Every once in a while, you see a press release that is so spectacularly misguided that it can’t be by chance. A little more rare is to see one that picks on a similarly dumb move by another company, and is then botched again by a publication that really should know better.
It started out with a press release about TSMC beating Intel to 3D chips, an intriguing proposition because TSMC’s 28nm process is not a FinFET process, and is not ‘3D’ in that sense. We will get to the abject stupidity of the whole 3D naming in a bit, but first, take a look at the the TSMC release, printed here as we received it, with minor formatting changes for print-ability. The original can be found here.
<Begin TSMC release>
Taiwan External Trade Development Council
News Release No. 21, 2011
TSMC May Beat Intel to 3D Chips Market
The chips boost the density of transistors in a single semiconductor by up to 1000 times.
FOR IMMEDIATE RELEASE. TAIPEI, TAIWAN – Taiwan Semiconductor Manufacturing Co., the world’s largest contract chipmaker, may beat out Intel to become the world’s first company to sell three-dimensional chips that boost the density of transistors in a single semiconductor by up to 1000 times.
The company could make its first 3D chips commercially available before the end of the year according to a person close to the situation who preferred to remain anonymous. The Taiwan firm’s year-end schedule matches that of its US competitor, which has set the end of 2011 for the launch of its 3D Tri-Gate chips.
Intel expects the Tri-Gate to be the world’s first commercial 3D chip. The technology has several layers of silicon stacked together, which allows it to achieve performance gains of about a third while consuming 50 percent less energy. That makes the chips especially suited for use in new generation technologies such as mobile phones and tablets.
The device is said to be the most significant advance in chip technology since the development of the chip transistor in the 1950s.
Shang-Yi Chiang, senior vice president for R&D at TSMC spoke about the new technology.
“This is definitely a new business opportunity for TSMC,” Chiang said. “We are building a patent portfolio now.”
Chiang also said the firm has been working closely with chip packagers and providers of design auto software to commercialize the 3D chip technology.
The new technology is expected to override a number of difficulties posed by traditional “planar” transistors, which can only move electrons across two dimensions. The three-dimensional structure of the next-generation transistors also allows chips to operate at lower voltage and with less leakage.
TSMC posted roughly US$1.2 billion in net profit in the first quarter of 2011. Morgan Stanley said the firm’s better-than-expected first quarter earnings reflected its efforts in controlling operating expenses and maintaining its gross margin.
Taiwan’s domestic chip industry accounts for nearly half of the market capitalization on the Taiwan Stock Exchange (TSE). Taiwan’s market grew from US$6.87 billion in 2009 to an estimated US$9.11 billion in 2010, representing a growth rate of 36.2%. Amid this growth, Taiwan’s government is taking steps to improve the competitiveness of the domestic industry. For example, Taiwan’s National Science Council (NSC) has been fostering ties between universities, research organizations and semiconductor companies to develop new technologies to enhance the island’s technological edge in the market.
Other major semiconductor firms in Taiwan include UMC (United Microelectronics Corporation) and Mediatek, which is a fabless semiconductor company.
Data shows that the global semiconductor material market rose 25 percent last year from the previous year to a value of US$43.55 billion. That represented a new high since 2007, when it recorded US$42.67 billion.
About Taiwan External Trade Development Council (TAITRA)
The Taiwan External Trade Development Council (TAITRA) was founded in 1970 to promote Taiwan’s foreign trade and competitiveness in world markets. Over the past 40 years, TAITRA has played a key role in the development of the Taiwan economy. TAITRA is jointly sponsored by the government and commercial associations and is viewed by all as the business gateway to Taiwan for the international business community. Please visit www.brandingtaiwan.org or www.taiwantrade.com.tw for more information.
Tel: <phone removed>
Mobile: <phone removed>
Email: jseidman<email removed>
Phone: <phone removed>
Email: michellewu<email removed>
<End TSMC release>
Lets start off by being blunt, whoever put TAITRA up to this debacle should be fired, and TAITRA should cut ties with TSMC over this, it is that bad. It looks like TAITRA was used by TSMC to try and put out a purposefully deceptive release about ‘3D’ chips. This uses TAITRA’s good name to buy TSMC a few headlines, and in the end makes both look abjectly stupid.
Why? What Intel did was a remarkable technical achievement, 22nm FinFETs. What TSMC is doing is Through Silicon Vias (TSVs), two totally different and unrelated technologies. FinFETs are all about the transistor itself, basically it goes from planar, IE parallel to the wafer, to somewhat vertical in order to minimize leakage. It is a very hard thing to do, and no one else is there, nor will they be for quite a while. TSMC on the other hand is putting vias in the wafer itself, and bonding chips together to stack them. This has absolutely zero to do with the transistors.
There are several companies selling stacked chips on the market, and Elpida just announced they are sampling a 4 DRAM stack using TSVs now. TSMC is far from a leader here, they are doing TSVs before some, but are far from first. Intel is unquestionably the leader in process technology, they are at least a full generation ahead of everyone else, period. This is probably why TSMC chose to compare themselves to Intel, ‘beating’ the leader to a very difficult technology is quite a feat. Then again, a press release saying, “TSMC is finally doing TSVs, beating several also-rans who probably won’t be in the business for much longer, but won’t be making FinFETs like Intel for at least 2 more years.” doesn’t have the same ring to it now, does it?
To make matters worse, half the press release is just plain stupid. “The Taiwan firm’s year-end schedule matches that of its US competitor, which has set the end of 2011 for the launch of its 3D Tri-Gate chips.” Think about that for a minute, this is akin to saying that Ford is planning on releasing a new SUV this fall, about the same time as a hippie commune in Vermont is going to get high and pick apples under the full moon, so Ford is a leader in organic farming. TSMC looks mighty stupid here, but that isn’t all of it.
In a classic case of not knowing when to stop digging a deeper hole, TSMC keeps going. “The new technology is expected to override a number of difficulties posed by traditional “planar” transistors, which can only move electrons across two dimensions. The three-dimensional structure of the next-generation transistors also allows chips to operate at lower voltage and with less leakage.” Fair enough, except that TSMC DOES NOT DO FinFETS, NOR DOES THE TECHNOLOGY THEY ARE MAKING A PRESS RELEASE ABOUT. Let me repeat that, TSMC is touting a technology in a release that they do not make, but the competitor who gets more space in the release does. TSMC can not make FinFETs, will not be able to make FinFETs at production levels for years, but is telling the world how great they are. Wow, just wow.
In situations like this, the best move is for a company is to simply shut up. There is no reason to issue a press release about the competition that only makes them look better and hurts your own image. The only way you could make it worse is to not tell the world what you are doing while dragging your friends in to the muck. Now you see why this release is so abjectly stupid? TSMC made Intel look good, made themselves look really stupid, and dragged TAITRA into it as well. As I said, whoever is responsible for this has unquestionably blundered. How this was conceived, much less released is truly a mystery.
That said, TSMC and TAITRA are not the only ones here that deserve blame, Intel needs some too. Why? Their press briefing on 22nm/FinFETs/Tri-Gate/3D was a masterclass in using the press in deceptive ways. Of late, Intel press briefing have been so bad that we basically don’t bother writing them up. When you are done with the webcasts, the most often asked question is, “How do I get that utterly wasted hour of my life back?”. We aren’t the only ones to feel this way either.
The press briefings have gotten so bad that there is no way it is by chance, Intel is purposefully trying to misuse the press in order to have their talking points repeated while not giving any chance for those pesky questions to be asked. In theory, it makes them look good and grabs headlines, but as you can see, problems remain. The fact that it is purposefully deceptive and unethical doesn’t seem to be important to the company any more. Think of Intel as a short bus version of Apple here.
Back to the TSMC/TAITRA stupidity though, Intel’s own moronic behavior opened up a huge hole for TSMC to drive a truck laden with marketing materials through. That hole is ‘3D’. What is 3D? What does it have to do with transistors? Do the Nvidia glasses help in fabricating them? Instead of doing the usual technical briefing while allowing the press to ask questions and inform their readers, Intel decided to dumb things down to worthlessness and push in order to push some talking points.
Given the awful state of the technical press, this exploitative behavior worked like a charm, go look at the headlines about 3D, skip the articles though, most of them make you feel dumb. The problem is that now the public and the mainstream press has no way of knowing what 3D is, only that it is ‘good’, derp derp derp. Intel dumbed them down, and TSMC saw an opportunity to exploit it. Fair enough, and normally we would applaud that kind of behavior. When someone is stupid, they deserve a slapping, and corporations are now people, in the US anyway. Instead of TSMC taking the high road, they were deceptive, made Intel look better, and took TAITRA down with them. Brilliant! The only losers are the readers. So far anyway.
That brings us to the press, specifically EETimes. With a title like EETimes, you would think they might have a vague clue about the technologies they are dealing with, right? If you read the story about the TSMC/TAITRA announcement, you will see that they not only have the technical knowledge to expand on some of the finer details, but they also miss the point entirely. EETimes takes the flag from the mud after TSMC dropped it, and, well, it is hard to come up with a euphemism for making it dirtier without using family unfriendly terms. Read the comments at EETimes to see how bad it is.
So, there we have it, a right mess. Intel dumbed things down to the point of drooling stupidity. TSMC had the opportunity to take the high road, but instead was deceptive and took TAITRA with them. Not feeling like enough had been done, EETimes decided to show the world how much the writers they fired over the last year were actually worth, and made me remove at least one bookmark. Who do I sue to get those brain cells back from reading all this crap multiple times? At least we didn’t have to put up with another abjectly stupid video of a highly technical and well respected process guy doing a dog an pony show in a press conference again. It is time to pour a lot more chlorine in the gene pool.S|A
Latest posts by Charlie Demerjian (see all)
- Thing go bump(gate) in the night for Nvidia’s GP100 Pascal GPU - May 3, 2016
- Cavium’s Octeon TX blends compute and packet moving - May 2, 2016
- Qualcomm releases Zeroth API to developers - May 2, 2016
- Another detail about Qualcomm server SoCs revealed - Apr 27, 2016
- AMD finally really honestly launches the dual Fury - Apr 26, 2016