Arm's lackluster Q2 outlook overshadows strong Q1 revenue growth
Execs blame gap in licensing revenues for lost momentum, while hawking AI
Brit chip design champ Arm posted its fourth consecutive quarter of growth on Wednesday with Q1 revenues up 39 percent year-over-year to $939 million and profits of $233 million.
But alas, the good times appear to have come to an end, with the chip shop forecasting substantially lower revenues for the second quarter.
In the second quarter, Arm expects to net $780 to 830 million – a decline of $109 to 159 million dollars compared to the prior quarter, depending on where the chips fall.
Speaking with analysts on Wednesday's earnings call, CFO Jason Child blamed the lacklustre outlook on the "timing of revenue recognition from licensing."
Child expects Q2 to be the lowest point for Arm during the 2025 fiscal year. But while revenues are forecast to fall, he predicted bookings to increase considerably in Q2.
The Q2 dip isn't expected to impact Arm's prior fiscal year revenue forecast, which remains unchanged at between $3.8 and $4.1 billion. Arm anticipates the largest share of this growth will come in the fourth quarter and be driven in no small part by growing demand for AI chips. However, this didn't stop Arm's share price from plummeting more than 12 percent in after-hours trading.
Speaking with analysts on Wednesday's call, CEO Rene Haas was keen to highlight AI as a growth driver during the first quarter of the 2025 fiscal year.
"We are seeing AI everywhere, which is driving demand for Arm's performant and power efficient compute platform," he explained.
While most large AI models are trained and inferenced on large clusters of GPUs and dedicated accelerators, these parts still require CPUs to manage them. "Every chip designed today requires a CPU and these are being designed with Arm in mind," Haas observed.
"The ultimate benefit of using Arm for an AI datacenter is customization," he added, highlighting Nvidia's Grace-Hopper GH200 Superchips as a prime example.
Announced in 2022, the GH200 marries a 72 Arm Neoverse V2 core CPU with an H100 GPU over a 900GB/sec NVLink-C2C interconnect. The chip's successor – dubbed Grace-Blackwell, which we looked at in depth last March – will use the same Grace CPU but pair it with twin Blackwell GPUs.
"We expect to see a similar trend with other accelerators that use Arm," he noted.
- Samsung confirms Exynos 2500 chip as profit skyrockets on back of AI memory boom
- Intel prepares to cut 'thousands' of workers
- AMD sold $1B of Instinct GPUs last quarter, driving triple-digit datacenter growth
- No love lost between Apple and Nvidia as iGiant chooses Google chips for AI training
On the call, Haas also touted Google's Axion CPUs announced this northern spring and AWS' Graviton 4 as evidence of Arm's growing influence in the cloud sector. Another example is Microsoft's Cobalt CPU, announced last fall, based on an Arm Compute Subsystem (CSS) design.
On the client side, Haas briefly touched on the raft of Copilot-compatible Windows-on-Arm AI PCs, but stopped short of actually naming Qualcomm. Considering that Arm is embroiled in an ongoing legal battle over the cores used in those systems, that's not surprising.
Along with AI, execs noted that the smartphone segment showed signs of recovery – royalties from sales were up 50 percent from the year ago quarter.
Of the $939 million in sales Arm realized in Q1, $476 million came from royalties (up 17 percent year on year), while licensing revenues topped $472 million – an increase of 72 percent from this time last year.
Haas told investors fresh license deals will eventually bear fruit between three and four years down the line. "I think it's a very, very good predictor of royalty growth." ®