Summary

  • Microsoft can earn two tranches of as many as 400,000 A10 Networks shares, at an exercise price of $0.01 a share, if purchases by Microsoft and its affiliates satisfy thresholds measured through 30 June 2027 and 30 June 2028. The actual spend thresholds and separate exercise conditions are redacted.
  • The warrant calls itself a sales discount, allowance or rebate to a customer and says A10 will recognise the share-based consideration as a reduction of revenue. It is therefore an incentive attached to sales, not evidence that the sales have already become backlog or revenue.
  • The 800,000-share ceiling is about 1.10% of A10’s 72.6 million shares outstanding on 31 July. That is a useful scale marker, not a forecast: no share vests without the contractual conditions, the denominator can change, and Microsoft may use cashless exercise.

The share count is public; the purchase threshold is not

A10 Networks filed the warrant on 5 August, two days after issuing it to Microsoft Corporation. The visible terms are unusually specific. Microsoft may purchase up to 800,000 A10 common shares for $0.01 each. The entitlement is divided into two blocks of 400,000 shares. The first purchase-measurement period ends on 30 June 2027 and the second on 30 June 2028. If the relevant threshold is missed, that tranche expires unvested.

Exercise comes later. Shares in the first tranche cannot become exercisable before 1 January 2028; the second tranche starts no earlier than 1 January 2029. The warrant’s drop-dead date is 3 August 2036, subject to its extension provisions. Microsoft can pay cash or use the cashless formula, which cancels a small part of the entitlement in place of paying the strike price. At the maximum share count, a cash exercise would send only $8,000 to A10.

That last number shows why the strike price should not be treated as the transaction’s value. One cent is the exercise price for a contingent equity incentive. It is not the price Microsoft pays for A10 products, the fair value of the warrant or the value of the commercial agreement.

The decisive schedule is missing from the public copy. Exhibit E, which contains the vesting schedule, is wholly redacted. Exhibit F, which contains the exercise conditions, is redacted too. The 8-K tells investors that vesting depends on “specified purchase thresholds”, but not whether the thresholds are measured in tens of millions, hundreds of millions or another amount. It also does not reveal whether vesting increases gradually inside each 400,000-share tranche or arrives at discrete gates.

The instrument is therefore a meter with an unreadable scale. Future filings may show that the needle moved. Today’s filing does not let an outsider convert 800,000 possible shares into a minimum dollar purchase commitment.

The warrant describes a rebate, not an investment cheque

Section 14 supplies the document’s most important economic label. A10 and Microsoft state that they intend to treat the warrant as a sales discount, allowance or rebate connected with Microsoft’s purchase of A10 products and services. A10 is to account for it as share-based consideration payable to a customer under ASC 718 and ASU 2025-04, recognised as a reduction of revenue under ASC 606.

That treatment changes how the headline should be read. A conventional supplier says that a customer buys equipment and the supplier records revenue. Here the supplier can also deliver equity value to the customer as qualifying purchases accumulate. The accounting aims to show that consideration on the same side of the ledger as a discount rather than presenting it as an unrelated corporate investment.

FASB’s 2025 clarification addresses exactly this structure. Share-based customer consideration may take the form of a warrant and may vest when a customer reaches a specified purchase volume or amount. Unless the supplier receives a distinct good or service in exchange, Topic 606 generally reduces the transaction price and revenue by the customer consideration. The measurement and vesting assessment use the stock-compensation guidance.

This does not mean A10 will subtract 800,000 multiplied by some current share price from one quarter’s sales. The amount and timing depend on the accounting measurement, the probability and pattern of vesting, the commercial facts and future reporting. The filed contract establishes the direction of presentation, not the final number.

It also gives the arrangement an incentive asymmetry. Microsoft’s purchase decisions help determine whether it receives low-strike A10 equity. If the resulting business lifts A10’s value, the customer can participate in that upside after satisfying the conditions. Existing shareholders receive the sales opportunity but may bear both revenue reduction and dilution. Whether the exchange is attractive depends on the purchases and cash contribution that the public thresholds do not reveal.

Why this is not backlog

Backlog language compresses several stages that should remain separate. A commercial framework can establish eligible products, pricing, governance and a future buying channel without making every prospective purchase non-cancellable. A purchase threshold can motivate spend without guaranteeing that the threshold will be reached. A vested warrant can evidence qualifying purchases without proving the margin or cash conversion of those purchases.

A10’s own filing defines remaining performance obligations as contracted, non-cancellable revenue not yet recognised because work remains unsatisfied or partly satisfied. At 30 June, the company reported $154.826 million of RPO: $93.230 million expected within one year, $48.702 million over the next two to three years and $12.894 million thereafter.

That snapshot predates the 3 August warrant. The filing does not say that the RPO includes Microsoft’s arrangement, nor does it split the balance by customer. It would be equally unsafe to take management’s description of a “multi-year agreement” and insert an invented contract value. Duration tells the market how long a relationship can operate. It does not reveal purchase minimums.

The right evidence ladder is narrower. First comes a commercial arrangement. Then qualifying orders may be placed. A10 must deliver products or services and satisfy its revenue-recognition obligations. Purchases count against the private vesting formula. Share-based customer consideration reduces reported revenue. Cash must still be collected, and only then can the sales opportunity be assessed against working capital, margin and dilution.

Calling the first rung backlog erases every rung that follows.

The unnamed 38% customer remains unnamed

A10 entered the arrangement from a position of rapidly rising customer concentration. Its June-quarter filing says one unnamed end customer represented 38% of revenue in both the quarter and the first half. The ten largest end customers supplied 58% of quarterly revenue. A single distribution channel represented 40% of quarterly revenue and 57% of gross receivables at period-end.

Those figures matter because a customer warrant can deepen the economics of a large relationship. They do not identify Microsoft as “Customer A”. The quarter closed on 30 June; the warrant was issued on 3 August; and A10 keeps the significant customer anonymous. Any sentence that equates the two would replace evidence with inference.

The disclosed concentration still sets the risk frame. A10 says its large-customer orders are often sizeable but irregular, follow long sales cycles and are difficult to predict. An accelerated order can lift a quarter; a delay can move revenue across reporting periods. The Microsoft warrant does not abolish that timing risk. It puts an additional incentive and accounting layer around qualifying purchases.

The mix also matters. Enterprise customers supplied 60% of second-quarter revenue, up from 40% a year earlier, while service providers fell to 40%. Product revenue rose to $49.024 million from $39.173 million, while service revenue edged to $31.113 million from $30.210 million. Management linked the new agreement to scaled AI deployment. The market test is whether that narrative becomes repeatable, cash-converting product and service demand without making one buyer the whole growth story.

The dilution ceiling is a scale marker, not an outcome

A10 reported 72,609,021 shares outstanding on 31 July. Against that fixed denominator, 800,000 shares equal roughly 1.10%, and one 400,000-share tranche equals roughly 0.55%. This arithmetic makes the maximum warrant understandable. It does not predict Microsoft’s eventual ownership.

The numerator is conditional. Shares must vest and become exercisable. Cashless exercise would issue slightly fewer shares than a full cash exercise. The denominator can move through employee awards, repurchases, settlement of A10’s convertible notes and other capital actions. Microsoft receives no present voting ownership merely because the warrant exists.

Nor should dilution be judged in isolation. A supplier may rationally exchange a bounded equity claim for a purchasing relationship that expands revenue, improves factory absorption or validates a product in a difficult market. The comparison must be between the incremental economic value of qualifying sales and the full customer consideration, service burden and concentration risk required to win them. The redacted thresholds prevent outsiders from completing that comparison today.

That is the real disclosure gap. Investors can see the maximum equity incentive and its time windows. They cannot see the purchase denominator against which the incentive should be priced.

What can actually prove the agreement

The first proof will not be a press-release adjective. It will be the accounting trail. A10 will need to measure the warrant, assess vesting, record customer consideration and explain material effects. Later spend certifications can determine whether either tranche vested. Changes in shares outstanding and diluted earnings per share will show whether contingent dilution became actual or probable.

Operating evidence matters just as much. At 30 June, total inventory had risen to $31.729 million from $18.032 million at year-end, with raw materials more than doubling to $22.375 million. Deferred product revenue had increased to $17.452 million from $2.783 million. Those movements predate the warrant and are not attributed to Microsoft. They are useful baselines, not deal proof.

The agreement earns its strategic label only if later reports connect demand to delivery: product revenue, gross margin, receivables, operating cash, RPO and customer concentration must move in a coherent way. A large order that consumes cash, receives a substantial equity rebate and fails to repeat could be less valuable than its gross revenue suggests. A durable deployment with attractive net revenue, service pull-through and controlled concentration could justify the incentive.

Until those facts arrive, the filed warrant should be read for what it is: a contingent price paid by a supplier to encourage a customer’s purchases. It makes future spending more likely to matter. It does not make future spending a present fact.

Sources