By Asif Jakwani

Efficient Power Consulting

www.efficientpowerconsulting.com

If you are in the semiconductor industry, you have been part of many bust-boom cycles. Each semiconductor bust-boom cycle has a unique peak and valley, which influence its duration. Proactively managing these cycles is crucial for corporate value creation. Although, current post COVID semiconductor cycle is similar in characteristic with previous ones, yet it is significantly different due to its much higher peak demand and longer duration, creating higher inventory glut in the entire supply chain and a longer recovery period. Besides over-inventory in the supply chain, additional capacity is coming online due to higher capital investments of the past few years making financial recovery elusive with lower factory utilization impacting margins and operating income.

One of the characteristics of this post COVID semiconductor boom-bust cycle is that it was coupled with very high inflation. To achieve a target inflation rate of 2%, Federal Reserve embarked on a highly restrictive fiscal policy resulting in significant demand destruction over the last couple of years. Although, restrictive fiscal policy is yielding predictable outcome on the inflation, it has created significant demand challenges. Looking back at this post COVID semiconductor cycle, one of the key learnings is that demand destruction and resulting inventory glut were expected once Federal Reserve embarked on a restrictive fiscal policy to reduce inflation. Also, peak semiconductor demand was highly exaggerated with double and triple bookings by customers and distributors resulting in over inventory throughout the channel.

I have been part of multiple of these semiconductor cycles and looking back, there were early warning signs in 2H2022/1H2023 to plan not only for short-term (adjust supply to meet true consumption) but also medium to long-term (modulate capital investments during the upcycles). Key learnings are to model macro-level fiscal policy impact to tactical and strategic actions. At times, these actions will likely be contrary to normal expected behavior, but these will yield better outcomes for the corporation once semiconductor demand enters its usual inventory correction and softer demand period.

By Asif Jakwani

Efficient Power Consulting

www.efficientpowerconsulting.com

Semiconductor market demand and supply are rarely in equilibrium.    Semiconductor bust-boom cycles are clear manifestation of this unique characteristic.   Although there is long latency in bringing new semiconductor supply to the market, primary reason of disequilibrium in demand and supply is continued penetration of semiconductor in highly disruptive new applications.  From the internet expansion of 2000s to smart phone revolution of 2010s to current decarbonization of transportation and power network to emerging Artificial Intelligence (AI) phenomena, new and disruptive applications have driven robust yet highly uneven growth in the semiconductor demand.  Although there are continued investments in additional supply, usually these are out of sync with demand due to long latency in bringing new supply.      

Although supply latency is not unique to semiconductor industry but the emergence of highly disruptive applications and resulting growth is very semiconductor specific.  In addition, these disruptive applications go through inflection and nominal adoption curve (innovators, early adopters, early majority, etc.) before reaching steady state.     These adoption and growth scenarios get further influenced by macro-economic environment, monetary and fiscal policies, regional and global political uncertainties, making semiconductor demand-supply planning extremely challenging. These additional complexities offer opportunities to companies not only to differentiate with innovative products but enhance customer and shareholder value by devising economically efficient yet resilient supply chain.  

Currently, high-level demand, supply and pricing alignment in the semiconductor industry happens during design-in/design-win phase.   Subsequently, annual volumes are finalized at the beginning of each year.  This static model of demand-supply and pricing alignment does not account for unique characteristics of new and emerging disruptive applications, high capital investments and long-latency of adding semiconductor supply.  This results in supply constraint during peak demand cycle driven by macro-economics and adoption of disruptive applications which drives over-booking, hoarding, higher capital investments and subsequent inventory glut, factory underutilization impacting financials of semiconductor companies.     Semiconductor market expect to reach over $1 trillion dollars by the end of the decade.   Robust semiconductor growth in coming decades offers an opportunity to innovators to adopt a new business model yielding a more efficient supply chain creating higher share holder value.

More on innovative demand-Supply model in coming weeks.   

A forward-looking efficient demand-supply model must rely on three key components of collaboration between semiconductor suppliers and customers.   First, long-term alignment on technology and product roadmaps that creates customer value both at device and system-level.   For customers to commit long-term growing demand, they must have the industry-leading technology and products that will enable them to offer differentiated products in their end markets.   Secondly, there must be strong collaboration between semiconductor suppliers and customers on long-term demand to ensure timely capital investments are made to bring supply of new technology and products on time.   This collaboration could be in the form of take-or-pay (TOP) long-term supply agreement (LTSA), pre-payment of future revenue on new technology, joint capital investments, consigning equipment, etc.  Semiconductor capacity is highly capital intensive and without strong and meaningful customer commitment, it is extremely challenging for BU GMs and Chief Financial Officers (CFOs) to make significant capital investments.   This financial commitment from customers is very critical to demonstrate that they have skin in the game.    

Finally, all forecasts are invariably inaccurate.  This means it will require an on-going adjustment of the forecast.   In the new model, customer and supplier must move to a 3-to-5-year demand-supply planning and commitment instead of annual volume commitment.   Secondly, there must be formal cadence on demand and supply planning and sufficient forewarning of changes in the forecast to ensure supply can be adjusted to demand.   

If the committed demand and installed supply is below the actual demand, there will be opportunities to improve supply with improving yields, increasing overall equipment efficiency (OEE), maximizing utilization, re-configuration of factory floor to maximize output, commissioning under-utilized equipment from other lines, etc.  This will drive better financials for both semiconductor suppliers and while meeting higher demand.  However, if demand is still outpacing supply and sufficient investment has not been made to meet the actual demand, this will have incurred opportunity cost to both semiconductor supplier and the customer.   However, further increase in capital investments to meet higher demand must consider macro-economic business and competitive environment to make sure it does not result in overcapacity in the future due to long latency of semiconductor supply.   

  If the committed demand is lower than the installed capacity, it presents significant challenge to the semiconductor supplier with equipment under-utilization resulting in poor financials for the corporation.   Most of the semiconductor suppliers are public companies with well publicized financial and capital investment models.   Severe under-utilization of the factories will invariably impact financials and will present significant challenge for the semiconductor suppliers.   This financial impact must be jointly shared by customers and semiconductor suppliers to demonstrate long-term commitment.   It may be possible that financial impact of under-utilization is minimized by moving other product lines, customers to newly installed equipment line.   Once capacity is committed to other customers or re-purposed to other product lines, it is impossible to make it available to the original customer, so these decisions require very close collaboration and will have meaningful consequences.      

As discussed above, demand-supply planning in semiconductor industry is extremely complicated with long latency of bringing new supply, continued yet uneven growth of new and disruptive applications, macro-economic cycles and geo-political uncertainties.  This has resulted in bust-boom cycles in the semiconductor industry resulting in capacity constrains during peak demand periods and overcapacity and factory under-utilization during softer demand.   A superior model requires a close collaboration and strategic partnership between semiconductor suppliers and customers starting with long-term technology and roadmap alignment to ensure industry leading solutions, collaboration on bringing timely new capacity with joint-investments, LTSA, and finally continued commitment to adjust supply to meet always changing demand.