July 10, 2023
About the Company
Nanya is a small Taiwanese DRAM manufacturer. They produce DRAM on trailing-edge process nodes (licensed from Micron) but are developing their own more advanced, technology. Nanya has about a 3% share of the DRAM market, a distant fifth, behind Samsung, Hynix, and Micron. The top three companies control a combined 94% of the DRAM market.
I don’t own any shares of Nanya and would not buy any, because they lack the scale and the technology to be competitive in DRAM. I keep an eye on their quarterly releases as a data point on the DRAM market.
Market Commentary and Q1 2023 Performance
The PC segment, the dominant market for Nanya’s products, entered oversupply first, so it should be the first segment to emerge. Therefore, Nanya is a leading indicator for the DRAM market. If that is the case, this quarter was a good sign for Micron shareholders. Net sales were up by 9.4% sequentially. This happened because shipments increased mid-teens %, more than enough to offset the mid-single digits % reduction in ASPs. Demand is starting to recover and price declines look to be approaching bottom. For the last six quarters, going back from Q2-23 (from newest to oldest), ASP declines have been: mid-single digits, high single digits %, mid-20s%, low-20s%, mid-single-digits %, mid-single-digits %. Bit shipments strengthened recently but are down low-twenties % from a year ago. That is pretty stunning. Nanya has reduced the number of DRAM bits it is shipping by more than 20% from a year ago. Gross margin was (11.2)%, a small erosion from the (8.6)% last quarter. The company didn’t include their gross margin graph for the second time this quarter. I bet when things turn back up that graph will reappear. It’s going to be glorious when it does.
Bit Shipments and Capital Expenditures
Nanya reduced their 2023 capital expenditures forecast from NT$18.5B (stated last quarter) to NT$15.0B (stated this quarter). This is a 28% reduction year-over-year. Approximately 50% of these expenditures will be for equipment. That is a lower percentage than the fraction of equipment spending in a full fab. This indicates that the company is building out the shell of their new fab but not filling much of it with equipment. In the second half of 2023, the company said production will be “dynamically reduced” by up to 20%. This is the same statement they made last quarter. It means they will dial up output when demand is there. The company does not have the cash margin for error that it’s big competitors do, so they will probably turn fab production up sooner, for the cash flow. Free cash flow in Q2 was negative NT$6.54B, an improvement from the negative NT$8.03B of free cash flow last quarter. Capital expenditures were cut in half. This fully offset the higher loss from operating cash flow.
2023 Market Outlook
They are calling 2Q23 to be the bottom of this DRAM cycle for demand. “Marginal to moderate demand rebound is expected in 2H’23.” Nanya expects “improvement” in supply-demand balance in 4Q23. That is a meaningless statement. A decline in the rate of ASP erosion indicates an improvement in supply-demand balance. Nobody knows anything and Nanya executives are at the front of that line.
- · Server Market: The company checked the box that they used the term “AI” in this deck, making the obvious statement that AI is favorable for server demand. Enterprise server demand is still weak and recovery here is key to improving the market.
- · Mobile Market: Demand in China will be key to recovery.
- · PC Market: Inventory gradually returns to normal levels. Shipments in the second half of 2023 are expected to be better than the first half.
- · Consumer Market: Demand for IP CAM, networking, industrial, and automotive applications are relatively healthy. This is a copy-paste from last quarter. As I said last quarter, DRAM softness is a virus that spreads to every sector eventually. It could be their exposure to these segments is small enough that they didn’t see the softness that other companies did. It is getting late in the cycle for any segment to still be unaffected.
Analyst Call
Prepared Remarks
· Nothing worth recording. The CEO mostly just read the slides to the audience.
Analyst Q&A
- · The market has bottomed in the second quarter. Shipments have recovered somewhat. While the recovery in Q2 is helpful, the decline has been so severe in the last year that they need to see more strength and momentum to feel good.
- · They see Q3 to be a mixed period and hopefully Q4 will be a balanced DRAM market.
- · Most of the capital expenditures reductions have been in equipment delays. They have not changed their fab construction schedule very much.
- · Q3 will “likely” have an improved blended ASP compared to Q2. This is a bold statement. DDR5 will likely improve before DDR4. This is because there is less inventory in DDR5 parts to consume. Qualify this prediction with the call the CEO made last quarter that the pricing inflection point would either be late Q2 or early Q3. We are in early Q3 and, while he didn’t say pricing had turned, he did predict the blended ASPs for the whole Q3 would be better than Q2. He could still turn out to be correct.
- · Bit shipments in Q3 are expected to be higher than shipments in Q2. Combined with their statement that blended ASPs will also increase, if these two forecasts come to pass, then Nanya’s Q3 results will clearly indicate that the downturn is over and the recovery in the DRAM market is underway.
- · DDR5 is “very minimum” fraction of their total DRAM production.
- · They believe, in the PC market, the customer inventories are not as much of a risk as the supplier inventories are.
Summary
I am tough on the Nanya CEO and the company predictions, with good reason. They turn out to be wrong a lot, more than most in memory. The theme of their inaccuracies is optimism on when the memory market will improve. With that caveat, the CEO made the prediction that pricing, as measured by blended DRAM ASPs, will be better in the third quarter than in the second. They also said demand will be stronger sequentially. The fact that demand did improve from Q1 to Q2, along with a small decrease in ASPs, gives this prediction credibility. Since PC DRAM went into oversupply first, it will likely be the first segment to recover.
-S. Hughes (cyclical long MU)