How Microchip turned the downturn around

Microchip Technology has been through tough times. Steve Sanghi, President and CEO of Microchip Technology, explains in an interview with Markt & Technik what measures were necessary to get the company back on track.

Markt & Technik: The semiconductor market is cyclical, yet Microchip was profitable for 104 consecutive quarters. 2024 and 2025 were exceptions. What were the main reasons for this?

Steve Sanghi: In those two years, Microchip mismanaged the post-coronavirus cycle from an operational perspective. We had successfully navigated previous cycles, such as the Asian financial crisis of 1997/98, the internet downturn of 2000 and the financial crisis of 2008/2009. At that time, we were considered ‘cycle mavens’ because we recognised the cycles early on and managed our business in such a way that no layoffs were necessary and we remained profitable throughout.

The 2022 to 2024 cycle was very deep, but that alone does not explain our situation. The decisive difference was our own actions. We had built up excessive inventories with our distributors and customers and were extremely rigid in dealing with existing orders. Customers could neither adjust nor postpone orders, even if they no longer needed the components.

At the same time, we invested heavily in building up manufacturing capacity to meet demand that later turned out to be unrealistic.

When customers and distributors began to correct their inventories, new orders declined sharply. As a result, our sales fell by 60 per cent and our profits fell by as much as 93 per cent. Looking back, it is clear that Microchip managed this cycle very poorly, and that was the key difference from previous cycles.

Today, the situation has improved significantly. You have announced plans to reduce operating costs to around 25 per cent in the long term. How do you intend to achieve this goal?

Historically, our operating costs have been around 24 per cent. During the post-Covid boom, our sales were very high, which led to extensive new hiring.

The company assumed that this revenue corresponded to real demand. When revenue then slumped by 60 per cent, these costs initially remained in place. When I returned as CEO in November 2024, our operating costs were 39 per cent of our revenue – driven by the slump in revenue and an excessive number of employees.

After 25 years of steering Microchip through cycles without layoffs, I realised that I had to break with this tradition. In March 2025, we therefore carried out our first global workforce reduction in 25 years, cutting around 2,000 jobs out of a total workforce of around 20,000 worldwide. Shortly afterwards, sales bottomed out and began to grow again. Together, these two factors have led to a reduction in our operating costs from 39 per cent to 32 per cent.

In the long term, we are aiming for around 25 per cent. We deliberately plan to allocate an additional percentage point or so to investments in future areas such as connectivity, networking, high-performance computing, FPGA and AI. The further reduction from 32 per cent to 25 per cent will be achieved primarily through sales growth, supplemented by productivity gains, for example through the use of AI in administrative processes.

What other measures were crucial to the turnaround?

Shortly after I returned to Microchip, I developed and implemented a nine-point plan to restructure the company. The first point concerned our manufacturing, which was simply too large for the size of our business. Accordingly, we had to make the difficult decision to close our Fab 2 in Arizona and our back-end fab in the Philippines, plus make minor adjustments in Thailand, but without layoffs, as this could be achieved through natural fluctuation.

The second point concerned the reduction of our own inventories. At 266 days, these were about twice as high as our target of 130 to 150 days. By reducing the utilisation of the remaining fabs and increasing deliveries from the warehouse, we were able to reduce inventory to around 199 days.

The third point was an in-depth analysis of all business areas. Five strategic pillars were formed from around 20 business units: ‘Analogue’ and “Microcontrollers” – this is what Microchip is known for. This is not the case for the other three, where we have restructured to strengthen them, including substantial investments. One of these is the ‘Connectivity and Networking’ division. Microchip recently introduced the world's first 6th generation PCI switch, which is manufactured using 3 nm structures. We also introduced a new family of 10BASE-T1S endpoint ICs for automotive applications. This area also includes our other components for various communication standards such as WiFi, Bluetooth, etc.

The fourth pillar is high-performance computing. This is where our FPGA activities are based, and Microchip is now the third-largest FPGA supplier alongside Lattice. This area also includes our timing products, which are another important product segment, as every electronic device needs a clock generator.

The fifth pillar is based on our licensing business and on ‘AI on the Edge’, a new business unit at Microchip. This area is not about high-performance GPUs, but about intelligence for the edge. Microchip recently licensed AI accelerators from CEVA, which are now being specifically integrated into our MCUs, network products, FPGAs, etc.

The idea behind this is as follows: let's take a car with eight cameras. You can send the raw data from each camera directly to a central GPU. Or you can bring intelligence close to the camera, i.e. ‘on the edge’. There, you can pre-process images, distinguish moving objects from stationary ones, separate shadows from real people, recognise three-dimensional information, and so on.

Microchip explicitly emphasised its power semiconductor activities for a while, but they no longer appear...

These components belong to our analogue division. Within this platform, we have our own analogue power business unit. There is also a mixed-signal business unit, a high-reliability business unit and our discrete business units.

So there are several business units that are grouped under the analogue or analogue power platform. This type of organisation was the third point in our nine-point plan.

The fourth point concerned rethinking megatrends. 5G is largely complete as a megatrend and has been replaced by ‘AI at the edge’. ADAS has not been removed, but has been reclassified as part of ‘Networking and Connectivity,’ as this approach extends beyond the automotive market to include industry, robotics and data centres. These megatrends serve primarily as clear priorities for sales and investment.

The fifth point concerned distribution. While other semiconductor manufacturers have significantly reduced margins and largely use distribution only for processing, we have only made moderate adjustments to margins in favour of Microchip, which was received positively by distributors, who had feared worse.

Point six concerned our customers. Our rigid stance after Covid and several price increases had cost us trust. A global programme of direct customer dialogue – including factory visits and on-site appointments – helped to largely normalise relationships within a year. This led to new sales growth, numerous design wins and a significant rapprochement.

Point seven was the introduction of a new long-term financial model with clear target values. The model used by the capital market until then dated back to 2022. It still contained many remnants from the peak period: extremely high gross margins, surcharges for rush orders, penalties for customers who did not accept parts. The model therefore included a lot of income that did not come from the actual product business.

Our new long-term model envisages a gross margin of 65 per cent, operating costs of 25 per cent and an operating profit of 40 per cent. And we are making very good progress in this direction.

Our gross margin bottomed out at 52 per cent, and now we are very close to getting back to a 6 in front of it. We have already talked about operating costs, and our operating profit bottomed out at 14 per cent, but we have now more than doubled it.

The eighth point defined the specific path to cost reduction, including staff cuts. This has reduced our costs by about half, and we want to achieve the other half through growth.

And the ninth point concerned the CHIPS Act? We had already sought support from the CHIPS Act, but during this process there was a change of government. Now, government money is no longer being given away; instead, the government wants shares in the company, as was the case with Intel, for example.

We have come to the conclusion that we do not need government money if we have to give up shares in return. If we need capital, we would rather do so through investors than through the government. That is why we have put our CHIPS Act activities on hold for the time being.

Are political tensions an additional cycle driver?

They make management more difficult, but they do not deepen the cycles. This is because the resulting effects are felt throughout the entire cycle. China is continuously pushing ahead with the development of its own domestic semiconductor industry, and the West will permanently lose part of its business there. At the same time, many Chinese manufacturers who export are relocating their production out of China – to Vietnam, to Malaysia – in order to avoid US tariffs. And Microchip is following suit. Whereas these customers used to buy from us in China, they now buy in Malaysia or Vietnam. So you may see a decline in sales in China, but those sales reappear elsewhere. In that sense, it doesn't deepen the cycle.

Of course, we have also lost business, whether with Huawei, ZTE, FiberHome or other Chinese security and surveillance companies.

Microchip is one of the last companies still active in the 8-bit MCU market. In your last presentation to investors, you emphasised the importance of PIC64 components. Does that mean that Microchip is also slowly withdrawing from the 8-bit segment?

No, you have to look at that separately. PIC64 came from a completely different business unit because its development was based on a contract with NASA. The background: the processor used in space today is the same one that was used in the Apollo missions – around 60 years ago.

About four years ago, NASA issued a call for tenders for the next-generation processor. The specification was clear: it had to be based on RISC-V – not ARM or anything else – and it had to be a radiation-hardened chip suitable for deep space missions. Microchip prevailed over all other suppliers and won the contract. We then developed this high-performance processor. Now we are taking this design and developing commercial versions for automotive and industrial applications.

This development has no impact on our classic MCU portfolio. Today, 32-bit MCUs are the largest segment, followed by 8-bit and 16-bit. We have significantly reduced our investments in 8-bit and shifted some of these investments to 32-bit. But we will continue to bring new 8-bit products to market.

What do you expect from 2026?

I think 2026 will be a good year for us – but also for the entire industry. From my point of view, the industry is currently dividing into two camps: AI and non-AI.

No concerns that the AI bubble will burst?

I'm not even sure if AI is already a bubble. If it is, then it's definitely not yet clear when it will burst – whether in two, three or four years. But one thing is clear to me: it won't be in 2026. On the contrary, 2026 will be a great year for AI. More and more companies are turning to AI, so investment will continue, power plants will be built, data centres will be built. AI will therefore continue to grow strongly this year.

And what about the non-AI sector?

Non-AI markets were really weak in 2024 and 2025. But these areas are now also growing strongly. We have doubled our profits from their low point, and our earnings per share have almost quadrupled compared to the low point.

And we are entering 2026 with this momentum: continuing to increase sales, continuing to increase gross margins, continuing to increase operating profits. That is why I believe that 2026 will be a very good year for Microchip.

Is this optimism somewhat surprising?

But it is justified: our customers and distributors have now used up their inventories, business has returned to normal and new orders are coming in. In addition, few new products were developed in 2022 and 2023, simply because the development engineers were busy qualifying replacement products. When the components became available again in 2024 and 2025, the development engineers returned to their actual tasks, i.e. developing new products. And these new products are now – after about two years – going into production. This is currently driving strong growth.

In addition, many suppliers such as STMicroelectronics, Infineon and onsemi have focused heavily on electromobility and invested heavily in SiC. However, EV growth failed to materialise because electric vehicles were very expensive and the market shifted back towards hybrids and combustion engines. We are not affected by this phenomenon, so there is no reason for us not to be optimistic about this year.

Matchmaker+ Anbieter zum Thema

zu Matchmaker+