STRATEGICALLY SPEAKING: How many MOCVD reactors is too many?

Jan. 26, 2011
An unprecedented level of orders for MOCVD reactors could create a glut in the LED market, but is unlikely to have much effect on high-end LED suppliers, says TOM HAUSKEN and Strategies Unlimited staff.

Insights Into LEDs & Lighting, from Strategies Unlimited


In early January, rumors circulated that Golden Concord Holdings in Hong Kong plans to invest $2.5 billion in LED manufacturing, including the purchase of 500 metal-organic chemical vapor deposition (MOCVD) reactors. That is on top of news of several companies buying more than 100 reactors each, all for use in China. There have never been orders for so many reactors in the history of MOCVD. And all these reactors will be used to make LEDs.

What’s sure is that it’s a bubble. What’s not as clear is just how big the bubble is.

The LED world needs more and better reactors. Older reactors are not cost-effective for the capacity needed to sustain the unprecedented growth in LED chip production, especially for LCD TV and monitor backlighting, and LED lighting.

We estimate that approximately 300-400 new reactors per year are needed through at least 2013 to meet the projected growth in LED demand. Actual reactor shipments for LED manufacturing went from about 250 in 2009 to nearly 750 in 2010, and may go even higher in 2011. Consequently, we believe that reactor shipments are exceeding even the rapid growth that the industry needs.

Some of the excess reactor ordering stems from the usual speculation: no one wants to get caught short of capacity, so some companies bet big.

Subsidies in China also encourage excess ordering. As anywhere else in the world, regional governments in China want the next “LED Valley” to be in their region. Many offer subsidies of as much as 8-10 million yuan (about $1.2 to $1.5 million) toward MOCVD reactor purchases, which typically cost around $2.5 million each for the current generation of high-capacity systems. Some subsidies are paid in cash, others are paid through tax discounts and other benefits. Some require that milestones be met.

The subsidies have influenced location decisions: they have attracted Taiwanese and Korean companies to build new joint-venture plants in China, some even locating in multiple regions in order to collect multiple subsidies. The subsidies also lower the barriers for local companies entering the LED business. Altogether, subsidies could end up supporting purchases of as many as 1,200 reactors, unless Chinese policymakers put on the brakes.

For the record, I wish everyone well, especially the reactor vendors and other capital equipment vendors. Selling capital equipment is a tough business, cycling strongly with industry ups and downs. It’s not necessarily the reactor vendors’ concern if the industry is overinvesting in reactors.

Reactor vendors could be faced with a deep cycle if the subsidies accelerate orders so much that the overcapacity leads to a stall in orders, perhaps to the point that companies fail and used reactors go back onto the market. Used reactors may be difficult to sell, though; the bargain may not justify the requalification necessary to fit a different process. In any case, some overcapacity may be okay for reactor vendors, as long as it’s not so much that their business swings into a deep cycle after the bubble bursts. Even then, that’s the business they are in: it’s all about riding the cycles.

It’s highly likely, however, that many of the reactors won’t contribute to worldwide LED production within the next five years. It’s hard enough to make functional LEDs, much less fully-specified LEDs. Just having an MOCVD reactor is not enough.

In fact, countering the surplus of reactors is a shortage of MOCVD engineers. There’s talk that an expert grower in China today can earn a $250,000 salary. Like CEOs and rock stars, there are only so many good ones for hire, and the compensation reflects that reality.

Consequently, none of this is likely to impact the market for high-end LEDs, which are the most difficult to make. The high-end suppliers know their process well, and are likely to expand capacity regardless of subsidies. The subsidies merely affect the “when, where, and how much,” not “if.”

It could be worse for suppliers of low-end LEDs, those with the loosest specs, sometimes even selling by the kilogram. The extra fab capacity could create a glut of these chips, forcing other suppliers to drop prices or even exit that market.

So who wins and who loses from the bubble?

The end-user of LED-based products benefits most. If anything, the excess manufacturing capacity puts downward pressure on LED prices, saving the customer money.

China also wins, no matter what. The subsidies are already partially successful; they have attracted foreign companies to expand capacity in China. That means employment and some transfer of technology. It’s a giant LED industry stimulus for the “University of China.” What is a few billion dollars in investment in LEDs for a country with money in the bank, anxious to move faster toward its own green energy future? We’ve seen how well China played its industrial policy hand in the photovoltaics and wind-energy industries. This is a similar game.

The bubble may turn out to be less than anticipated. There is talk that Chinese regional governments are pulling back on the subsidies, letting some air out of the bubble. If the subsidies aren’t withdrawn altogether, new policies may require that reactors be installed by mid-2011 (or at the latest, the end of the year) in order to receive a subsidy.

What is sure is that many players will make some money. The end-user gains, no matter what. China gains some experience that it didn’t have, and more domestic production to serve its vision for a greener future. If overcapacity leads to a glut of LEDs, the most pressure will be on lower-tier LED suppliers, possibly leading to a shake out. But if it doesn’t get too bad, very few investors will be stuck with expensive paperweights.