The first thing to understand about Babu Mandava's career is that he keeps choosing the part of the technology stack most people never see. A modem chip disappears inside a box. A baseband processor turns radio waves into a usable connection without asking for applause. A glass substrate sits beneath a cluster of components, carrying signals, heat and power while the product above it gets the name.
The second thing is that his timing has repeatedly put him inside an awkward transition. He has worked where an established system is reaching its limit but the successor still looks conditional: digital signal processing before broadband was commonplace, WiMAX before the mobile market settled around LTE, and advanced packaging while the chip industry learns that shrinking a single transistor cannot solve every computing problem.
Mandava's documented career begins in 1983, the year he earned a Master of Technology in electrical engineering and computer science from the Indian Institute of Technology Kanpur. He spent the next three years at Semiconductor Complex in India, designing for telecom and defense applications. The job placed him close to both the logic of a circuit and the constraints of making one useful.
A career written in bottlenecks
At Zilog, Mandava worked on microprocessors and microcontrollers, including an early single-chip combination of a microcontroller and digital signal processor. At Cirrus Logic, he became an IC design manager and architected DSPs for wireless and V.34 modem products, along with an ARM-based modem controller. The product names now read like exhibits from the pre-broadband internet. The underlying problem remains current: move more information through limited power, silicon and bandwidth.
In 1994, Mandava co-founded Centillium Communications. His engineering work there included the architecture of a SIMD/VLIW digital signal processor used to enable multichannel ADSL modems and high-density voice-over-IP chips. Centillium went public in 2000 and later reached a peak market capitalization of roughly $3 billion. It was the first proof in his career that a component hidden deep in the system could become the center of a sizable public business.
Yet the detail worth stealing is not the market-cap number. It is the movement from architect to operator. Semiconductor startups cannot keep product, process and customers in separate rooms. Someone must understand enough of each to decide which compromises preserve the company. At Centillium, Mandava took on engineering and general-management roles of increasing responsibility. The circuit designer was learning to manage the machine around the circuit.
The standard that moved underneath him
Mandava co-founded Beceem Communications in 2003 and served as chief executive. The company built chips for mobile WiMAX, contributed to the IEEE 802.16e effort and helped define the first Mobile WiMAX profile. In 2009, Beceem held a reported 65 percent of the market for mobile WiMAX chips.
The number sounds commanding until you remember what the market was doing. Wireless standards do not win on engineering alone. They require carriers, devices, spectrum, capital and timing to line up. LTE was gathering force. Beceem responded with a platform designed to support both LTE and WiMAX rather than treating one standards bet as an identity.
In October 2010, Broadcom announced an agreement to acquire Beceem for approximately $316 million, net of cash assumed. The buyer's logic was revealing: Beceem offered a team with experience across 4G systems and a faster route to multimode LTE and WiMAX products. Mandava then served as a Broadcom vice president of engineering, leading the team responsible for the company's first LTE multimode baseband chipset.
A company can be attached to a standard; an engineering capability can travel. Beceem's market evolved, but its knowledge of radio, baseband design and mobile broadband retained strategic value. Mandava's 2010 talk about building the company carried the title “Toughing it Out to Glory.” The title is less polished than an acquisition announcement and more useful. Infrastructure companies spend a long time being early, expensive and difficult before their timing looks good in retrospect.
From designing chips to organizing their ingredients
In January 2017, Mandava joined Cadence Design Systems as senior vice president of research and development for the IP Group, later described as its senior vice president and general manager. Cadence sells the design tools and reusable intellectual property that chipmakers use to assemble complex systems. In his first year, he refocused the IP strategy, secured key customer wins and built a stronger leadership team. He stayed through 2021.
The move widened the lens. At Beceem, the object was a 4G product. At Cadence, the object was a portfolio of building blocks that other companies would turn into products. That makes his next job feel less like a pivot than a continuation.
3D Glass Solutions, founded by Jeb Flemming in Albuquerque, works on glass-ceramic materials and processes for electronic packages. Its technology uses semiconductor-style patterning and etching to form three-dimensional microstructures, passive RF components and through-glass connections. Those structures can help multiple dies and components communicate inside one package. The company addresses RF electronics, photonics, sensors and high-performance computing.
Glass sounds fragile because windows are our reference point. In a package, different properties matter. 3DGS emphasizes low transmission loss at high frequencies, thermal stability and the ability to create dense interconnects in a dimensionally stable material. The substrate can host passive devices and connect components vertically, reducing the distance a signal must travel.
Mandava's work is not to invent glass. It is to turn a specialized process into a dependable manufacturing system. That distinction appears throughout his public comments. When 3DGS raised $30 million in Series C funding in April 2023, he said the capital would expand production wafer capacity and accelerate the product roadmap for RF and data-center heterogeneous integration. The investor group included Walden Catalyst Ventures, Intel Capital, Lockheed Martin Ventures and Applied Ventures.
A factory is an argument made concrete
The current chapter stretches between Albuquerque and Bhubaneswar. In April 2026, Indian officials and 3DGS broke ground on an advanced packaging and embedded-glass-substrate facility in Odisha. The project carries a planned investment of ₹1,943.53 crore, with commercial production expected by August 2028 and full-scale volume targeted for August 2030. Planned applications include data centers, AI, 5G and 6G, automotive radar, defense electronics, aerospace and photonics.
The capacity targets make the ambition tangible: roughly 70,000 glass-panel substrates and 50 million assembled units a year, plus 3D heterogeneous-integration modules. At the groundbreaking, Mandava described large substrates with high dimensional stability, dense interconnects and low loss. “This is just the beginning,” he said.
A month later, Odisha, Intel and 3DGS signed a separate memorandum of understanding to explore an advanced glass-core packaging-substrate facility in the state. The proposal remains conditional on approvals, government support and other business and regulatory requirements. Its importance lies in the relationship it sketches: Intel contributing process expertise, 3DGS bringing glass technology, and Odisha trying to assemble the surrounding manufacturing ecosystem.
Mandava has known Intel chief executive Lip-Bu Tan since 1997. In a public post, he wrote that they had worked across five companies where Tan was an investor, director or, at Cadence, chief executive. Tan joined the 3DGS board after the 2023 financing. Long relationships matter in semiconductors because the technical cycles are slow and the capital demands arrive before certainty does. Reputation becomes a form of working capital.
What persists when the market changes
Former colleagues have described Mandava as creative, intellectually energetic and focused on execution. One remembered a leader who pushed a work-hard, play-hard culture. Those impressions fit the visible pattern without turning it into mythology. He has stayed close to technical detail while repeatedly accepting responsibility for financing, teams, roadmaps and customers.
The arc from a DSP to a glass package is not as wide as it first appears. Both are answers to coordination problems. A processor coordinates instructions. A modem coordinates a device with a network. A package coordinates multiple components with one another. Each creates more useful performance from pieces that would be limited alone.
There is a practical lesson in that consistency. A durable career need not follow one product. It can follow one constraint. Mandava's constraint is the movement and integration of information under physical limits. The tools changed from modem controllers to baseband chipsets to substrates. His working question did not.
The Odisha factories will take years to answer whether 3DGS can scale its process as planned. That long calendar is the point. In deep technology, the announcement is not the achievement; it is a public starting line. The material must perform, the process must yield, the package must qualify and the customer must reorder. After four decades, Mandava has chosen another layer where all four have to meet.