He sold companies to Cisco, F5, Cascade, and Ericsson. Then he turned around and started writing the checks.
SAN FRANCISCO - TAIPEI // FILED FROM THE GP DESK
Cheng Wu - General Partner, Taiwania Capital
The Lede
On a Tuesday in February 2025, Cheng Wu publishes a white paper. The subject is enterprise AI and how messy company data becomes business intelligence. The byline reads Cheng Wu, General Partner, Taiwania Capital. It does not mention that he has done this dance four times already - that he has stood on the operator's side of the table watching the same wave crest, then crash, then crest again. Networking gear in the mid-90s. Web switching in the late-90s. File virtualization in the 2000s. Mobile video in the 2010s. Now: data infrastructure for AI. The pattern is not coincidence.
Wu is the kind of investor who reads pitch decks the way a former chef reads menus. The garnish does not impress him. He is looking at how the kitchen is wired.
Born and educated in Taiwan, schooled in computer science in the American Midwest, deployed to the Boston networking corridor and then to Silicon Valley, he now spends his time bouncing between San Francisco and Taipei, putting Taiwania Capital's growth-stage money into companies that look a lot like the ones he used to run.
His resume does not have a clean line. It has four of them. Arris Networks. ArrowPoint. Acopia. Azuki. Each one founded, each one sold, each one for a price that an entire venture portfolio would happily settle for.
The number people quote, when they remember to quote it, is six billion dollars.
By the Numbers
The Exits
- to Cascade Communications
High-density internet access gear at the moment the modem rack went from optional to oxygen.
- to Cisco Systems
Web content switching, taken public and folded into Cisco at the absolute peak of the first internet wave.
- to F5 Networks
Intelligent file virtualization for enterprise storage. Founded right after he left Cisco.
- to Ericsson
Broadband wireless multimedia delivery. He called it virtualization for video.
The Story
The first time, in 1995, he was building modem aggregation gear for internet service providers nobody had heard of yet. Arris Networks went to Cascade Communications inside of a year. To anyone else this would have been the career. For Wu it was the warm-up.
The second time was the one people remember. ArrowPoint Communications opened its doors in 1997 with a bet that the web would soon need a different kind of switch - one that understood HTTP, not just packets. By 2000 it had IPO'd and was inside Cisco Systems for $5.7 billion. Wu stayed on as a Group VP for two years, learned how a $400 billion company actually moves, then walked.
The third time, in 2002, he co-founded Acopia Networks. The thesis was unglamorous: enterprise file storage had become a tangle of NAS boxes and nobody could find anything. Acopia made a switch that put a single namespace over the mess. F5 Networks bought it in 2007 for $210 million.
The fourth time, again in 2007, he was already deep into Azuki Systems. The bet there was that video was about to eat mobile networks alive and somebody would have to deliver it across every screen, every format, every carrier. Ericsson eventually agreed. Acquisition closed in February 2014.
Then he did the thing that operators rarely actually do - he kept going. He founded APL Software in 2015. He led the angel round for Appcito alongside Atlas Venture and March Capital. He took a board seat at Verana Networks. He served as strategic advisor to Ubiik, the Taiwanese IoT firm.
In August 2020, he formalized the second act and became a General Partner at Taiwania Capital. The fund - sometimes described as the venture arm of Taiwan's national innovation strategy - had a mandate to bridge Silicon Valley and Asian deep tech. They needed someone who could speak fluently in both rooms. Wu had been doing that for thirty years already.
The Timeline
The Investor
Taiwania Capital was set up to be a bridge. On one side: Taiwanese semiconductor manufacturing, contract production, and engineering depth that the rest of the world still depends on. On the other: American software, AI, and enterprise distribution. The fund deploys early- and growth-stage capital into both, often into companies that need to live on both shores at once.
Wu's portfolio of past investments and board seats reads like a who's-who of plumbing-layer tech: Verana Networks, where he is a director. Ubiik, where he serves as strategic advisor. Appcito, which he led the seed round for. Wren AI, the open-source GenBI startup Taiwania backed alongside the 2025 white paper.
The thread that runs through all of it is unglamorous, and that is the point. Wu invests in infrastructure - the things the next layer up will be built on top of. Switches, then file systems, then video delivery, now data and AI substrate. The pattern is not a portfolio. It is a worldview.
The 60-person Taiwania team operates out of Taipei City, in a building on Keelung Road, and out of San Francisco, where Wu does most of his West Coast work. Two clocks, two passports, one thesis.
If you have built four companies that survived to acquisition, you develop a nose for the unsexy detail that determines whether a startup ships. Most pitch decks describe the product. Wu reads them for the assumption hidden in slide four.
"The acceleration of IPTV is triggering the investment cycle on the telco side," he said back in 2013, when he was running Azuki. That sentence dates the moment, but the underlying instinct does not. Catch the cycle on the upswing. Sell the picks. Get out before the wave breaks.
The Receipts
The Quotes
The Outtakes
After ArrowPoint sold for $5.7B in 2000, Wu took a Group VP seat at Cisco. Most founders would have parked there or jumped to a beach. He stayed two years, watched how a giant moves at scale, then quit to found Acopia. The reason that matters: every subsequent company he built was designed by someone who had actually seen what an acquirer's roadmap looks like from the inside.
Acopia was sold to F5 in 2007. Azuki Systems was founded in 2007. The overlap was not an accident - Wu was already deep into the next thesis while the previous one was closing. Few operators do this and live to tell about it.
Before he ever printed a Taiwania business card, Wu was leading angel syndicates. Appcito's $7.5M Series A in the mid-2010s was co-led by his check alongside Atlas Venture and March Capital. The path from operator to investor was not a leap. It was a slow, quiet walk.
His business cards read San Francisco. His firm's office is on Keelung Road in Taipei. The dual posture is not optics - it reflects three decades of moving capital, talent, and ideas across the Pacific in both directions.
Indiana University awarded him the Presidents Circle Laurel Pin in 2016, recognizing benefactor status. The 2009 Career Achievement Award from the School of Informatics had already established the through-line: the kid from Hsinchu who showed up in the American Midwest for grad school never really left the alumni rolls.
Wu does not show up much in social feeds. When he wants to make a point about where enterprise AI is going, he writes a white paper and ships it with a portfolio company. The signal-to-noise ratio is closer to a research note than a thread.