United States. Congress. Senate. Committee on the.

The Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) online

. (page 59 of 140)
Online LibraryUnited States. Congress. Senate. Committee on theThe Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) → online text (page 59 of 140)
Font size
QR-code for this ebook


the logic function and about four years in the memory function from present
solid state devices. CCD's would typically be used as long serial shift registers
in buffer memories ; bubbles would be used to replace core or other semiconductor
devices in the main memory since they are extremely high density devices. If
IBM uses CCD's in the buffer, cache or scratch pad memories, and bubble memory
in the CPU, the other computer mainframers would not be able to introduce
competitive systems for about five years after IBM announced FS. Furthermore,
the semiconductor houses' computer market share would erode l)ecause it is not
certain that today's devices would be compatible with bubble memory tech-
niques. The only other company known to be advanced in the bubble memory
techniques is Bell Telephone Laboratories. In summary, the other computer
mainframers would not have a source of supply since both IBM and BTL pro-
duce for in-house use only, and the semiconductor companies might not be able
to produce.

In order to show that the semiconductor industry would remain a very viable
and strong competitor if IBM Components were spun off, I have constructed
three tables showing industry statistics, company statistics, and projected growth
rates for selected devices through 1978. Table IV shows U.S. factory sales of
all types of semiconductors for the decade 1969-1978. The evidence is conclusive
that the semiconductor industry is a growth industry, and that that growth is
expected to continue through 1978 (most forecasters do not go beyond 1978
because of expectation of IBM's FS announcement). The largest market .segment
is computer and business equipment with a 39.3 percent share in 1969 and a 37
percent share in 1978, excluding IBM in both cases. The second largest .share
is total automotive and consumer which was 17.4 percent in 1969 and is expected
to rise to 18.6 percent in 1978. Fairchild forecasts the data on a worldwide con-
sumption basis as shown in Figure 6.

If IBM were introduced into the marketplace as a supplier, initially it would
acquire some share of the computer and business equipment market, which would
be the memory segment of that market. Intel, the leading semiconductor mem-
ory house, has recently published forecasts for the memory market tlirougli
1976 which are shown in Table V below. Intel's estimates for memory apply
entirely to the computer memory market which includes mainframes, small
business computers, industrial minicomputers, terminals, communications equip-
ment, and microcomputers.



5427



TABLE V

SEMICONDUCTC"! MEMORY MARKET

IMillions of dollars!



Type


1973


1974


1975


1976


RAM

ROM PROM.


86

53


257
104
39


333
101
33


408
102


Registers ...


. - 30


36








Total


169


400


467


546


MEMORY PENETRATION
|ln percent]


Sector


1973


1974


1975


1976


Computers

Terminals

Communications


_ 13

- 80




39
95

loo'


52

100

1

100


72

100

2


Other


-_. 100


100


Total _..


16


38


51


69



Source: Intel.

There are, however, many other new applications for semiconductor memory
in noncomputer areas such as telephony, automotive marlvcts, home appliances
and security systems, games and disk replacement. If we assume that the com-
puter semiconductor memory market expands to $650 million in 1978, and that
IBM can capture an unrealistically high 50 percent of that market, then in
1978, IBM would have a 22.4 percent share of the total market for computers
and l)usiness equipment (650 million divided by 50 percent=$325 million, divided
hy .$1446 million=22.4 percent. However, that would result in only 8.3 percent of
total U.S. factory sales of semiconductors of an estimated $3.9 billion.

The next question to be answered is the size of IBM Comiwnents relative to
the other semiconductor companies in the industry. Texas Instruments is the
industry leader with total sales in 1973 of $1.3 billion, of which about $650 mil-
lion is estimated to be semiconductors, and of that, about $350 million is inte-
grated circuits. Motorola is the next largest with $400 million in total semi-
conductors, followed by Fairchild and National Semiconductor. The ten largest
producers have total sales of about $1.8 billion, with only $50 million in esti-
mated .shipments going to IBM. I estimate that IBM Components has total
volume of about $600 million, all in integrated circuits based upon the assump-
tion that the company purchases all its discretes from outside suppliers. IBM
would therefore be the largest integrated circuit company, but Texas Instru-
ments would still be the top semiconductor producer in the industry. Table VI
gives some pertinent data for each of the ten largest companies.

In attempting to set up an income statement for IBM Components as a sepa-
rate operating unit, I have used Intel as the most realistic model since it is the
lowest cost producer in the industry. Manufacturing costs are obviously low
because of its extremely high run rates on a limited number of circuits. I have
u.sed depreciation allocations higher than the rest of the industry, including
Intel, because IBM's production process is highly automated and because the
company uses very conservative accounting. The industry typically depreciates
semiconductor manufacturing and test equipment over a 2^ year period, and
I have a.s.sumed a two-year period for IBM. As previously discii.s.sed. IBM .spent
$730 million on all research and development in 1973, and probably $70-$100
million of that was allocated to semiconductors. Cost of sales is therefore 70
l>ercent of Components' revenues. Determining marketing and general and ad-
ministrative expenses was somewhat more difficult, and so I used the ratio ex-
perienced by the newest company in the indu.stry. Advanced Micro Devices, at
19 percent. Intel's SG&A expenses have been running at only 11 percent but
that ratio is now shifting to a much higher level as the company begins to ex-



5428

pand its market penetration outside of the computer memory market. Putting
all these hypotliese together, IBM Components would liave an operating margin
about the same as Texas Instruments, Fairchild, and Motorola, but lower than
Intel, Mostek, and Advanced Micro Devices, which probably can be attributable
to relative size. The variations which occur bj changing the various factors is
shown in Table VII.

In conclusion, I believe this simple groui) of tables shows that IBM would be a
viable competitor in the semiconductor industry and that the other companies
which now dominate the industry would not be placed at a competitive disadvan-
tage. Furthermore, the other computer companies would be able to compete much
more effectively if they could use IBM's semiconductor memory devices which
will surpass those of the independent semiconductor companies in the 1975-76
time frame. No attempt has been made to estimate revenues for IBM's test equip-
ment operations, but the total market today is probably $150-$200 million. It is
highly cyclical and is currently in a boom year. The inclusion of revenues from
this source could increase IBM Components Division revenues by as much as $25
million. TI and Fairchild also build this type of equipment but revenues are not
available. The major companies would probably retain their present competitive
characteristics.

TABLE VII

IBM COMPONENTS, INC.

[Dollar amounts In millionsl

1972 1973 1973

Sales

Cost of sales:

Manfacturing _..

Depreciation

Research and development

Total - . - -

Selling, general and advertisement

Total expenses _ _ . .

Pretax income _. _ _



[The tables and figures mentioned follow :]



$600


Percent


$600


Percent


$600


Percent


$3C0
21
73


50.0
3.5
12.2


$330
21
9C


55.0
3.5
15.0


$330
21
80


55.0
3.5
13.3


394
120


65.7
20. C


441
120


73.5
20.0


431
90


71.8
15.0


514
86


85.7
14.3


561
39


93.5
6.5


521

79


86.8
13.2



5429



o =
< E






O^ CvJ to OO O) ij



OO a>CT>tD CTJ 00






>^^r^oo



C300 OOO
OOCDCSOO
OOOOOC3



^:^ CT) oo •— < r^ to

r*"" 0> CT> OO ** PQ



r CO iDto ^•



lo CO ^* CM to r*.



(J)CM. - ^hC



^- OO r^ uj LU to



OltT) CO OOCNJ Oi



OfSJCNI ^-«^<



>>5 c = o



5430



Figure I



fO



I



z -












^J



N



^r^ [^n [\



JIL



/em



V*'



Figure 2





1


T^TTfL p^£-r/^y /A/co/r}£




K/i


VA


v^ F^



J8m



SP£^/iAY ^/OAlertUfiL A^CIS BUfitiC06MS CoMTKOLOi^n



j2/n^73 erc£Pr spstetr /t4i^o (s/jf/?^).



5431



Bee>



6co



4oo



% 2e0



Y

I



>:



Figure 3





1

1^








7Z>T-/)C /R4rO er^AJD/raesS












^




1^


'y.^


-y.


K^




fSm


: ^*


9£ett


Y A«/


VSYta


Figure 4 '



TcfT/tL C/9S// /jA/O £QOil//lC£f)jrS



t:^"^ f-^



/a^ .SAAeer //PAmt*^eu a/ca oueecueMS CoMrsecMm






5432



TABLE II



Company



Price



1973

earnings

per share



Shares

Price/ outstanding

earnings (millions)



Advanced Memory Systems' $9

Advanced Micro Devices' ^

Burroughs - - 88

California Computer Products.. /

Control Data 20

Data Products 3

Fairchild 30

Honeywell - 54

IBM 198

Intel' -..- - - 34

Memorex' 2

Mostek' 12

Motorola. -._ 48

NCR 28

National Semiconductor 12

Potter Instruments 2

Sperry Rand - 33

Storage Technology' °

Telex - 2

Texas Instruments 84



($0. 19)

.98

3.00

1.57

3.70

.22

5.12

5.12

10.79

1.41



-X

7.1
29.3

4.5

5.4
66.0

5.9
10.5
18.4
24.1



1.89


6.3


2.95


16.2


3.10


9.0


1.40


8.6


(.83)




3.27


10.1


1.28


7.0



3.67



22.9



1.3

2.4

39.0

3.1

15.9

6.8

5.1

19.2

146.9

6.2

4.3

3.9

28.0

22.8

11.3

2.8

34.4

3.3

10.5

22.8



Market

value

(millions)



$11.7

16.8

3, 432.

21.7

318.0

20.4

153.0

1,036.8

29, 086. 2

210.8

8.6

46.8

1,344.0

638.4

135.6

5.6

1,135.2

29.7

21.0

1,915.2



1 Traded over the counter.



5433



Shsckley. Bardeen
Bratten



Bell Laborator\es



Hoeml
Amelco



Hall
Molectro



General
Hicro Elect.



Phllco-Ford
HIcpoElect.
(closed 1968)



Electronic
Arrays



Noyce
Intel



Advanced
Micro-Oevlces



HonolUhlc
Memories



n



Freund,Hoerni
Union Carbide



Bobb

American

Micro S>stOTi»



Shockley

Shockley
Transistor




Clevite




ITT


'











Noyce
Fairchild



Baldwin
Rheem



Texas
Instru.



Rothlein
National



James
51gnet1cs



Lee.Hugle
smconix



Hugle
Stewart Warner



Sporcli
National



Moerni
Intersil



Raytheon



I Westlnghouse I



O-



Adv. Memory
Systems



Signetlcs
Mem. Systems



Closed
1968



Cogar
Co gar



FIGURE 5. SEMICONDUCTOR INDUSTRY GENEALOGY
(Courtesy of Electronic News. January 25, 1971)



5434



TABLE III.— HISTORY OF SEMICONDUCTOR INTRODUCTIONS VS. IBM USAGE



Year Industry introduction



IBM usage



1912... Vacuum tube

1946... Sperry builds first computer

1947... Transistor invented .._

1952... Texas Instruments gets license from BTL and makes
germanium transistors.

1953 701 introduction using vaccum tubes.

1954... Texas Instruments makes silicon transistors

1958... Texas Instruments invents the integrated circuit

1959 _ - - 1401 introduction using termanium transistors.

1961... Texas Instruments begins making simple integrated
circuits.

1962... Fairchild makes transistors with double diffusion planar
process.

1964... Fairchild makes integrated circuits with double diffusion 360's introduced using solid logic technology-hybrids,
planar process.

1955... Texas Instruments announces transistor-transistor-
logic (TTL).

1967... Small scale integration— Industry-wide

1969... Medium scale integration— Industry-wide; Motorola
announces MECL-10,000 emitter-coupler-logic; Amer-
ican Micro-Systems— MOS circuits, metal-oxide-
silicon.

1970... Intel announces P-Channel MOS process for memories; 370/155 and 165 using monolithic systems technology
Texas Instruments— Bipolar memories. with core memory.

1971... Intel announces N-Channel silicon gate MOS process 370/135 and 145 using MST with bipolar memory-370/
No. 1103 memory circuit 1-K device. 195 using MST and core.

1972... 1-K and 2-K devices available— ASP i— $14 370/158 and 168 with MST and MOS memory using 1-K

devices— 370/125 with MST and MOS.

1973... 4-K MOS devices announced— 1-K— ASP— $11 370/115 with MOS 1-K devices.

1974 . 4-K MOS devices shipped in limited quantities by Tl, Switch 1-K bipolar to 4-K MOS in 135 and 145.
Intel and Mostek— ASP of $4, 1-K— ASP— $3.

1975... 4-K MOS generally available but ASP is still $14 4-K MOS now cheap— ASP— $6-$8.

1976... l-K-ASP-$2-4-K-ASP-$8 4-K-ASP-$4-$5.

8-K available— ASP— $14 16-K or 32-K available.

• Average selling price.

TABLE IV— UNITED STATES FACTORY SALES OF SEMICONDUCTORS
[Dollars in millions]





1969




1973




1974




1978




End use market


Amount


Percent


Amount


Percent


Amount


Percent


Amount


Percent


Consumer


$194
22

487
169
169
200


15.6
1.8

39.3
13.6
13.6
16.1


$250
50

825
310
362
283


12.0
2.4

39.7
14.9
17.4
13.6


$300
90

911
358
418
310


12.6
3.8

38.1
15.0
17.5
13.0


$475
250

1,446
586
699
449


12.2


Automotive. _

Computer (Business Equip-
ment)

Communications


6.4

37.0
15.0


Instrumentation and controls.
Government


17.9
11.5






Total


1,240 -.




2,080




2,387 ..




3,905 ...


























67.7 .




14.8 ..




63.6 .





















Source: RCA.



5435



t/3 -O

O a>

^ 1

^ !

o ^

2: -

o *>



E
■o —



CTJ •
00



2 I



I if



OOf-iO»CSJ



oo^H ^^m ^^



10

o 00 ^-^ -^ ro r-^ to ■
00^ ^^ fee-



r^ CO CT> CM CT> <— ' c

o *-i 00 ' ai^^<r



01 •— ' o CJ o •— • CT>



00^ — ^'



3— r-LD t£3



• C

«— ' 00 ooosj in



od^f



OOr-> 1— >'



■-= ' m "^ ra E E '






m "^ o Q.S 2 TO a) a.
cc_i— 000.1— ZUJ



-•I



id

So:



5436



Figure 6



FAIRCHILO CAMERA AND INSTRUMENT CORPORATION



WORLDWIDE SEMICONDUCTOR MARKET
End-Equipment Market Consumption Patterns



%




PERCENT OF TOTAL


CONSUMER

COMPUTER

INDUSTRIAL

GOVERNMENT


DOLL


ARS


N BILLIONS


$




1 ' 1

50 40


' 1 ' 1 ' 1 '
30 20 10


1 1 1
1.0


1
2


' 1 ' 1 ' i
3.0 4.0 5.0




31


V//////,./y./


^,^y^


1.684


32








1


3.649














30


VA;-iy'-/ ,r ;


^^•x^/^


1.672


31


1






1


3.515














28


\^t^->^^^^^:^;


1>^^^


1.568


30








1


3.372






^








11


t^


0.584


7




1


0.834


w//,


1980 r







1974



Senator Hart. Our next witness is Royden C. Sanders, Jr., president
of Sanders Associates, Inc.

I have a message here from Senator Mclntyre. He hopes you will
repeat what you personally told him. We are going to hear from you
now.

STATEMENT OF ROYDEN C. SANDERS, JR., PRESIDENT, SANDERS
ASSOCIATES, INC., NASHUA, N.H.

Mr. Sanders. Thank you, Mr. Chairman. I am Royden Sanders of
Sanders Associates and we have a prepared statement today. We will
submit it for the record.

Senator Hart. It will be printed in full.

[Mr. Sanders' prepared statement appears as exhibit 1 at the end
of his oral testimony.]

Mr. Sanders. With your permission I will summarize our position.

First of all, we appreciate the opportunity to share our views with
you. Sanders Associates is predominantly an electronics company,
with total sales of $170 million, about $25 million of which is derived
from the commercial data processing market, principally in cathode
ray display terminals used to communicate to and from data processing
systems. That relatively low figure may class us as one of those cottage
industries which Mr. Katzenbach referred to on Tuesday. But because
we are at the forefront of the distributed processing and the display
terminal market we feel we have carved an important niche in the
fastest growing segment of the industi*y.

Our field is dominated by IBM and we have had, and continue to
have, our private antitrust 'differences with IBM. The nature of those
differences has been discussed by some of the previous witnesses. Mr.
Biddle touched upon IBM's use of software to disconnect competitive



5437

<levices. We have suffered from that and we intend to take care of that
privately or by private antitrust action. But today we want to discuss
a much larger threat, and our views on how to deal with it.

In addition to terminals, Sandei-s has technical expertise and busi-
ness interest in the communications satellite field. Our insight into the
domestic satellite situation, coupled with our experience in data dis-
plays, places us in a unique position to register the enormity of IBM's
latest move: Acquisition of controlling interest in the CML Corp.
C^IL has been approved by the Federal Communications Commission
to start domestic satellite service.

Domestic satellites are the key to low-cost communications. A prop-
erly implemented switched satellite system providing neutral or trans-
parent interconnections to all users.'with no built-in bias toward one
equipment supplier, would be very beneficial to the American public
and the industries which supply it.

Actually, IBM has sold the business community two or three times
the main-frame capacity it needs or can efficiently use at this time.

With the introduction of IBM's nonswitched satellite very little of
this excess capacity will become available or useful. On the other hand,
a properly implemented switched satellite system will make this excess
capacity useful, and could consequently reduce computer users charges
two or three times. This could usher in an era of vigorous competition
in the field of distributed processing, which you have heard so much
tibout.

However, IBM's entry into the communication picture gives it a
means to control all elements of the distributed systems: The central
computer, the communication paths, and those remote devices such as
display terminals connected by the communications facilities. Having
oversold the main-frame market today IB]M will seek to control the
growth rate of the emerging market to its own advantage.

^"\niat's more, the IBISI method of control will make it next to im-
possible for meaningful competition to arise, giving a new twist to_ a
A'ery successful anticompetitive tactic they have employed for years in
tlie'data processing market. IB]M will keep secret until the last possible
moment the means of interconnection of devices using the communica-
tions systems, and they will be so complex when they are released that
competition will never catch up, IB]\I will own the systems and, there-
fore, the markets from computer tiirough communications to a wide
varictv of applications-oriented terminals.

In fact, technology exists today that would enable IBM to simply
bypass much of the local telephone plants now operated in urban areas
by A.T. & T. for business communications by the u?e of a verv small
rooftop antenna that communicates directly with the satellite. IB]M has
the voice/data switching equipment already in production Avhich
would complement its traditional data processing line to allow it to
completely take over all business information handling and commu-
nication needs. This is the integrated business system market just open-
ing up.

The user community will be confronted with a dilemma of acce])ting
the fidl IBM product line offering or rumiing the risk of acquiring
superior independent equipment only to find it disconnected by IBjM's
capricious manipulation of the interface. If an independent does man-
age to supply a device that hooks on in competition with an IBM busi-

40-927 — 75 39



5438

ness product and there is a problem in tlie IBM commnnications sys-
tems, whose installation do you think will receive primary attention b}'
an IBM-dominated CML? '

]\Iany more examples could be given, but. in short, the ]:)otential for
harm is far too high to allow any relationship between IBM as a sup-
plier of communications services and IBM as a supplier of business or
terminal equipment. The difficulties of tlie A.T. & T.-Western Electric
vertical arrangement are far too clear to let a similar situation develop.

Soilie means is needed by Government to insure that the degree of
concentration I have forecasted here — total domination — does not oc-
cur in this new^ field that is being created out of two old ones; comput-
ers and communications. Our prepared statement places this market
at $250 billion by 1980, and IBM could easily end up with the lions
share of that.

Because of its concentration on the American Telephone & Tele-
graph Co. and its lack of jurisdiction over data processing, the Fed-
eral Communications Commission is not in a position to fully safe-
guard the public interest. In fact, there is no Government operation
equipped to handle the problem created b}^ IBM's entry into communi-
cations. Those who are viewing IBM's entry as a panacea for the
A.T. & T. problem are making a catastrophic mistake.

Whether or not the industrial reorganization bill is a vehicle to ac-
complish the type of Government control required, I cannot honestly
say. I do think the current method of Government antitrust prosecu-
tion is very ineffective. The most glaring example is the curent Gov-
ernment litigation against IBM. From our brief contacts with the liti-
gation team they appear to be understaffed, especially when compared
to the legions of lawyers retained by IBM. They have had to accept
a cutoff date of 1972 insofar as their case is concerned, and that is un-
fortunate for many of IBM's most anticompetitive plans conceived
previously are only now being implemented. Many aspects of the com-
puter industry are not being covered by Government prosecution.
This can only lead to a judgment or, and this we fear the most, a near
term consent decree that is designed to punish only some of the past
sins and whicli does not recognize the change in the market IBM is in
the process of forcing and then occupying.

But something must be done, giving the immediacy of the problem we
have outlined. A dedicated judicial or Government mechanism, witli
the resources to accommodate continually changing market circum-
stances and the power to enforce policy in that environment, is re-
quired for solution to the IBjNI problem ; today's, which Justice is at-
tacking in a limited fashion, and a broader one of the CML acquisition.

We wish this committee well in efforts to find solutions to those
pressing national problems.

Thank you very much.

Senator Hakt. Thank you verj' much for a very effective summary.

Mr. O'Leary?

INIr. O'Leary. ISfr. Sanders, let me see if we understand the signifi-
cance of this acquisition that you have made a reference to.

If I misstate things, or go wrong, please don't hesitate to correct
me.

As I understand it, it used to be that computer users generally have
their computer on the premises, and that we are now moving into an
era where the user may have terminals for intelligence, or a series



5439

of computers in different parts of the country linked together by a
communications line.

Do I understand that your fear is that by virtue of this acquisition^
IBM will be able to oiler the whole package ; in other words, the data
processing? And the computers will bounce an electronic impulse off
a satellite and link up these various computers and terminals intO'
one service and one system ?

Mv. Saxders. Weil, there is no question about the fact that either'
the satellite system being proposed hj IBM. nonswitched system or the?
much superior switched system, which we have referred to, will pro-
vide superior computer results to the customer.

The difference in the switclied system is that when you have a
switched system the computer terminal Avill have the ability to access
more than one computer in an effective useful manner. And it Avill be
very much similar to the computer terminal ; the computer terminal
will then have the utility of the way you can dial any other telephone
in the countrv.

So you will go from a slave to one particular computer into some-
thing that can access any of the data bases, any i^lace in the country,
with an effective cost mechanism for distributing those costs.



Online LibraryUnited States. Congress. Senate. Committee on theThe Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) → online text (page 59 of 140)