How Hock Tan Built Broadcom into an Indispensable Power in the AI Boom
By Ben Lee | 01 Jul, 2026

Tan's knack for making key acquisitions and applying financial discipline has made Broadcom not only highly profitable but also indispensable to both the hardware and software sides of the AI revolution.

Hock Tan, the 73-year-old CEO of Broadcom, rarely gets the same billing as Jensen Huang of Nvidia, Sam Altman of OpenAI, and the researchers building the models reshaping the world. 

Tan doesn't give TED talks about the future of humanity.  He does not post on social media.  He is not given to grand pronouncements.  What he does, with the consistency of a Swiss watch, is acquire companies, cut costs, protect franchises, and compound returns. Over two decades of doing exactly that, Tan has quietly assembled one of the most consequential technology companies in the world — and when the AI boom arrived, Broadcom was already perfectly positioned to profit from it.

Gifted Malaysian Scholarship Student

Hock Tan was born in 1951 in Penang, Malaysia, the son of a modest family in a country still finding its postcolonial footing. In 1971, he earned a scholarship to the Massachusetts Institute of Technology, the most coveted destination in the world for engineering talent. By 1975, he had completed both a bachelor's and a master's degree in mechanical engineering — a remarkable achievement for a young man who had crossed the Pacific with little more than intellectual ambition.

He went on to earn an MBA from Harvard Business School, where he absorbed the financial and strategic frameworks that would later define his business approach. His early career took him through General Motors and PepsiCo, two enormous industrial organizations where he learned how large enterprises are managed, optimized, and turned toward profit. He then returned to Southeast Asia, serving as Managing Director of Hume Industries in Malaysia before co-founding Pacven Investment, a venture capital fund based in Singapore.

These experiences gave Tan something rare: a dual fluency in engineering and finance. He understood what made technology tick and what made businesses generate cash. That combination, it would turn out, was the precise toolkit the semiconductor industry needed.

MIT-Harvard Business School Playbook

In 1999, Tan took over as CEO of Integrated Circuit Systems, a small Pennsylvania chipmaker that most people in the technology world had never heard of. It was here that the Hock Tan playbook began to crystallize. The formula was not complicated: find companies with valuable, mission-critical technology embedded deep in their customers' infrastructure; acquire them; eliminate redundancy; raise margins; and repeat. ICS became a proving ground for this approach, and when the company was sold to Integrated Device Technology, Tan was ready for a larger stage.

That stage arrived in the form of Avago Technologies. In 2005, a consortium of private equity investors bought the semiconductor division of Agilent Technologies — itself a spin-off of Hewlett-Packard — for $2.66 billion, and brought Tan in as president and CEO. Avago was a sprawling collection of semiconductor product lines serving wireless communications, industrial automation, fiber optics, and enterprise storage. It was, in short, exactly the kind of undervalued, underleveraged technology asset that Tan had learned to love.

Over the next decade, he transformed Avago through a series of increasingly bold acquisitions. He bought LSI Corporation, a storage and networking chip company, for $6.6 billion in 2014. He acquired PLX Technology, Emulex, and Brocade Communications. Each deal followed the same rhythm: identify the enduring, hard-to-replace product lines within the target company, preserve and strengthen those franchises, and jettison everything else. Customers sometimes grumbled. Employees braced for layoffs. But the financial results were, by almost any measure, extraordinary.

Avago-Broadcom Transformation

In 2015, Tan engineered the transaction that would make him a semiconductor legend: a $37 billion acquisition of Broadcom Corporation, one of the most recognizable names in networking chips. The deal was remarkable not only for its size — it was one of the largest semiconductor acquisitions in history at the time — but for its audacity. Avago was technically the acquirer, but it was buying a company far better known than itself. Tan's solution was characteristically pragmatic: keep the Broadcom name. Why fight brand recognition when you can simply absorb it?

The newly enlarged Broadcom did not pause. In 2018, Tan reached for Qualcomm in what would have been the largest technology deal ever recorded, a $103 billion bid that briefly convulsed the industry. President Donald Trump blocked it on national security grounds, citing concerns about Chinese influence in the semiconductor supply chain. It remains the deal that got away — though, in retrospect, the failure may have been a blessing in disguise. Qualcomm's integration challenges would likely have consumed years of management bandwidth, and the $103 billion price tag would have been a staggering debt burden.

Undeterred, Tan turned toward enterprise software. The $18.9 billion acquisition of CA Technologies in 2018 and the $10.7 billion purchase of Symantec's enterprise security business in 2019 puzzled some observers. What was a semiconductor company doing buying mainframe software? The answer, in the Hock Tan framework, was simple: CA Technologies and Symantec's enterprise division both had large, sticky, deeply embedded customer bases that generated reliable recurring revenue. They were franchise businesses disguised as legacy software companies. Tan saw that immediately.

The crown jewel came in 2023: the $69 billion acquisition of VMware, the cloud computing infrastructure giant whose software underpins the server virtualization layers of thousands of enterprises worldwide. It was the largest software deal in history, and it transformed Broadcom overnight into a company with one foot in silicon and one foot in software — a hybrid structure uniquely suited to the AI era.

The AI Dividend

When the artificial intelligence boom erupted in earnest following the launch of ChatGPT in late 2022, Broadcom was not a bystander. It was a load-bearing wall.

The first pillar of Broadcom's AI advantage is custom silicon. As hyperscalers like Google, Meta, Microsoft, and Amazon began pouring hundreds of billions of dollars into AI infrastructure, they also began to recognize that Nvidia's general-purpose GPUs, while extraordinary, were not always the most efficient solution for specific workloads. Custom chips — designed precisely for a given task — could do certain jobs faster and at lower energy cost. Broadcom, with decades of experience designing Application-Specific Integrated Circuits, was one of very few companies in the world capable of producing these custom accelerators at scale. Google's Tensor Processing Unit, the chip that powers much of the company's AI infrastructure, has been designed and refined in partnership with Broadcom. More recently, OpenAI and Anthropic have joined the roster of companies working with Broadcom on custom silicon.

The second pillar is networking. As AI training clusters have scaled from hundreds to hundreds of thousands of interconnected chips, the networking fabric that ties those chips together has become as important as the chips themselves. Broadcom dominates the Ethernet switching market that serves as that fabric. Its Tomahawk 6 switching chip has become a cornerstone of the industry's push toward Ultra Ethernet, a standard designed to bring the scale and openness of Ethernet networking to AI training workloads that were previously the exclusive domain of Nvidia's proprietary InfiniBand network. In a data center full of Nvidia GPUs, there is a good chance that Broadcom is running the highways between them.

The third pillar is VMware. In the enterprise, AI does not live in a vacuum. It runs on virtualized infrastructure, in hybrid cloud environments, inside the data center architectures that VMware has defined for the better part of two decades. As companies race to deploy AI across their operations, they are doing so on the VMware stack that Broadcom now controls. The software business has generated controversy — Broadcom's pricing restructuring after the acquisition angered many customers — but it has also generated extraordinary cash flows that fund continued investment in the semiconductor side of the business.

Triumphant Numbers

The financial results of Tan's approach are difficult to argue with. Broadcom's fiscal year 2025 revenue reached $63.9 billion, an increase of more than 20 percent year over year, with AI-related revenue alone surging 106 percent to $8.4 billion in a single quarter. Net income came in at $23.1 billion — a net margin of roughly 36 percent, among the highest in the semiconductor industry and a direct testament to the margin discipline that Tan has enforced across every business he has ever run.

Broadcom's stock has risen more than 3,000 percent since Tan took the helm, outpacing nearly every technology company of comparable size. Its market capitalization has at times exceeded $1 trillion, placing it in rarefied company alongside Apple, Nvidia, Microsoft, and Alphabet.

Looking ahead, Tan has given investors a remarkable target: more than $100 billion in AI chip revenue by 2027. Given the trajectory of AI infrastructure investment and Broadcom's embedded position across custom silicon, networking, and enterprise software, that figure may prove conservative rather than ambitious.

Quiet Architect

Hock Tan will almost certainly never be as famous as Jensen Huang. He lacks the leather jacket and the theatrical flair. He does not narrate the future of humanity from a stadium stage. What he does instead is something arguably harder: he finds enduring value where others see complexity, acquires it at a fair price, runs it with iron discipline, and waits for the world to catch up with why it matters.

The AI boom did not create Broadcom's importance. It revealed it. The networking chips, the custom accelerators, the enterprise software infrastructure — all of it was already there, accumulated over two decades of deliberate, unsentimental deal-making. When the demand for AI compute exploded, Broadcom did not have to pivot. It simply had to answer the phone.