Share This article
AMD released its Q2 2015 results today after warning last week that its figures would miss initial expectations. The results for the company’s APU business were brutal. Total company revenue was $ 935 million, down 8% from Q1 and 35% year-on-year. That decline was driven by decreased sales of both APUs and graphics solutions. Gross margin was 25%, down 7 percentage points from Q1.
The principle driver of the decline was the falling volume of APU sales and a one-time charge of $ 33 million related to non-production of 20nm hardware and the need to shift these designs to 14nm FinFET technology. As we’ve previously discussed, it’s not AMD’s fault that the 20nm node didn’t pan out — that one is on TSMC and GlobalFoundries — but it further hit the beleagured company. Excluding that charge would’ve left AMD with a gross margin of 28%.
AMD reported an operating loss of $ 137 million and a net loss of $ 181 million, with $ 829 million in cash-on-hand and $ 2.27 billion in debt. That $ 820 million dollars is still above the $ 600 million cash level AMD has set as its preferred minimum (executives today indicated that the company could manage with an unspecified lower amount of money), though below the $ 1 billion ideal target that the company had previously set. Going forward into Q3, AMD expects revenue to increase sequentially by 6% (plus or minus 3%) with a non-GAAP gross margin of 29% and a cash and cash-equivalents level of $ 700 million.
If AMD holds that level, it’ll be edging close to the minimum values it previously set for itself. The rumors that Lisa Su was meeting with investors and financial groups suddenly make a great deal more sense. Due to shifts in the PC market, the company no longer expects to return to profitability in the back half of the year and has cautioned that its outlook for Q3 is heavily qualified on the behavior of the volatile PC market.
Searching for a bright spot
Financial sheets normally make for poor graphics, but this one needs to be seen.
It’s shocking how quickly AMD’s APU business has essentially disintegrated. Rory Read promised that AMD would reach a break-even point of deriving 50% of its revenue from new sources, and 50% from existing business. That’s completely flip-flopped, now — AMD’s embedded, semi-custom, and enterprise segments (which basically means the PS4 and Xbox One) are responsible for 60% of the company’s revenue.
It wasn’t long ago that AMD’s graphics business was worth nearly $ 300 million a quarter, much less its APU and GPU business combined. AMD reported that it won a new embedded customer in Q2, though it gave no details on when it might recognize revenue from the win or how much revenue it expected from the new business. AMD did report that it believes its channel inventory has been cleared out, which could help move products in the second half of the year.
Demand for consoles remains healthy, which will continue to provide AMD with absolutely critically needed revenue, but the state of the decline in the APU and graphics business suggests that more is badly needed. Unfortunately, it’s not clear how much more AMD has to give in the short-term. Fury and Fury X both launched in Q2, and we’d expect AMD to see some reinvigorated sales from both GPUs, but discrete graphics was never a huge profit driver for the company, and its operating income for discrete sales (back when it separated them from CPU sales) was typically in the 6-9% range. Fury and Fury X could offer some relief, as could the upcoming Fury Nano and dual-GPU Fury solution, but they aren’t going to plug a $ 147 million hole in APU profits.
According to Lisa Su, sales of APUs were driven down by slowing notebook sales in advance of the Windows 10 launch. That makes sense — plenty of shoppers might choose to wait to see what new systems will arrive, including for back-to-school — but it’s not clear how much rebound we can expect in this segment. Su noted that there are 35 wins for Carrizo coming with Windows 10 and expects a rebound in Q3. Desktop chips, A10 APUs, and Fury X were all noted as being a positive impact on Q2, though ramp was ongoing for the new video cards. Seattle, AMD’s Cortex-A57 server CPU, is expected to finally launch in the back half of the year.
Analysts asked if the wafer-supply agreement with GlobalFoundries would be fulfilled according to the $ 1 billion target AMD previously set. AMD indicated it had only spent ~$ 400 million on wafer purchases to-date, but that it did not expect to take an additional financial hit from doing so.
AMD has previously said that it expects to begin recognizing income from additional embedded deals starting in 2016 and that’s welcome news indeed. Most publications, including this one, have assumed that Zen and the follow-up family of HBM-equipped Zen-based APUs will launch in mid-2016 (for Zen CPUs) with APUs possibly appearing by holiday 2016 or early 2017. Those are our own estimates, not AMD’s, but whenever the launches come, they can’t come quickly enough. Intel’s announcement that its own 10nm hardware will be delayed to 2017 opens the door for AMD to compete against Intel on 14nm with Zen vs. Kaby Lake. Su noted that AMD just finished taping out its first few 14nm designs, which would put them on track for a Q2 – Q3 2016 launch — though she didn’t explicitly say which 14nm products had taped out.
This entry passed through the Full-Text RSS service – if this is your content and you’re reading it on someone else’s site, please read the FAQ at fivefilters.org/content-only/faq.php#publishers.