The Symantec ProxySG bandwidth pooling clause that inflates your 2026 renewal.
The bandwidth pooling clause sits roughly two thirds of the way through a typical Symantec ProxySG enterprise agreement. It is not headlined. It is rarely raised at signature. It defines how concurrent throughput is measured across the buyer's set of ProxySG appliances, and it governs how that measurement maps to the licensed throughput entitlement. The contracts written from 2018 through 2022 treated the pool as a simple sum across appliances. Throughput across the whole estate was averaged to a steady state number and benchmarked against the licensed pool. The 2026 reading we are seeing in renewals and true ups is materially different. The pool is now measured at peak, not steady state. Peak is defined as the highest five minute window in the measurement period. Buyers who built their estate against the old reading are finding that the new reading puts them above the licensed pool, sometimes by 20 to 35 percent, and the renewal quote is being built against the new peak rather than the historical average.
The clause language has not changed in most contracts. The interpretation has. Broadcom's deal desk and audit teams have aligned on a reading of the existing clause that treats peak as the measurement point. The contract language permits the reading, which is why this is a trap rather than a dispute. Buyers who read the clause to mean steady state, and who built their estate to that interpretation, are inside the four corners of the contract being told that the steady state interpretation was never the obligation. The reading is defensible. It is also commercially expensive.
How the clause reads, in plain English
The clause typically defines a licensed bandwidth pool in gigabits per second, lists the in scope appliances, and states that the buyer's total measured throughput across the in scope appliances shall not exceed the licensed pool. It does not define measurement frequency. It does not define averaging window. It does not define peak treatment. The absence of the definition is the trap. The buyer's interpretation in the 2018 to 2022 period, supported by account team conversations, was that the measurement was a fair representation of normal operating load. The buyer's bandwidth telemetry was sized to that. The 2026 interpretation is that the measurement is the highest five minute window observed, and that the licensed pool is the cap on that window rather than on the average.
The five minute window is not arbitrary. It is the default reporting granularity in the ProxySG management plane. The audit teams have read the absence of an averaging definition as a green light to use the granularity the platform reports natively. The reading is consistent with how the platform reports. The buyer side argument that this was never the spirit of the clause is true, and is also weaker than the buyer would like.
Where the price impact comes from
The price impact runs through three paths. The first is the peak versus average gap. In buyer estates we have measured, the ratio of five minute peak to steady state average sits between 1.4 and 2.1. Buyers sized at the steady state are running 40 to 110 percent above the new measurement point. The licensed pool was set against the steady state. The new measurement says the pool is being exceeded. The renewal quote is built on a true up of the difference, plus a forward year licensed pool sized to the new peak, plus an uplift for the new pool size.
"The clause did not change. The reading did. The contract language permits the new reading, and the buyer who signed against the old interpretation is being asked to true up against the new one."Symantec ProxySG Engagement Lead, The Desk
The second path is the support tier reset. Many ProxySG contracts couple the support tier to the licensed pool size. When the pool grows to absorb the new peak, the support tier resets upward in the seller's rate card. The support uplift is typically 12 to 18 percent of the new licensed pool cost. It is rarely separately negotiated. It appears as a line on the renewal and is moved through as if it were arithmetic.
The third path is the multi year hold the seller offers as a concession in exchange for accepting the new pool size. The hold caps the per gigabit rate but does not cap the pool size, which is now reset upward. The buyer reads the hold as a price concession. The seller reads it as a volume lock. Both readings are correct. The lock matters more than the hold over a 36 month term.
What the buyer can do about it
There are three corrective moves we have seen work. The first is a measurement window amendment. The buyer asks the seller to amend the clause to specify the measurement window and the averaging method. Broadcom has, in some renewals, accepted a 60 minute averaging window rather than a five minute peak, in exchange for a small uplift on the per gigabit rate. The trade is usually favourable to the buyer because the 60 minute average is much closer to the original steady state interpretation and the licensed pool requirement drops materially.
The second corrective move is an estate segmentation review. The buyer's network team can often identify ProxySG appliances that are no longer in scope, have been replaced by cloud delivered web gateway equivalents, or are running test workloads that should not have been counted in the pool. Reducing the in scope appliance count reduces the measured total throughput. In our 2025 to 2026 sample, the median in scope reduction available to buyers who ran the review was 18 percent of the appliance fleet.
The third corrective move is a transition concession. Buyers who can credibly state that the ProxySG estate is being consolidated into Cloud SWG over the next 24 months have been able to negotiate a hold on the existing pool size at the old interpretation, in exchange for committing the Cloud SWG ramp. The trade is buyer favourable only if the Cloud SWG move was already on the roadmap. If it was not, the trade locks the buyer into a Broadcom platform they had not committed to.
The numbers
What we have seen on live deals
A European bank's first renewal under the new reading came in 31 percent above the prior contract. The bank's network team produced a 60 minute averaging measurement that, when applied, brought the estate inside the existing licensed pool. The seller initially declined to accept the alternative measurement. The buyer's legal team pointed out the contract did not specify five minute granularity and that the deal desk's chosen granularity was a unilateral interpretation. The seller accepted the 60 minute average in exchange for a 4 percent uplift on the per gigabit rate. The buyer closed the renewal 3 percent below the prior contract rather than 31 percent above, with the measurement amendment written into the new contract.
A North American retailer was less fortunate. The retailer had not measured under either interpretation and could not produce telemetry to support a corrective position. The renewal closed at 19 percent above the prior contract. The retailer has since instrumented the estate and expects to be in a better position at the next renewal anniversary.
The takeaway
- The ProxySG bandwidth pooling clause language in most contracts written before 2023 has not changed. The Broadcom interpretation has shifted to a five minute peak reading that the original buyer signed against a steady state reading. The reading is defensible inside the contract and the renewal quote is being built against the new peak.
- Buyer side corrective moves are available. A measurement window amendment, an estate segmentation review, and a transition concession each move the closing renewal materially. In our 2025 to 2026 sample the median total reduction available after the corrective moves was 21 to 34 percent against the seller's opening.
- The corrective work requires buyer side telemetry. Buyers who cannot produce throughput measurement under their preferred interpretation are operating without the principal evidence the conversation needs. The instrumentation work is unglamorous and is the foundation for every other move.