In the autumn of 2022, as the semiconductor industry weathered a broad consumer downturn, AMD's Lisa Su offered a measured defense of her company's $49 billion acquisition of Xilinx — arguing that the deal's true purpose was never to chase the PC cycle, but to plant AMD more deeply in the durable soils of data centers, aerospace, automotive, and defense. The quarterly numbers were imperfect, yet the strategic architecture held: diversification, not dominance in any single market, was the hedge AMD had purchased. In a moment when the industry's short-term fortunes dimmed, Su's message was essen
AMD CEO: Xilinx acquisition 'playing out exactly as hoped' amid chip downturn
The acquisition thesis has played out exactly as we had hoped
Mark
When Su says the acquisition is playing out "exactly as hoped," what does that actually mean in a downturn? Isn't that when deals usually look bad?
Mimi
Not this one. The whole point of buying Xilinx was to move away from PCs—which are collapsing—into embedded systems and data center, which are still growing. So the downturn is actually proving the thesis right.
Mark
But the company still missed earnings expectations. How is that a win?
Mimi
It missed because of weakness in PCs and gaming. The segments that matter for the future—data center up 45 percent, embedded holding steady—those are doing fine. The miss was expected and telegraphed. The real story is what's not falling.
Mark
What about the intellectual property synergies Su keeps mentioning? That sounds like corporate speak.
Mimi
It's not. She's talking about combining AMD's processor design with Xilinx's AI acceleration technology. That's a real product opportunity. But they've only been under the same roof for nine months. The integration is still ahead.
Mark
How much does Intel's delay on Sapphire Rapids matter here?
Mimi
It matters a lot. AMD gets to ship Genoa first, gain market share, and prove the architecture works before Intel shows up. That's the kind of window AMD has been winning with for years.
Mark
Is there any risk the downturn gets worse and kills the whole strategy?
Mimi
Sure. If the downturn deepens, even data center could weaken. But right now, that segment is the most resilient part of AMD's business. It's the hedge that makes the Xilinx deal look smart.
Le Pouls
- AMD's Q3 revenue rose 29% to $5.57 billion, yet still missed Wall Street's expectations, and the company braced investors for a flat Q4 as PC and gaming markets continued to erode.
- The consumer hardware slowdown exposed how dependent legacy chipmakers remain on cyclical markets — a vulnerability AMD had been racing to escape through the Xilinx deal.
- Xilinx's embedded portfolio — serving aerospace, automotive, telecom, and defense — emerged as a stabilizing force, with that segment and data center the only two expected to grow sequentially into Q4.
- Data center revenue surged 45% year-over-year to $1.6 billion, validating AMD's pivot and signaling that enterprise and cloud demand could absorb shocks that consumer markets cannot.
- AMD's next-generation Genoa data center chip was set to launch within days, targeting Intel at a moment when Intel's competing chip faced delays into 2023 — a rare and deliberate opening.
- Su pointed toward a longer horizon where AMD's processor IP and Xilinx's AI acceleration capabilities merge into a unified portfolio, suggesting the acquisition's most consequential chapter has not yet been written.
In the autumn of 2022, as the semiconductor industry weathered a broad consumer downturn, AMD's Lisa Su offered a measured defense of her company's $49 billion acquisition of Xilinx — arguing that the deal's true purpose was never to chase the PC cycle, but to plant AMD more deeply in the durable soils of data centers, aerospace, automotive, and defense. The quarterly numbers were imperfect, yet the strategic architecture held: diversification, not dominance in any single market, was the hedge AMD had purchased. In a moment when the industry's short-term fortunes dimmed, Su's message was essentially philosophical — that the right acquisition is one whose wisdom becomes clearest precisely when conditions grow difficult.
When Lisa Su spoke with CNBC the morning after AMD's third-quarter earnings, she was not pretending the numbers were flawless. Revenue had grown 29 percent to $5.57 billion — a meaningful gain — but it had still fallen short of analyst expectations, and AMD had already softened the blow with a warning in early October. The fourth quarter outlook was essentially flat, with personal computers and gaming graphics cards expected to weaken further as consumers pulled back.
What Su wanted to talk about was the part of the business that was holding. AMD's data center revenue had climbed 45 percent year-over-year to $1.6 billion, and the embedded segment — the category inherited from Xilinx, serving aerospace, automotive, telecommunications, and defense — was also expected to grow sequentially into Q4. These were the two pillars AMD had been building toward when it paid $49 billion for Xilinx earlier in the year, and in a downturn, their relative resilience made the strategic logic feel vindicated.
The timing of AMD's next move added urgency to the story. The company was preparing to launch its Genoa data center processor the following week, a chip engineered to challenge Intel directly on both performance and power efficiency. Intel's competing next-generation chip had been delayed into 2023, leaving AMD a window it intended to use. Genoa was already shipping to select customers, with broader production ramping through the quarter.
Looking further out, Su expressed genuine enthusiasm about what the Xilinx integration might eventually produce — a combined portfolio weaving AMD's processor architecture together with Xilinx's artificial intelligence and acceleration capabilities. The acquisition had closed less than a year prior, and the deeper work of merging the two companies' technologies was still underway. AMD also maintained that new U.S. export restrictions on advanced chips to China would not materially affect its business. The near-term picture was complicated, but the direction Su was pointing remained consistent: away from consumer cycles, and toward the infrastructure that powers everything else.
Lisa Su sat down with CNBC the day after AMD released its third-quarter earnings, and the message was clear: the company's $49 billion bet on Xilinx, which closed in February, was working out. Not perfectly—the semiconductor industry was in a downturn, and AMD's results had come in slightly softer than Wall Street expected—but the strategic logic behind the deal was holding up.
The numbers told a mixed story. AMD's third-quarter revenue climbed 29 percent to $5.57 billion, a solid gain on paper, yet it still fell short of analyst forecasts. The company had already warned in early October that things would be disappointing, so there were no real surprises. Looking ahead to the fourth quarter, AMD projected revenue of roughly $5.5 billion, essentially flat from Q3. The weakness was concentrated in the traditional consumer markets: personal computers and gaming graphics cards were both expected to deteriorate further as the macroeconomic environment tightened.
But here's where the Xilinx acquisition mattered. The only two segments AMD expected to grow sequentially in the fourth quarter were data center and what the company calls "embedded"—the category where Xilinx's business lives. That embedded portfolio serves aerospace, automotive, telecommunications, and defense industries, markets that were holding up better than the consumer PC business. Su told Jim Cramer that the pairing had already begun to pay dividends. "With the Xilinx portfolio, we've been able to actually help accelerate the growth in that embedded business when you look at the supply chain and just all the demand for those products," she said. "It's worked out very well."
The data center division was the real bright spot. AMD's data center revenue had surged 45 percent year-over-year in Q3 to $1.6 billion, and the company expected that segment to grow even further in Q4. This resilience mattered because it showed that AMD's diversification strategy—the whole reason for buying Xilinx—could cushion the company during industry downturns. While PCs and gaming cratered, data center and embedded kept the lights on.
Su was particularly bullish about what comes next. AMD planned to unveil its new Genoa data center processor the following week, a chip designed to compete directly against Intel's Sapphire Rapids. The company had already begun shipping Genoa to select customers, with production ramping up through the fourth quarter. Su emphasized that Genoa would extend AMD's advantage not just in raw performance but in power efficiency—a growing concern as energy costs climbed globally. The timing was significant: Intel's next-generation chip had faced delays and wouldn't arrive until 2023, giving AMD another window to gain ground.
Looking further out, Su saw even bigger opportunities in combining AMD's processor technology with Xilinx's artificial intelligence and acceleration capabilities. "I'm actually even more excited about what we'll see in the future as we integrate some of our products together with the AMD processor IP with some of the Xilinx AI and other acceleration IP," she said. "We think it's going to be a fantastic portfolio." The acquisition had closed less than a year earlier, so the real work of stitching the two companies together was just beginning.
One last detail: AMD said that new U.S. export restrictions on high-end chips to China wouldn't materially affect its business. The company had made the same claim when the rules were first announced in August, and it was sticking to that line. For now, the focus remained on data center dominance and the long-term potential of the Xilinx deal.
Citations marquantes
With the Xilinx portfolio, we've been able to actually help accelerate the growth in that embedded business when you look at the supply chain and just all the demand for those products. It's worked out very well.— Lisa Su, AMD CEO
I'm actually even more excited about what we'll see in the future as we integrate some of our products together with the AMD processor IP with some of the Xilinx AI and other acceleration IP. We think it's going to be a fantastic portfolio.— Lisa Su, AMD CEO
Envie de l'histoire complète ?
Lire l'original sur CNBC ↗
Business