Slashdot Mirror
AMD, Which Lost Over $2.8 Billion In 5 Years, Takes a Hit After New Report (arstechnica.com)
An anonymous reader quotes a report from Ars Technica: On Monday, AMD's stock price plunged nearly 9 percent after a report by Morgan Stanley, a major investment bank, which found that "microprocessor momentum" has slowed. According to CNBC, a new report by analyst Joseph Moore found that "cryptocurrency mining driven sales for AMD's graphics chips will decline by 50 percent next year or a $250 million decline in revenue. He also forecasts video game console demand will decline by 5.5 percent in 2018." As per AMD's own SEC filings, the company lost over $2.8 billion from 2012 through 2016. However, new releases from AMD suggest that it may be on something of a resurgent track. As Ars reported last month, AMD's Ryzen and Threadripper processors re-established AMD's chips as competitive with Intel's.
AMD had a fantastic Q3 and predicted a slower Q4 (as expected), and the stock has fallen a ton in the past few days. It really makes no sense.
Well, the stock has gone from $2 to almost $15 and then slumped back to $11 so it's not like the market has really lost faith but the expectations for AMDs recovery were maybe exaggerated. Right now AMD is probably billing a lot of semi-custom revenue for SoC that go into the launch of Xbox One X and the PS4 Pro was last year, next year will be a slow year. Also Q3 is generally when Microsoft/Sony buy chips for their Christmas sales. Ryzen and TR is doing well in some markets, but Intel has pretty aggressively slashed prices to close the launch gap and enterprise customers take a long time to trust AMD EPYC servers.
Vega is an okay response to Pascal as a GPU but it's better as a crypto-mining card, the question is how quickly does the Ethereum market move to FGPA/ASIC custom chips like Bitcoin has, will the ICO bubble pop and so there's a lot of risk there. And don't forget that AMD has sold off and licensed a ton of assets and got a pretty high debt burden. And the PC market is in a general slump, nothing their fault but it doesn't help. With Zen they've put enough product on the table that they seem to be out of immediate danger and breaking even, but I still think they have a long way to go towards stable growth and profit.
Live today, because you never know what tomorrow brings
The Jump from $2 to $15 priced in all the growth you're talking about. In fact it priced in more growth assumptions. If AMD doesn't hit all the targets that got priced in (look to the consensus earning estimates made last quarter that covers the next year) the price is going to fall.
EPYC is still on full allocation to cloud providers and OEMs only. It's impossible to buy on the individual market. This makes it apparent that it's popular but how popular we won't know until AMD tells us how many they sold, it could be simply that their production runs have not been large enough to tackle demand and that's a scary though considering we a more than a full quarter away from when production started. I'm not encouraged by the fact that Supermicro hasn't bothered finalizing their single core motherboard designs. I'm willing to bet they are having production issues and they haven't sold anywhere near what everyone thought they would. For example if the 0.08cent per share earnings that was the consensus for 4Q (I'm going off memory here not verifying, my intent is the scale not the exact figure) ends up being half that you'll see the price fall to $7.
Personally I believe AMD is missing a critical envelope here in the individual server market where I think they would do relatively decent on the linux server market. They trying to ensure Cloud computing and OEM's get first pick. I understand why they did this, they wanted the companies that will absolutely love their high core counts and expanded PCIe bus to get first shot and it helps the cloud providers are happy to buy thousands of these at a time and can fully utilize them right out of the door because their clouds run Linux (ie they won't be waiting on Microsoft to update the windows kernel). But if Google, Facebook and Rackspace get all the production for the next 6 months there never will be an individual market for these chips where they could build some lasting execution. The cloud companies are happy to jump around and buy from anyone who can provide the most compute per dollar per watt. You don't get long term purchases in that market, where in the individual market you'll build brand awareness and are far more likely to get return purchases. The PCIe advantage in particular I believe could draw off a significant number of Intel sales in the individual server market and benefit AMD for years to come.
In the end all that's going to matter is can AMD remain competitive with Intel and avoid missteps while continuing to execute. They've had bad management IMO the last couple decades, pursuing fads with no long term vision and significantly overpaying their executives. Only by cleaning out the MBA's and stocking management with engineering experience and a desire to compete do they stand a chance here. They need to start executing and planning to remain competitive with Intel. Ultimately this will decide if Bankruptcy eventually takes them as that $2 stock price portended but Ryzen barely avoided.