zz A brief history of Apache and its IPO
时间:12-12
整理:3721RD
点击:
by John Cooley
Holliston Poor Farm, P.O. Box 6222, Holliston, MA 01746-6222
On Monday, Apache Design Solutions, Inc. (APAD) filed to IPO for $75 million
as a publically traded stock on the NASDAQ.
The last company to successfully IPO in EDA was 10 years ago. It was Magma,
which raised $63 million back in 2001.
REACTIONS TO AN APACHE IPO:
"Apache gave us a few initial headaches in dynamic analysis in their
early days; I have to give them that. But that was in 2006 and it
was ONLY for dynamic power analysis. VoltageStorm was and still is
the gold standard for static. In 2008 we introduced the Encounter
Power System, which passed Apache on the dynamic side.
Apache's other problem is we've built VoltageStorm directly into
Virtuoso and Encounter. At best, RedHawk is an add-on tool. We
now have power integrity and in-design signoff across both digital
and analog flows; including chip & package domains. This seriously
constricts Apache's business prospects."
- Dave Desharnais, marketing group director at Cadence
Another Apache weakness is also its strength.
"Apache has been Andrew Yang's baby since Day 1 of the company. He
built the company up from zero himself. Andrew works tirelessly and
is involved in ALL aspects of the company, from deep R&D questions to
customer sales to lawyer issues to accounting concerns to the exact
DAC Apache booth layout to minor marketing decisions. He does it
ALL. Without Andrew, Apache would fall apart. Aart was the same
way early on in Synopsys until he eventually learned to let go and
delegate. It took some time for him to get there, though."
- An anon Apache rival
"I think Apache's well run," says Gary Smith of Gary Smith EDA. "Andrew
might be the only one running it, but it's a well run EDA company. Their
market share numbers are fantastic."
2009 IC Power Market Share
Apache : : ########################################## (73%) $42.3 M
Cadence : : ############## (24%) $13.8 M
Sigrity : : ## (3%) $1.5 M
"Getting these numbers in a critical market like power is a significant
achievement for a small company like Apache," continued Gary. Their tools
are so good customers buy Apache tools outside of the Cadence package deal."
"For the noise market, Apache is weak because noise/SI is tied to timing
as far as most chip designers are concerned."
2009 IC Signal Integrity & Noise Market Share
Synopsys : : ######################## (52%) $24.0 M
Cadence : : ########### (24%) $11.0 M
Quantic : : #### (8%) $3.7 M
Sigrity : : ### (7%) $3.3 M
Apache : : ## (5%) $2.2 M
OEA : : # (2%) $1.1 M
OPTEM : : # (2%) $0.8 M
"I laughingly call this the PrimeTime Effect, because Synopsys got this
noise/SI lead from owning the timing market."
"Keep in mind that these are 2009 market share numbers," warned Gary. "For
timing I see Magma Tekton and Extreme-DA making some serious in-roads.
Synopsys is in trouble here. PrimeTime is so slow, I'm seeing engineers
using PT only for the final sign-off to fab. They're using Magma Tekton or
Extreme-DA for day-to-day design because they're 30X to 40X faster."
APACHE'S STRENGTHS:
Since I'm an engineer who only knows technical and NOT finance, I asked a
friend who knew financial stuff to look at the Apache IPO S-1 filing:
"First, they're profitable. Second, they have $26 M in cash, their
total operating expense is $29 M with an operating income of $6 M.
This means they have a year's worth of expenses socked away in cash.
They have no long term debt obligations, so all that cash is real.
This makes Apache's IPO a growth IPO and not a survival IPO. Some
companies burn through cash and IPO to survive; not the case here."
He also liked that APAD revenue grew 3X over 5 years.
Apache Design Solutions, Inc. (APAD) Revenue
2006 : : ############### $14.7 million
2007 : : ################### $18.7 million
2008 : : ########################## $25.7 million
2009 : : ################################### $34.6 million
2010 : : ############################################ $44.0 million
My friend then broke down the APAD financials in term of % of revenue to see
how it stacked up against its peers:
2010 Metric SNPS MENT CDNS LAVA APAD
---- ---- ---- ---- ----
Sales & Marketing 24% 37% 33% 33% 29%
General & Admin 8% 11% 9% 14% 11%
R&D 33% 32% 40% 36% 26%
Cost of Goods 21% 15% 17% 15% 20%
Operating Margin 12% 2% -3% -4% 13%
He liked what he saw there, too.
On the technical side, APAD has one flagship product called RedHawk, which
does IR-drop analysis on ICs. It's mostly sold as time-based SW licenses;
which minimizes revenue bumps & troughs from last-minute sales negotiations.
Customers: Intel, Samsung, TI, ST, Sony, Hynix, LSI, Qualcomm, ARM, Toshiba
Powerful customers like those ensure that Synopsys/Cadence/Mentor/Magma will
keep their own proprietary tools interoperable with Apache's tools.
Sales are 59% US, 15% Europe, 14% Japan, 14% Non-Japan Asia Pacific.
APACHE'S WEAKNESSES:
- Apache must sell against Synopsys/Cadence/Mentor/Magma bundled deals,
but they've doing that since its founding in 2001.
- The S-1 filing warned that many key APAD employees were fully vested.
Not good. They could walk once they get their money out of Apache.
- The S-1 filing warned that the APAD senior execs do not have experience
managing a public company. Same with their accounting staff. There's
a learning curve here that could hurt the company.
- Apache tools "focus exclusively on power-efficiency and noise-immunity"
for IC's and chip packages, which is a *limited* market. That 3X revene
growth every 5 years probably isn't going to be forever.
- Apache is Andrew Yang and Andrew Yang is Apache. Something happens to
him and the company dies. That will have to change.
Pushing this analysis aside, a great number of people in the EDA vendor
world (from both big & small co's) really want to see Apache's IPO succeed.
"I think Andrew did a phenomenal job coming out of the recession unscathed,"
said Rajeev Madhavan, CEO of Magma. "I hope he has a successful IPO. I can
sympathize that he'll now have to deal with more accountants and lawyers
than he ever had to before. Welcome to the world of Sarbanes-Oxley."
"The stock performance of how public EDA companies have done for the past
few years has been excellent; much better than what the DJIA has done," said
Kathryn Kranen, CEO of Jasper-DA. "It's well over 300 percent. The people
who track stocks love that. We hope to follow in Apache's footsteps."
"Yes! We finally have a company with a successful exit that doesn't involve
selling out to the Big 3!," said another anon EDA vendor.
"This is a good thing for EDA. We need more IPOs," said yet another.
"Taking 10 years to a liquidity event is definitely not breaking records,
nor is their $75 M potential IPO, but it is important," conceded Cadence's
Desharnais. "Apache's filing is good news for EDA and more evidence of a
return to growth in EDA overall. Congrats to Andrew and his team!"
Or, as an EDA start-up CEO said to me under his breath: "It's about #@$%-ing
time someone made some %@#$-ing money in this $#@%-ing industry!"
- John Cooley
DeepChip.com Holliston, MA
P.S. And for those into dirt & details, here's a brief history of Apache.
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
1995: - Andrew Yang, a professor at University of Washington, founded
a fast SPICE simulation company called Anagram with two of his
grad students, Jack Yao and Henry Chang. Andrew was the CEO
and Ed Cheng was the President of Anagram.
1996: - Anagram was bought by Avanti. Andrew remained at Avanti under
Gerry Hsu for a little over a year or so.
1998: - Andrew "retires" from Avanti; took a year off in Washington.
He was the lead investor in CadMos (bought by Cadence), Mojave
(bought by Magma), Innologic (bought by Synopsys), Ultima
(bought by Cadence) and Sapphire (later merged with Sequence).
2001: - Andrew Yang founds Apache Design Solutions, Inc., with Norman
Chang and Shen Lin both from HP. Also acquired a tiny company
that sold a true SPICE simulator called "NSPICE".
- Early investors: Andrew Yang, Andy Bechtolsheim and Intel Capital.
2002: - In an IR-drop world dominated by Simplex VoltageStorm, Iota, and
Avanti AstroRail, tiny Apache first introduces Tomahawk at DAC.
In it's niche Tomahawk makes a big splash because it benchmarked
10X faster using 1/2 the memory. See DAC'02 #21.
Tomahawk was initially a static IR-drop tool then.
- Trident was Apache's first IR-drop customer. Broadcom and TI
were also early customers.
2003: - Broadcom helps transition Tomahawk from static IR-drop analysis
to dynamic.
- Cadence at first denies a need for dynamic and then later relents.
They promise VoltageStorm will have dynamic in 6 months. It didn't
actually happen until years later.
- Patrick Wei, a newly hired Apache field engineer, posed using false
credentials to enter the Sequence DAC'03 suite for a demo. Wei gets
caught. EDA users and vendors alike laugh at this story until
Patrick Wei is fired and Vic Kulkarni, CEO of Sequence, sues Wei.
Opinion turns on Vic & Apache for going too far. See ESNUG 416 #4.
2004: - Apache renames Tomahawk to RedHawk because of the association of
to the Tomahawk cruise missiles in the first US Gulf War. All
references to "Tomahawk" on the Apache-DA.com site are magically
replaced with "RedHawk" -- even the old press releases change!
- Due to acquisitions, IR-drop becomes Synopsys AstroRail vs. Cadence
Simplex VoltageStorm. Apache small but growing. See SNUG'04 #15.
- Series B funding.
2005: - Apache acquires Baynacre which later becomes PsiWinder, to analyze
clock jitter in the presence of simultaneous power-grid and signal
coupling noise.
2006: - John Cooley does a DeepChip survey asking EDA users about Apache.
38 customers gush about Apache in ESNUG 455.
Item 1: Apache RedHawk vs. VoltageStorm vs. Sequence CoolTime
Item 2: Users on the Apache RedHawk PowerGate option
Item 3: Users on the Apache RedHawk FAO option
Item 4: Users on Apache PsiWinder
Item 5: Apache NSPICE vs. Synopsys HSIM & HSPICE
Item 6: Busted! No users on Apache SkyHawk!
Item 7: Users rank Apache Customer Support exceptionally well
Item 8: Exceptional people who work at Apache -- naming names
Item 9: Secrets of the Apache future product Roadmap
Item 10: What would Users Change about Apache
Item 11: What would Users NOT Change about Apache
And now let's look at those revenue numbers again from the S-1:
Apache Design Solutions, Inc. (APAD) Revenue
2006 : : ####### $14.7 million
2007 : : ######### $18.7 million
2008 : : ############# $25.7 million
2009 : : ################# $34.6 million
2010 : : ###################### $44.0 million
Interesting timing here. I love it when my readers get it right!
- Apache announced Sahara-PTE, which does temperature impact on
leakage, timing, reliability, and voltage drop.
2007: - Apache makes its first jump into IC-to-package power and EMI
analysis with its new Sentinel tool suite.
- Apache acquires Optimal Corporation, a package-to-PCB power
and SI toolset, for $7.5 million cash.
2008: - Sentinel-PI receives the "Best New Product" award at DAC'08.
- 53 customer companies worldwide.
2009: - Apache acquires archrival Sequence Design, Inc. for $5.3 million
cash. Vic Kulkarni, the former CEO of Sequence, joins Apache.
- TSMC Ref Flow 10.0 includes RedHawk, Totem, and Sentinel tools.
- Andrew Yang's "honour" is questioned when Apache's company bloggers
hide behind Techguri.com, a site falsely posing on its surface as
an unbiased EDA web site. See ESNUG 482 #1.
2010: - Apache presents in Singapore on IC-package thermal co-analysis
in a 3D-IC environment.
- Their PathFinder full-chip ESD tool was introduced.
- Apache in TSMC Ref Flow 11.0 and new TSMC AMS Ref Flow 1.0.
- RedHawk 10.2 released.
- 116 customer companies worldwide.
2011: - March 15th, Apache announces its intention to IPO for $75 million.
Holliston Poor Farm, P.O. Box 6222, Holliston, MA 01746-6222
On Monday, Apache Design Solutions, Inc. (APAD) filed to IPO for $75 million
as a publically traded stock on the NASDAQ.
The last company to successfully IPO in EDA was 10 years ago. It was Magma,
which raised $63 million back in 2001.
REACTIONS TO AN APACHE IPO:
"Apache gave us a few initial headaches in dynamic analysis in their
early days; I have to give them that. But that was in 2006 and it
was ONLY for dynamic power analysis. VoltageStorm was and still is
the gold standard for static. In 2008 we introduced the Encounter
Power System, which passed Apache on the dynamic side.
Apache's other problem is we've built VoltageStorm directly into
Virtuoso and Encounter. At best, RedHawk is an add-on tool. We
now have power integrity and in-design signoff across both digital
and analog flows; including chip & package domains. This seriously
constricts Apache's business prospects."
- Dave Desharnais, marketing group director at Cadence
Another Apache weakness is also its strength.
"Apache has been Andrew Yang's baby since Day 1 of the company. He
built the company up from zero himself. Andrew works tirelessly and
is involved in ALL aspects of the company, from deep R&D questions to
customer sales to lawyer issues to accounting concerns to the exact
DAC Apache booth layout to minor marketing decisions. He does it
ALL. Without Andrew, Apache would fall apart. Aart was the same
way early on in Synopsys until he eventually learned to let go and
delegate. It took some time for him to get there, though."
- An anon Apache rival
"I think Apache's well run," says Gary Smith of Gary Smith EDA. "Andrew
might be the only one running it, but it's a well run EDA company. Their
market share numbers are fantastic."
2009 IC Power Market Share
Apache : : ########################################## (73%) $42.3 M
Cadence : : ############## (24%) $13.8 M
Sigrity : : ## (3%) $1.5 M
"Getting these numbers in a critical market like power is a significant
achievement for a small company like Apache," continued Gary. Their tools
are so good customers buy Apache tools outside of the Cadence package deal."
"For the noise market, Apache is weak because noise/SI is tied to timing
as far as most chip designers are concerned."
2009 IC Signal Integrity & Noise Market Share
Synopsys : : ######################## (52%) $24.0 M
Cadence : : ########### (24%) $11.0 M
Quantic : : #### (8%) $3.7 M
Sigrity : : ### (7%) $3.3 M
Apache : : ## (5%) $2.2 M
OEA : : # (2%) $1.1 M
OPTEM : : # (2%) $0.8 M
"I laughingly call this the PrimeTime Effect, because Synopsys got this
noise/SI lead from owning the timing market."
"Keep in mind that these are 2009 market share numbers," warned Gary. "For
timing I see Magma Tekton and Extreme-DA making some serious in-roads.
Synopsys is in trouble here. PrimeTime is so slow, I'm seeing engineers
using PT only for the final sign-off to fab. They're using Magma Tekton or
Extreme-DA for day-to-day design because they're 30X to 40X faster."
APACHE'S STRENGTHS:
Since I'm an engineer who only knows technical and NOT finance, I asked a
friend who knew financial stuff to look at the Apache IPO S-1 filing:
"First, they're profitable. Second, they have $26 M in cash, their
total operating expense is $29 M with an operating income of $6 M.
This means they have a year's worth of expenses socked away in cash.
They have no long term debt obligations, so all that cash is real.
This makes Apache's IPO a growth IPO and not a survival IPO. Some
companies burn through cash and IPO to survive; not the case here."
He also liked that APAD revenue grew 3X over 5 years.
Apache Design Solutions, Inc. (APAD) Revenue
2006 : : ############### $14.7 million
2007 : : ################### $18.7 million
2008 : : ########################## $25.7 million
2009 : : ################################### $34.6 million
2010 : : ############################################ $44.0 million
My friend then broke down the APAD financials in term of % of revenue to see
how it stacked up against its peers:
2010 Metric SNPS MENT CDNS LAVA APAD
---- ---- ---- ---- ----
Sales & Marketing 24% 37% 33% 33% 29%
General & Admin 8% 11% 9% 14% 11%
R&D 33% 32% 40% 36% 26%
Cost of Goods 21% 15% 17% 15% 20%
Operating Margin 12% 2% -3% -4% 13%
He liked what he saw there, too.
On the technical side, APAD has one flagship product called RedHawk, which
does IR-drop analysis on ICs. It's mostly sold as time-based SW licenses;
which minimizes revenue bumps & troughs from last-minute sales negotiations.
Customers: Intel, Samsung, TI, ST, Sony, Hynix, LSI, Qualcomm, ARM, Toshiba
Powerful customers like those ensure that Synopsys/Cadence/Mentor/Magma will
keep their own proprietary tools interoperable with Apache's tools.
Sales are 59% US, 15% Europe, 14% Japan, 14% Non-Japan Asia Pacific.
APACHE'S WEAKNESSES:
- Apache must sell against Synopsys/Cadence/Mentor/Magma bundled deals,
but they've doing that since its founding in 2001.
- The S-1 filing warned that many key APAD employees were fully vested.
Not good. They could walk once they get their money out of Apache.
- The S-1 filing warned that the APAD senior execs do not have experience
managing a public company. Same with their accounting staff. There's
a learning curve here that could hurt the company.
- Apache tools "focus exclusively on power-efficiency and noise-immunity"
for IC's and chip packages, which is a *limited* market. That 3X revene
growth every 5 years probably isn't going to be forever.
- Apache is Andrew Yang and Andrew Yang is Apache. Something happens to
him and the company dies. That will have to change.
Pushing this analysis aside, a great number of people in the EDA vendor
world (from both big & small co's) really want to see Apache's IPO succeed.
"I think Andrew did a phenomenal job coming out of the recession unscathed,"
said Rajeev Madhavan, CEO of Magma. "I hope he has a successful IPO. I can
sympathize that he'll now have to deal with more accountants and lawyers
than he ever had to before. Welcome to the world of Sarbanes-Oxley."
"The stock performance of how public EDA companies have done for the past
few years has been excellent; much better than what the DJIA has done," said
Kathryn Kranen, CEO of Jasper-DA. "It's well over 300 percent. The people
who track stocks love that. We hope to follow in Apache's footsteps."
"Yes! We finally have a company with a successful exit that doesn't involve
selling out to the Big 3!," said another anon EDA vendor.
"This is a good thing for EDA. We need more IPOs," said yet another.
"Taking 10 years to a liquidity event is definitely not breaking records,
nor is their $75 M potential IPO, but it is important," conceded Cadence's
Desharnais. "Apache's filing is good news for EDA and more evidence of a
return to growth in EDA overall. Congrats to Andrew and his team!"
Or, as an EDA start-up CEO said to me under his breath: "It's about #@$%-ing
time someone made some %@#$-ing money in this $#@%-ing industry!"
- John Cooley
DeepChip.com Holliston, MA
P.S. And for those into dirt & details, here's a brief history of Apache.
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
1995: - Andrew Yang, a professor at University of Washington, founded
a fast SPICE simulation company called Anagram with two of his
grad students, Jack Yao and Henry Chang. Andrew was the CEO
and Ed Cheng was the President of Anagram.
1996: - Anagram was bought by Avanti. Andrew remained at Avanti under
Gerry Hsu for a little over a year or so.
1998: - Andrew "retires" from Avanti; took a year off in Washington.
He was the lead investor in CadMos (bought by Cadence), Mojave
(bought by Magma), Innologic (bought by Synopsys), Ultima
(bought by Cadence) and Sapphire (later merged with Sequence).
2001: - Andrew Yang founds Apache Design Solutions, Inc., with Norman
Chang and Shen Lin both from HP. Also acquired a tiny company
that sold a true SPICE simulator called "NSPICE".
- Early investors: Andrew Yang, Andy Bechtolsheim and Intel Capital.
2002: - In an IR-drop world dominated by Simplex VoltageStorm, Iota, and
Avanti AstroRail, tiny Apache first introduces Tomahawk at DAC.
In it's niche Tomahawk makes a big splash because it benchmarked
10X faster using 1/2 the memory. See DAC'02 #21.
Tomahawk was initially a static IR-drop tool then.
- Trident was Apache's first IR-drop customer. Broadcom and TI
were also early customers.
2003: - Broadcom helps transition Tomahawk from static IR-drop analysis
to dynamic.
- Cadence at first denies a need for dynamic and then later relents.
They promise VoltageStorm will have dynamic in 6 months. It didn't
actually happen until years later.
- Patrick Wei, a newly hired Apache field engineer, posed using false
credentials to enter the Sequence DAC'03 suite for a demo. Wei gets
caught. EDA users and vendors alike laugh at this story until
Patrick Wei is fired and Vic Kulkarni, CEO of Sequence, sues Wei.
Opinion turns on Vic & Apache for going too far. See ESNUG 416 #4.
2004: - Apache renames Tomahawk to RedHawk because of the association of
to the Tomahawk cruise missiles in the first US Gulf War. All
references to "Tomahawk" on the Apache-DA.com site are magically
replaced with "RedHawk" -- even the old press releases change!
- Due to acquisitions, IR-drop becomes Synopsys AstroRail vs. Cadence
Simplex VoltageStorm. Apache small but growing. See SNUG'04 #15.
- Series B funding.
2005: - Apache acquires Baynacre which later becomes PsiWinder, to analyze
clock jitter in the presence of simultaneous power-grid and signal
coupling noise.
2006: - John Cooley does a DeepChip survey asking EDA users about Apache.
38 customers gush about Apache in ESNUG 455.
Item 1: Apache RedHawk vs. VoltageStorm vs. Sequence CoolTime
Item 2: Users on the Apache RedHawk PowerGate option
Item 3: Users on the Apache RedHawk FAO option
Item 4: Users on Apache PsiWinder
Item 5: Apache NSPICE vs. Synopsys HSIM & HSPICE
Item 6: Busted! No users on Apache SkyHawk!
Item 7: Users rank Apache Customer Support exceptionally well
Item 8: Exceptional people who work at Apache -- naming names
Item 9: Secrets of the Apache future product Roadmap
Item 10: What would Users Change about Apache
Item 11: What would Users NOT Change about Apache
And now let's look at those revenue numbers again from the S-1:
Apache Design Solutions, Inc. (APAD) Revenue
2006 : : ####### $14.7 million
2007 : : ######### $18.7 million
2008 : : ############# $25.7 million
2009 : : ################# $34.6 million
2010 : : ###################### $44.0 million
Interesting timing here. I love it when my readers get it right!
- Apache announced Sahara-PTE, which does temperature impact on
leakage, timing, reliability, and voltage drop.
2007: - Apache makes its first jump into IC-to-package power and EMI
analysis with its new Sentinel tool suite.
- Apache acquires Optimal Corporation, a package-to-PCB power
and SI toolset, for $7.5 million cash.
2008: - Sentinel-PI receives the "Best New Product" award at DAC'08.
- 53 customer companies worldwide.
2009: - Apache acquires archrival Sequence Design, Inc. for $5.3 million
cash. Vic Kulkarni, the former CEO of Sequence, joins Apache.
- TSMC Ref Flow 10.0 includes RedHawk, Totem, and Sentinel tools.
- Andrew Yang's "honour" is questioned when Apache's company bloggers
hide behind Techguri.com, a site falsely posing on its surface as
an unbiased EDA web site. See ESNUG 482 #1.
2010: - Apache presents in Singapore on IC-package thermal co-analysis
in a 3D-IC environment.
- Their PathFinder full-chip ESD tool was introduced.
- Apache in TSMC Ref Flow 11.0 and new TSMC AMS Ref Flow 1.0.
- RedHawk 10.2 released.
- 116 customer companies worldwide.
2011: - March 15th, Apache announces its intention to IPO for $75 million.