wall Street is “losing patience”. “Investors are pulling the kill switch.”
The stock has plunged to levels not seen in a decade.
Why does this matter?
Not because the world is short of the kind of chips that intel is struggling to make. On the contrary, we are just slowly exiting a cyclical glut. If Intel does manage to establish itself as a chip designer and a significant maker of chips (fab operator) that will offer more choice to buyers. But this is not why Intel really matters.
Why Intel matters is geopolitics and the new political economy of industrial policy.
In a world of geoeconomics, decoupling, derisking and disentangling from China, Intel is claiming the role as a national champion. As Susannah Glickman explains in this excellent piece for Phenomenal World Intel has since the late 1960s been at heart of the state-business nexus that defines chip manufacturing and the political economy of high-tech everywhere around the world.
In the last few years under the sign of the new industrial policy, Intel has attracted giant subsidies both from the US, via the CHIPS act, and in Germany. All told, the subsidies committed to Intel around the world, at least on paper, run to more than $20 billion, for one, private company, not in lguarantees, but in hard cash.
Back in 2023 when the industrial policy hype was at its max, I was a sceptic. I did a bunch of TV and other interviews in Germany questioning the logic of Berlin’s public subsidy for Intel. What I was worried about was:
- The staggering capital intensity of the industry. This is a sector where private companies will spend over $10 billion per quarter in capex. Source: Semiconductor IntelligenceIn an industry operating on this scale and at this pace, is the public balance sheet big enough to play? I know that sounds like a crazy question, but the CHIPS act allocates .. drumroll … $39 billion in subsidies for creating or expanding U.S.-based semiconductor fabs and an additional $11 billion in research and development. And the money takes many, many months to disburse. In the world of semiconductors that is the spending of one player for one year. For Germany, hobbled by the debt brake, this is very deep water to be playing in.
- The semiconductor industry, an ultra high-tech commodity sector with long lead times, is dominated by the vicious swings of a classic commodity cycle (think “hog cycle”). In 2023 policy-makers were harking back to the post-COVID chip shortage, whereas at that very moment we were heading into an industry wide downturn and glut. Again, the industry moves far faster than politics. Source: WSTS
- Furthermore, industrial policy makes most sense at an early stage where it can hope to exercise something like a formative influence, or in industries with many players and plenty of opportunity for benchmarking and competition. The global semi-conductor industry, by contrast, is a vast oligopoly in which the winning and losing hands were already distributed. Taxpayer-funded industrial policy has virtually no chance to creatively reshape the industry. Anyone wanting to play the chip game on a substantial scale has to get in bed with the big players. And amongst those, intel is the big loser.
- Intel was once a world leader. It formed a combination with Microsoft and IBM in the 1980s which shaped the early stages of desktop IT. But in the 2000s it missed the boat on chips for cellphones. Then on the fab side it made fundamental missteps in manufacturing. And then it missed the boat on AI. So now it is playing catchup on all fronts. On chip design, as the Economist notes, Intel is miles behind:This year it expects to sell $500m of its Gaudi ai chips. Nvidia sells $20bn of its ai chips each quarter. What is more, success in the market for ai chips is about more than the chips themselves. Nvidia sells networking gear that ties together hundreds or thousands of its processors. It also has cuda, a software platform, which allows customers to fine-tune the chips.
- Finally, more superficially, Intel’s leadership are no doubt highly skilled engineers, but the current CEO is perhaps best described as “weird”. Back in 2022 in seeking to goad Congress into acting faster on the CHIPS act, he remarked about the EU: “This complex 27-member socialist union . . . is now ahead of the US by a solid six months.” Unless this was intended tongue in cheek, and that does not seem like Pat Gelsinger’s style, this is someone who does not understand that he does not understand the people he is dealing with.
Amidst the hoopla of securing the Intel commitment to Germany, I though a fragile coalition in Berlin was underestimating the risks of getting into bed with this firm. I thought the same was probably true for Bidenomics too Washington, but Washington, at least potentially, has resources to play in the big leagues. Intel is an American company. Added to which, Intel was promising to build the “silicon heartland” in Ohio and, you know, when it came to the heartland there was no arguing with the Bidenomics people.
Online in press releases from a few months back, Intel pitches a story of global expansion and its range of partners.
In the US, we are expanding our existing operations in Arizona, New Mexico, and Oregon, and investing in two new leading-edge chip factories in Ohio. We have submitted all four of our major project proposals in Arizona, New Mexico, Ohio, and Oregon to the US Department of Commerce’s CHIPS Program Office. These projects are estimated to represent over $100 billion of US manufacturing and research investments over the next f ive years. In the EU and Israel, we have announced a series of investments spanning our existing operations in Ireland and Israel, as well as a planned investment of more than $33 billion in Germany to build a leading-edge wafer fabrication mega-site. We have also announced our plans to invest up to $4.6 billion in an assembly and test facility in Poland, and the start of high-volume manufacturing using Intel 4 technology and extreme ultraviolet (EUV) technology in Ireland. … Intel teams across the globe are installing new tools, delivering new clean rooms, and completing construction of new buildings. To give an idea of the sheer scale of operations, in 2023, about 145,000 tons of steel—roughly the weight of 20,743 African elephants—was used in our construction and expansion projects, Our construction teams also poured over 2 million cubic yards of concrete—enough to build the Empire State Building 32 times over.
Source: Intel
But, all this talk about factories and machinery and steel and concrete, cannot hide the fact that twelve months on from the German announcement and two years on from the CHIPS act, in terms of actually mass-producing ultra high-end chips that are competitive at the global level and able to generate profit, Intel is way behind its competitors and the gap will not close anytime soon.
CEO Gelsinger tells a story to analysts of building two world-class companies: the Fab side and the product-development side. The product designer will be able to choose not just intel production facilities, but fabs anywhere in the world. The fabs will make not just intel chips, but will be open for business from all comers, which makes them particularly attractive from a US strategy point of view.
But as one Timothy Prickett Morganremarks, this may be a good patter but, “that is a lot calmer than the situation really is. This is going to be hard to watch over the next couple of years, and even harder to do.”.
The results that caused the stock value to crash tell the story.
Here ends the Chartbook item
BLUNDELL
I did set out to pull down a paragraph for that magnanimous pusillanimous awesome short essay + photo-essays Little🚀🧨🚀🧨🚀🧨🚀Man @X account – mine doesn’t get stuck up Concentric Economic ⭕️ 10 – Blundell 1995 Australia 🦘🌏 – GEA.
But several times a day industrious professional writers everywhere are irritated if not annoyed – some of us even mentally fit to ‘kill’ – about the time-wasting – as much 2 or 5 minutes an hour – overlaid, or to use a mechanical metaphor (hydraulics) fancy-arsed but data-jumble incident ridden monumentally useless ‘5G’ information ‘servo systems’ whose only known or ever demonstable purposes are to sell deskilled kids crap product and even more execrable deracinated – unrooted in lived experience or regional socio-economic-cultural or political transactions. So to end the long macro story for some micro WHAMMO the machine wouldn’t let me neatly excise a carefully selected say 220 Character paragraph easily so i copied the whole ‘shooting match’ and published it here AS A GREAT PROMO FOR THE AUTHENTIC HISTORIAN GUY❗️
John Blundell
Post Socratic Logic, Neurolinguistics, VG Childe’s 1940’s ‘household’, or oikonomia (Greek οἰκονομία), usually translated as MICRO ‘governance’ which he spelled OEkomenia I recall & Philosophy of Science
Billionaire-🙈🙊🙉🐒s with Olivettis 