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 97 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 97 of 140)
Font size
QR-code for this ebook


the need for compatibility, the PCM companies could not deliver a replacement
for a new product until after IBM had delivered the product, and if the products
went in on a two year lease, the PCM would be prevented from marketing a
replacement until two years after deliveries began. An IBM analysis showed
that a term lease plan which protected tlie market for twenty months would
turn PCM profits on the initial placement into losses. The analysis assumed
that the initial placement of the PCM machine would be lost at the time IBM
brought out a replacement product, but that the company could remarket the
machine at a lower price to another customer. The remarketing effort would only
incur juarketing and maintenance charges because no manufacturing or develop-
ment would be necessary and the profits from the follow on customer would make
up the losses on the initial placement for a profit on the entire product life. How-
ever, the analyst concluded that l)ecause profits were deferred for so long,
the company would have difficulty developing a follow on product and therefore
that with an effective term lease plan "PCM corporate revenues lower — no
funds for mfg., eng. — dying company I" (2.5)

At the recommendation of the Blue Ribbon Task Force, the IBM Manage-
ment Review Committee approved the Fixed Term Plan on May 25, 1971, and
announced it two days later. The Fixed Term Plan (FTP) allowed one or two
year leases on most tape, disk, and printer products. The customer received
an 8% discount over the month to month rental for a one year lease and a 16%
discount for a two year lease. All extra use charges (charges above the basic
monthly rental, which only entitled the user to 176 hours per month) were
eliminated on products xnader the FTP, making the effective discount for a two
year lease between 20 and 35%. If a product on a one year lease was cancelled
before the lease period was up. the customer was charged a penalty equal to
two and one half times the monthly rental charge : if a two year lease was broken
during the first year the penalty was five times the monthly rental. Card
equipment. CPU's, and System 3 equipment were not eligible for FTP. Before
this time IB]M had rejected the idea of long term leases because it felt there
was no way to justify leases on some products without appl.ving it to the entire
product line, but by May, 1971. the company was so concerned with its peripheral
competition that it downgraded the possibly adverse public relations aspects of
the FTP for peripherals only. The discounts available with the FTP, put IBM
prices l)elow PCM prices. IBM estimated that the FTP would cost the company
$31.5 million in 1971 and $44.1 million in 1972. but that the reduction in PC:M
com.petition resulting from FTP would bring in enough increased revenues in
1973 and later to make the total effect of FTP a gain in revenue of ?714 million
between 1971 and 1975. (26)

The potential danger to the PCM's posed by the FTP was quickly recognized.
The plan was announced ^lay 27. Between May 26 and May 28, Telex stock dropped
14%. Potter Instruments stocks dropped 10%, ]\lemorex dropped 15%. and
Marshall Laboratoi'ies dropped 18%. Potter, Telex, and Memorex continued to lose
value on the stock market throughout the next month. The following selection of
connnents from the trade press following the FTP announcement indicates clear
comprehension of the intent of the price cuts : (27)



5673

". . . the competition will be shooting at IBM leftovers: new delivered pe-
ripherals, leasing company installations, and — camaraderie be damned — each
other. . . .

. . . Too much slick pricing maneuvering on the part of IBM just could arouse
the Justice Dept.. which has maintained a deafening silence in the wake of IBM's
recent actions (price cuts, etc.) that have sent shock waves throughout the in-
dustry. . . . The new deal, affecting an estimated 80% of IBM's installed perphi-
erals on rent, was aimed squarely at independent manufacturers of plug-to-
plug compatible peripherals which System/.36U users have been installing at big
savings in both price and performance. . . . IBM not only lowered prices, it also
lowered tlie ability of its competitors to get cash to finance a negative cash flow
b.v loweriTig their profitability."

IBM's FTP was an immediate success with its customers. Within a month of
announcement, 40% of IB^NI's eligible customers had signed up for the new plan,
a coverage not expected by IB]M until six months later. Ninety percent of the new
370 peripherals (3330 disk drives and 4320 series tape drives) were being in-
stalled under FTP during June 1971. IBM considered the 3330 protection e.specially
significant because the first I'Cil 3330 had been announced but none had yet
been delivered by May, 1!>71. IBM concluded that in spite of the assumed 10%
price advantage of the independents on the 3330, '"it is assumed that near-term
3330 erosion will be contained until the FTP contracts approach maturity ; By
that time, Winchester, Iceberg, the 3330A/B and the 333M will all be available
as customer options and should hold the market for IBM." (2S)

The PCM companies reduced prices following the FTP in order to bring their
prices under IBM's, but not enough to fully restore the pre-FTP diffei-ential. An
IBM study of reactions to the I'TP six weeks after the announcement reported that
"Telex and Memorex have maintained their list price differential to the 2319B,
but they did not absorb the percentage decrease attributed to IBiPs additional
usage', and that Ami>ex and Potter had made ".selective price reduction". (29)
Another study showed that PCM discounts ranged from 4% to 28% on 2314-type
disk systems, depending on the configuration, length of contra <:-t. and PCM com-
pany. (30) The combination of smaller price differentials, penalties for removing
IBIM equipment early, and customer uncertainty about the long term viability of
PCM companies in the light of IBM actions reduced the order rate substantijxlly.
IBMs December, 1971 Quarterly Product Line Assessment reported that after tlie
FTP, the PCM tape order rate declined 62% and the disk order rate declined
48%. (31) The PCM order rate for the entire year following the FTP announce-
ment was 44% lower than for the year preceding the announcement. (32)

The FTP marked the end of the plug compatible companies as successful, jn-ofit-
able competitors to IBM. but it did not stop them from operating. Between the
instigation of the FTP and the end of 1973, none of the major PC:\I companies were
profitable, their sales declined, and their stock market values plummeted. In the
two years following the FTP, Telex's stock market value dropped from .$198 mil-
lion to ,S48 million. (33) The companies' depressed earnings and stock values
made it difficult to i-aise capital for development of new products and made IBM's
prediction that FTP would lead to PCM's becoming "dying companies" seem quite
accurate.

While the PCJI companies were feeling the pinch of the lower prices under the
FTP, IB:m made up the lost revenue through a price increase in CPUs and memo-
ries. On July 2.S, 1971 (two months after the FTP announcement), IBM announced
rental and purchase increases in 3G0 and 370 CPU's, channels, card equipment,
and maintenance. The increases varied from 4% to 8% on different pieces of equip-
ment, with some maintenance charges increased up to 25%. The purchase in-
creases became effective the date of announcement and the rental inerense ninety
days later in accordance with the notice of price change provisions in IBM rental
contracts. The rental increases were then further delayed by the August 15, 1971
price freeze which was instituted between the announcement and effective date of
the rental increase. In August. 1971, one of IB:\rs analysts wrote concerning the
price increase : "Although there will probably be adverse reaction on the part of
some customers, the net effect of the FTP and price changes will not significantly
increase his total costs and no systems decreases were forecast." (34) IB:\I was
able to cut prices on one set of products and raise them on another without chang-
ing planned business volumes, while its competitors were put at a disadvantage
through the maneuver because of the difference in barriers to entry of
peripherals and CPU's.



5674

Following the May, 1971 price cuts associated with the FTP, IBM took no
further action to control the peripherals companies until August, 1972. By that
time, the PCM companies had begun deliveries of replacement disks for the high
performance 3330 system on the 370 machines, and replacement memory for the
larger 370 models. In order to prevent a recurrance of the 360 problems on the 370,
IBM prepared a set of price and product revisions, designated the SMASH pro-
gram, which were announced on August 2, 1972.

The 3330 was initially announced in June, 1970 as part of the 370 announcement.
The system consisted of a 3830 controller (which rented for $2400 per month)
which could handle up to four two-drive 3330's which rented for $1300 per month
for two drives. The smallest configuration available cost a total of $3700 (con-
troller and one two-drive bos) and gave the user 200 million bytes of capacity,
equal to seven drives on the 2314. If a user needed more than eight drives, a second
controller was necessary. In August, 1972, the 3830 controller was withdrawn
from the market and replaced by the 3830 II. The 3830 II was similar to the 3830
except that it could control up to sixteen drives instead of eight, and part of the
control electronics was missing. The missing part was put into a modified 3330
&0X, called the 3333. The user needed one 3333 box to go with each eight drives;
the minimum system would be one 3830 II controller and one 3333 and the maxi-
mimi system would be one 3830 II, two 3333's, and six 3330's for a total of sixteen
drives.

The price of the 38.30 II controller was set at $2025 per month ($37.5 lower than
the 3830). the price of the 3333 was set at $1627 per month ($327 higher than the
3330), and the 3330 boxes remained the same price. Thus the total price on a mini-
mum configuration only dropped from $3700 to $3652, a 1.2% decrease. Because of
the higher capacity of the 3830 II than the 3830, tlie cost of a sixteen drive con-
figuration dropped from $15,200 per month to $13,079, a 14% cut. The cuts due to
the 3830 II were relatively minor and easily explainable by technological change
which allowed a higher capacity controller. However, the 3830 II could also be
put under the covers of the CPU. If the 3830 II was put into a 370/135, it was
called an Integrated File Adapter (IFA) and rented for $700 per month instead of
$2025 if it was in a sepai-ate cabinet. If it was put into a 370/145, it was called a
Integrated Storage Adapter (ISC) and rented for $1150 per month. If it was put
into a 370/158 or 168. it was called an Inboard Director and rented for $2200
per month for two 3830 II's able to control up to thirty-two drives.

A single controller was not available inside the CPU of the 158 or 168. The
integrated controllers performed the same function and contained the same cir-
cuitry as the 3830 II stand alone imit. The IBM specifications stated: (35) "The
Director is packaged in two forms : a. as a second level product for integration
into a GPU frame such as the 145, Olympus [158] or Pisces [168] . . . this pack-
aging will be referenced as an IBD (Inboard Director), b. as a model of the
3830."

The low price of the integrated versions of the 3830 II produced substantial
price cuts. On the 370/135, the savings for a complete 3330 system ranged from
37% for a two drive minimum system to 23% for a sixteen drive system, com-
pared with the previous 3330 prices. On a 158 or 168, the price actually went
up for a two drive configuration because of the requirement that two controllers
be purchased. On more normal configurations for the large machines, the cost
of a 16 drive system dropped from $15,200 per month to $13,254 per month
(12.8%) while the cost of a 32 drive system dropped from $30,400 per month
to $24,308 per month (20%).

The 3330 revisions were both a price cut and a product manipulation designed
to make competitive attachment more difficult. The integrated controller meant
that independents would have to redesign their products to match IBM's inter-
face between drives and controllers rather than replacing both di'ives and con-
trollers as a unit. The changed interface between drives and controller through
putting control electronics with one of the drives meant that if an independent
had matched IBM's drive-controller interface in anticipation of integration, he
would still have to redesign his products to make them compatible. In addition
the integration of the controllers allowed IBM to practice price discrimination
and tying together of products without it being as obvious as if the products were
ih separate boxes.

The August, 1972 announcement also culminated a long discussion of how to
protect memory from the PCM companies. The first memory competition had
come in 19()9 from Amiiex on the Large Core Storage units for the 360/50 and 65.
The LCS was a core memory which was much cheaper and slower than main



5675

memory. In early 1970, the competition was extended to main memories for tbe
360. At that time, the amount of competition was insignificant and IBM felt that
it could profitably avoid taking any action to protect the 360 memory, but was
concerned to avoid loss of memory on the 370 machines being prepared for an-
nouncement. In April, 1970, T. V. Learson, IBM president, wrote: (36)

'•Originally, I advised them to price memories and CPU's both at 30%. They
readily agree that CPU's cannot be easily duplicated whereas memories can be
easily duplicated on a plug-in basis. I would, therefore, conclude that we should
bave 25% profit on memories, higher price on CPU's and up [sic] with an overall
profit in the 32% range."

As a result of the concern with possible competition, an IBM group was
authorized to study the potential profitability of competing with IBM in memory
at various IBM prices. Because the primary concern was with new entrants
rather than current competition, the study was done both for a hypothetical
new company and also for an established company moving into the memory com-
petition business. The analysis showed that IBM's primary protection was the
time required for the company to become established. The study assumed that
the company would begin planning and hiring in June, 1970 (at the announce-
ment of the initial 370 machines) and would not ship its first memory until
July, 1972. Because of the necessity for renting at least part of the memories
4ind the discount required over IBM in order to sell, the new company would
not break-even until around 1974 if the IBM price was $16-$18,00O per month
per megabyte [one million character capacity] of memory, and until 1976 at
.$12,000. At IBM prices of $12,000 and above, the summary charts showed good
profitability for the company beyond 1976, and at $10,000 and below, no break-
even until the far future if at all.

For an established component company, the analysis showed that it could
enter memory competition and reach breakeven point by 1973 with an IBM price
of $18,000 per month per megabyte, and by 1975 with a price of $10,000-12,000.
with high profitability in later years for all prices $12,000 and above. The
study concluded that the planned IBM pricing on the 370 ($12,000 per mega
byte) left "marginal viability" for a new company, but that established com-
ponent companies would be likely to enter the field because they could make
20% return on investment for a $10 million investment. (37) The IBM study
showed that after the start up period, the PCM companies would make a 23%
profit margin at an IBM price of $12,000 per month. (38)

Further IBM studies of potential memory competition showed additional
protection. By 1976, IBM expected to have $61 million per month rental value
of memories on 370 machines. Of this total, IBM estimated that 26% was pro-
tected through minimum memory sizes allowed with machines, 36% through cus-
tomers who would resist a mix of IBM and independent equipment, and 15%
because it was in low density geographical areas where the independents were not
expected to compete. This left only 23% of the memory as a possible target
for the independents regardless of the exact price set by IBM. An IBjNI memo
just before the final pricing decision on the initial 370 models states : (40)

"Memory OEM stvidies indicate not much concern over new small companies
starting uv — either ferrite or monolithic; considerable (and, I believe, proper)
concern over established components manufacturers such as TT, Motorola, Fair-
child, etc. Their estimated 23% penetration by end of 1976 is limited primarily
by market factoi's and not significantly by IBM slopes of $5, 6 or 7K. [5K slope
in this context is $5,000 per month per 512,000 bytes of memory, or $10,000
per megabyte] if we assume that the proposed system prices are correct, (which
are based on $6K slope), then any change in slope should be offset in the CPU
prices, I would rather have tbe profit in the CPU and would thus be against a $7K
slope. The $5K slope produces certain gaps in the structure of the line. There-
fore, I recommend we go with the $6K slope.

The exposure that the M9 memories [370/155 and 165 memories] might face
massive FET replacement in 1974 appears impossible. Even CD's schedule would
not permit significant shipment until 1975."

Based on the estimated lag between IBM introduction and competitive ship-
ments and the expected customer resistance to competitive memory, IBM adopted
the price of $12,000 per month per megabyte for the memory on the 370/155 and
165, announced June 30, 1970. The 370/155 and 165 memories were magnetic
cores, the standard computer memory component since the first generation.
The 370/145 introduced an all semiconductor memory, referred to by IBM as
phase 2i or bipolar technology. It was much faster than the existing core mem-



5676

ories and was not considered immediately subject to competitive attack. The price
was set as $9600 per montli per 512,000 bytes, a 60% increase by byte over
tlie core memory price of the 155 and 165. The memory technology was expen-
sive for IBM to develop and the relatively high price for 145 memory still
only yielded a 26% profit margin on memory, below IBM's target of 30%.

By the time of the 370/135 announcement in March, 1971, IBM was beginning
to be concerned about competitive replacement of semiconductor memory as well
as the core memory on the 155 and 165 models. However, because the 135 used
the same technology as the 145, the company felt constrained to change the same
price in order to have a rational pricing pattern. The 370/135 analysis stated :
"Memory sizes installed on the M135 are partially a function of projected PC^NI
memory" penetrations. While it was assumed the PCM impact could occur from
the 96K model on up, Forecasting's understanding of this recent PCM phenom-
enon was still based on limited data, much judgment and wide confidence
limits "(41) Given the memory price determined by the 145, IBM could only
vary the price of the CPU. A low CPU price generated maximum systems
installed and thus the maximum demand for memory and input-output equipment
(subject to replacement by PCM's) while the higher price on the CPU increased
CPU profit but reduced demand for peripherals. The solution found to the
dilemna was to bundle a large amount of memory with the CPU and charge the
lower CPU price. This had the advantage of maximizing overall profits while
protecting at least the bundled portion of memory from competition. The 370/135
analysis stated: (42)

"The more riskv and vulnerable parts of the system, due to PCM, were the
memory and the I/O. Since their contribution to overall profit was most directly
related to CPU quantity, a higher CPU price would result in less quantity and
thus less memory and I/O dependency. The tradeoff then was whether to put
price into the CPU to achieve profit from it instead of from large quantities of
memory and I/O obtained with a low price (and profit) on the CPU.

While Finance favored the higher CPU price, lower quantity ration.^le. the
MRC had a third way of looking at the problem. They felt that raising the
M135's minimum entry model from 72K to 96K of memory made strategic sense
without losing the company many orders — either the custom.er needed 96K to run
his applications, diagnostics, etc.. or he would accept a T55 [370/125] instead.
To make the 90K model easier to afford and ensure a minimmn loss of accept-
ances, tlie MRC decided to lower the price level of tlie CPU by $300."

At the time of the 370/135 announcement. IB:M was beginning to see that its
previous forecasts had underestimated the ability of PCM's to deliver replace-
ment memories, particularly for the core based 155 and 105. A task group was
set up to study alternatives for reducing the threat. Basically, the plans con-
sidered were various combinations of mimimum memories and reduced memory
prices combined v.-ith increased CPU prices. An extreme case involved aban-
doning the concept of variable amounts of memory and introducin'g more models
each with a fixed memory size. The reduction in memory options would protect
the memory market because IB:?! would refuse to sell a CPU without the mini-
mum memory, and at that time it was thought to be difficult or impossil)le to
install memory beyond the IBM specified maximum for the machine model
involved. The Various potential strategies were considered risky because of the
possil)ility of an antitrust suit over tying memory to CPU's and of the IP.:\I credi-
bility problem that would bo raised if the company introduced replacement
machines right after the old ones had been delivered. B. M. Hochfeld, a member
of the task force, wrote in March. 1971 : (43)

"I cannot think of a good rationaTe for drastically increasing memory mini-
mums to provide a very limited performance increment . . . restricting PCMs
from 60-70% of their market without price competition would almost certainly
provoke legal consequences and I wonder what contingency I can provide against
a civil triple damage suit . . .

"Furthermore, the loss of systems seems very likely to exceed gains from
PCISts. For example, a two megabyte 165 excludes one third of the potential
market at that level and many of these customers are likely to be captured
by competition.
" "In addition, our purchase customers are likely to be sorely tried by our
obsoleting of their equipment within eighteen months of acceptance. Their
realization that IBM sold them dead-end equipment with no Advanced Function
capability and no new programming is not likely to endear this customer group
to us and will present some complelx problems to our salesmen."



5677

While Hochfeld was uncouviuced of the wisdom of tying memory to CPU's, he
did feel that IBM ought to price the memory low enough to discourage com-
petitors. He wrote: (44)

"Furthermore, when you deal in monolithic memory, either CP or HMS, where
many of the larger potential competitors such as T. I. Fairchild have not yet
entered the market, it seems clear that their judgment on entry will be signifi-
cantly influenced by our pricing.

". . . We would want to be quite sure that the impact of a price cut on PCMs
is proi)erly identified so that a valid financial analysis can be made."

On July 1"), 1971. the Data Processing Group presented its recommendations
for memory protection to the ilanagement Review Committee. The fiinal recom-
mendations included pricing action and minimum memory but without actuall.v
withdrawing the old machines an€i introducing new ones as had been considered.
The Group recommened increasing the rental and purchase price of the 370
CPU's by 13-23%, while reducing the price of the various 370 memories liy 27-
439^. In addition, mininumi memory sizes would be imposed ranging from 256.000
bytes on the 135 (compared with 96,000 bytes at announcement three months
earlier) to two megabytes on the 165.(45) The Management Review Committee
re.iected the Group's recommendations "on the basis that pricing actions involv-
ing rental increases of more than 5-6% would be an irrational business move."
After further consideration, the Group reported back that "any plan formulated
within paramters outlined by 'SIRC is inadeciuate to meet objectives." After



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 97 of 140)