Mark Pickavance looks at the sorry financial state of AMD and wonders how much longer this can go on
Losses for AMD are something of a regular occurrence, but even its most ardent supporters drew breath when AMD recently announced its straight fourth quarterly loss.
The third quarter for 2015 saw the plucky chip maker blow another $197 million (£127m), bringing the total for this year to $557 million (£360m). Revenue was $1.06bn, so for every pound AMD spent, it only made about 80 pence back. If there is an upside to any of this, it’s that revenue was only expected to be $995m, so that was marginally better than expected.
What’s colouring lots of views on AMD, though, is its track record over the last 17 years, but balancing those good times against the less wonderful still leaves it an eye-watering $8bn in the hole.
Even ignoring the lengthy history of losing money, what concerns many analysts is that year on year, things aren’t improving; they’re actually getting worse. Year on year, revenue for AMD has declined by a whopping 26%, and should that continue, the next two quarters could be catastrophic.
Included in the figures are a $65m write-down of unsold APU stock that AMD clearly over produced, a tendency that seems endemic in the tech sector.
In an attempt to placate angry investors looking for the nearest fire exit, AMD announced that it would be spinning off its assembly and test facilities in a joint venture with Nantong Fujitsu Microelectronics. The Chinese business bought 85% of those assets, and that put $371m back in the kitty.
Dr Lisa Su, AMD president and chief executive, also continues to talk confidently: ‘We continue to take targeted actions to improve long-term financial performance, build great products and simplify our business model. The formation of a joint venture of our back-end manufacturing assets is a significant step towards achieving these goals and strengthening our balance sheet.”
But the reality is that the stock market has heard it all before, and shares declined roughly 3% on the day of the announcement.
Where From Here?
The fourth quarter of 2015 might be even worse, depending on whom on Wall Street you listen to. Most analysts are pitching just a $950m turnover, a full 10% down on Q4 2014, mostly down to seasonal declines in its semi-custom product lines.
Ironically, it’s these and non-PC product lines that AMD is pinning its hopes on in 2016, because PC sales are likely to decline further as people realise that they can actually do lots of typical PC jobs with just a tablet or their phone.
According to AMD’s latest roadmaps, the second half of 2016 will see its first FinFET node chips, a technology that could ultimately lead down to 7nm and even 5nm fabricated chip parts.
The actual chip maker, Global Foundries, was once part of AMD that spun off in 2009 and is now owned by IBM. It has declared its first FinFET technology is ready for production, starting with the more modest 14nm scale. That’s still a vast improvement over the 32nm or 28nm, on which the majority of AMD’s processors are currently built.
That puts it on par with Intel’s Skylake fabrication, and it has the possibility to take AMD even smaller down the road.
Despite promising AM3 ‘Vishera’ FX processors in 2015, there haven’t been any, and next year AM4 will arrive with the new Zenclass processors and the end of AM3+. Since 2013, the desktop side of the AMD equation has effectively languished, and socket AM4 will not only revitalise that line but also provide a single socket for both FX and APU product lines on the desktop.
From what information has been made available, it looks like Zen will switch AMD from modular CMT (Clustered Multithread) to SMT (Simultaneous Multithreading) micro-architecture, as used by Intel. This allows for more per-core performance rather than hoping that multi-core operations deliver a better overall experience, and it allows for larger dies with more computing power on them.
As with the Fury GPU, AMD badly needs some headlines that place its CPU and APU product lines at least on a par with what Intel is doing, rather than lagging behind.
AMD needs to ditch the association it has with the discount end of the computing spectrum and deliver products that people are happy to pay a profitable amount to own. Otherwise, the future isn’t going to be bright.
Final Thoughts
As more time passes, it seems like the heady days of the Opteron launch are long behind AMD, possibly never to return. Seen with the benefit of hindsight, those ground-breaking chip launches that shook Intel to its very processor intensive core look increasingly like AMD merely woke the sleeping giant, rather than delivered any lasting redress to the computing status quo.
The lack of any meaningful development on the FX line until recently, where the last new processor turned up more than a year ago, strongly indicates it no longer want to compete in the desktop CPU market with Intel. However, its roadmaps say otherwise, confusingly.
It’s entrenched in the desktop PC, budget laptop and micro PC markets with its APU products, a series that has never really delivered the combination of CPU and GPU power that AMD once promised. Will 14nm actually make the APU shine at last?
It’s also bought into the ARM revolution, although with Qualcomm and others banging away in that sector, it could be hard for AMD to be heard above that cacophony. Should it just focus on x86, even if that is a shrinking market?
The sad reality is that had it not bought ATI, there really wouldn’t be much of an AMD story to tell these days, and even that came with a sting in the tale. After years of chasing Nvidia somewhat fruitlessly, it’s now delivered the Fury, only for the steam to go out of the discrete video market. That’s not helping the bottom line, though there always appears to be some good reason why it’s never as healthy as it might have hoped.
Market fluctuations aside, the problem AMD has is balancing its ambitions with the scale of its business in what is currently a shrinking market for its products. These are largely the same problems that Intel must contend with, but it has a much larger pile of cash and significantly more patient investors than AMD. Even Intel is catching a cold with the slowdown in PC sales, but the weaker AMD might be getting pneumonia if it doesn’t pick up soon.
Along with cash, AMD also seems to be bleeding talented people, as its resident chip guru Phil Rogers left to become Compute Server Architect at Nvidia only a few weeks ago, probably taking with him some off its better product ideas. Rogers had been at AMD for 21 years, so this exit can’t be brushed off as youthful ambition.
If AMD is to survive, there needs to be a root-and-branch exercise in reducing its costs, or 2016 will probably be the last spin of the dice for this unique company.