Cisco Says It's a Software Company Now. It Bought That Identity for $28 Billion.

Here is a number Cisco wants you to remember: in fiscal 2024 its annual recurring revenue hit $29.6 billion, up 22% in a single year, with total subscription revenue reaching $27.4 billion — 51% of all revenue.¹¹⁰ It is the kind of figure a CFO puts on the cover slide — proof that the company that once sold you a $40,000 router has finally become a software business that bills you forever. And here is a number from the same filings that Cisco would rather you set aside: $4.3 billion of that ARR did not come from a single customer signing up for anything. It came from a wire transfer.¹ Cisco didn't grow that revenue. It bought it.

The official story is that Cisco engineered a textbook transformation — from selling boxes to renting outcomes, from one-time hardware sales to recurring software. The real story is quieter and more expensive: the model conversion is genuine, but the headline that announced its triumph was largely manufactured by the single largest acquisition in the company's history, while the engine that actually pays Cisco's bills was quietly stalling.

The pivot you can see, and the one you can't

Start with what is real, because it is real. Cisco began moving its existing software toward subscriptions years ago, and the work shows. By FY2024, 89% of all software revenue was subscription-based, and software subscription revenue itself grew 15% to $16.4 billion.¹ That is not a press-release illusion — that is a billing model genuinely rebuilt under existing products. When a maintenance contract becomes a license becomes a subscription, the same customer relationship simply gets repriced into a smoother, more durable stream. Investors pay more for a dollar of recurring revenue than a dollar of one-time hardware, and Cisco earned that re-rating on its software base honestly.

The trouble starts when you move from percentages to dollars. The eye-catching figure is not the 89% — it is the leap in absolute ARR to $29.6 billion and the 22% growth rate. And that leap has a name. Cisco completed its acquisition of Splunk on March 18, 2024 for roughly $28 billion in cash, the biggest deal it has ever done.² Splunk arrived carrying $4.3 billion of ARR on its back.¹ Strip that out, and the organic ARR growth is a fraction of the advertised 22%. The model conversion is the slow-burning truth; the headline number is the acquisition wearing it as a costume.

What the subscription glow is hiding

A pivot to software is supposed to be the answer to a hardware business that has stopped growing. In Cisco's case it is also a flattering light cast over one. FY2024 total revenue fell 6% to $53.8 billion — the first annual decline since FY2020, when revenue fell roughly 5% — and product revenue dropped 9% on a segment basis.¹⁴⁹ The decline was not spread evenly. Networking, still by far the company's largest segment at $29.2 billion of product revenue, bore the brunt: after a modest gain in Q1 FY2024, it posted double-digit year-over-year declines in each of the final three quarters of the fiscal year — down 12%, 27%, and 28% respectively.¹¹ Then came the quarter that made the trend impossible to wave away: in Q1 FY2025, networking revenue fell 23% year over year to $6.75 billion, as enterprise budgets visibly rotated toward AI infrastructure and cybersecurity.⁵ That is the core franchise — the thing Cisco has owned the way Coca-Cola owns the red can — contracting by nearly a quarter in a single year.

This is the cannibalization choice in its trickiest form. The classic version — Netflix killing its own DVD business, or a software firm torching license revenue to chase subscriptions — is a deliberate decision to eat your own margin for a better future. Cisco's version is different: the core wasn't being cannibalized by the new model so much as eroded by the market underneath it, and the software story was being used to paper over the gap. The recurring revenue isn't replacing the hardware that's shrinking. It's just changing the subject.

The bill for buying an identity

Acquiring revenue is faster than building it, but it is not free, and Cisco's own filings are unusually candid about the cost. The Splunk deal did not magically lift earnings. Cisco's SEC filing disclosed that the acquisition had a negative $0.04 impact on non-GAAP EPS for fiscal 2024, with accretion not expected until fiscal 2026 — not FY2025, as some commentary assumed.³ In other words, Cisco paid $28 billion for $4.3 billion of ARR and a more impressive subscription mix, and the immediate effect on per-share earnings was negative. The recurring-revenue story bought the company a software valuation argument; it did not buy near-term profit.

That single observation is the tell. FY2025 looks like recovery on paper — total revenue up 5% to $56.7 billion, subscriptions at 56% of revenue, operating cash flow up a striking 30%.⁶⁷ But growth that disappears the moment you remove one acquisition isn't organic recovery; it's the consolidation effect of folding a full year of Splunk into the comparison. The business is lapping its own purchase. Take Splunk out, and the underlying line was still going the wrong way.⁸

The fair case: maybe buying it was the smart move

The honest objection is that this read is too cynical, and there's a real argument on the other side. Building a credible security-and-observability software franchise from scratch, fast enough to matter, may simply be impossible for a hardware incumbent — the talent, the products, and the install base all live elsewhere. Splunk gave Cisco a genuine, market-leading asset and a foothold in exactly the AI-and-security budget categories that are draining away from networking.⁵ If those are the budgets of the next decade, paying $28 billion to be in them is defensible, even shrewd. And the cash generation is not fake: $14.2 billion of operating cash flow, up 30%, is a real number a hardware-in-decline story would struggle to produce.⁶ A pivot that takes a year or two to turn EPS-positive is not a failed pivot — it's an investment with a payback schedule, and Cisco's own filing puts accretion at FY2026.³

All true. The point is not that Cisco made a bad decision. The point is that the scoreboard it's holding up — ARR up 22%, subscriptions past half of revenue — describes an acquisition's arithmetic far more than a transformation's. Those are two different claims, and conflating them lets a company collect credit for organic conversion it hasn't fully earned while the segment that still funds everything quietly shrinks.

Cisco is doing something genuine and something flattering at the same time, and it's letting the second borrow the credibility of the first. The model conversion — license to subscription, sale to stream — is the slow, real, hard-won part. The $29.6 billion ARR with its 22% growth rate is the fast part, and the fast part came with a $28 billion price tag and a negative EPS hit on arrival. A company can buy its way into a software identity. What it can't buy is the thing the number was supposed to prove: that it grew into one. Cisco is becoming a software company. It's just paying retail to look like it already is.