Miku Yamada - MIP 23’
- BA Law and Political Science Mikk Raud - MIP 22’ - MA Economics & Mgmt - BA Law and Politics David Sprague - MIP 22’ - BS Systems Engineering Member #2 - BS Mathematics 23’ Abeer Dahiya - BA Economics - BS Mathematical & Computational Science Final Problem Statement How should the US Government augment the domestic semiconductor workforce through education and innovation initiatives to increase its semiconductor sector competitiveness? Original Problem Statement U.S. semiconductor procurement is too heavily dependent on TSMC, which creates a substantial vulnerability in American telecommunications, automotive, military and healthcare supply chains, especially in the event a PRC invasion of Taiwan, or other kinetic disruptions in the Indo-Pacific. TOTAL INTERVIEWS: 37 Team ShortCircuit Youngjun Kwak - MS Management Science & Engineering - BSFS International Political Economy
Getting out of the building...
Academia Think tanks Investors Industry Government
From chip shortage to a
lot more Different chips, different issues Different players, different priorities We realized it’s not just about manufacturing
Our trajectory Week 1 Manufacturing?
Week 3 Not Manufacturing? Week 4 R&D? Week 5 R&D? Week 6 Education? Week 7 Education!
First steps... Week 1 Sky
is falling - PLA shows of force around Taiwan demonstrate a potential threat to U.S.’s semiconductor supply chain Week 2 Focus on one aspect of semiconductor production, and public-private partnership solutions “If you can’t trust your manufacturer, you can’t trust your chips” (Stanford EE) Week 3 Kinetic disruptions are not the issue...it’s a long- burn problem “China’s not going to disrupt flow from TSMC’s fabs...it depends on them” (Think Tank)
...Moments of key discovery Week
4 US stronghold in design won’t last forever “The US is like a polar bear sitting on a floating piece of ice that is constantly melting.” (Think Tank) Week 4 Paradigm shift is coming - in packaging and beyond “The new industry will be about system partitioning, as opposed to single chips. This paradigm shift will give opportunities for new players.” (Stanford EE) Week 5 Hardware innovation is key for long-term success “It’s a major problem that some suggest there’s no need to understand how technology works in the deepest levels.” (Stanford EE) Week 6 Pivot from manufacturing to R&D and education “Government’s money shouldn’t just go to TSMC or Intel. Cash should go to advanced R&D and students. ” (Former EDA Executive) Week 6 But - manufacturing is still critical in the long term “Nothing is more important than bringing cutting edge fabs to the US. We need to colocate manufacturing with R&D.” (Manufacturer) Week 7 Start from K-12 “We need to start way earlier in the education pipeline, addressing inspiring students and getting them curious. Just like Einstein got hooked by seeing a compass.” (Think Tank)
Key stakeholders - risks and
priorities Lack of sufficient research funding Loss of high quality researchers Academia Better research funding & incentives to work at US universities & research orgs Lack of tax incentives Strict immigration policies Industry Talent & workforce development & retention Sparse deal flow High capex, low initial RoE, high B2E/strong IP Investors “Priming the pump”, believable success of VC model to hardware Heavy manufacturing dependence on East Asia China’s advances US Govt Build fabs & eventually build up manufacturing ecosystem Hardware coursework less accessible Less employment opportunities vs. software Students Develop hardware curricula encouraging entrepreneurship & innovation Risks are multifaceted & shared by many stakeholders Challenges are interconnected Education is the missing piece
“I worked on a hardware
startup and I talked to a few VCs, as soon as they were hearing hardware, they would lose interest.” - Student “China will very soon have their own design and EDA capability.” - Industry “New ways of doing things won’t come from Intel or the likes, they start from universities.” - Academia “All high-tech requires skills that are not grown overnight. If you lose those skills, you also lose skills to innovate the next big thing.” - Think tank Why education & innovation? Potential for new paradigm shift Eroding US strengths Essential to maintain advantage Overlooked & underfunded
Investment Strategy for Education and
Innovation (Using NASA Case Study) KEY TAKEAWAY: Education should receive a quarter of every billion invested in the semiconductor industry
Our Solution: Incentivize Hardware Education
and Innovation K12-STEM initiatives to spur early interest Improving undergraduate access to industry Underwrite appropriate financial risk Design courses to establish link between hardware & software ‘Fab in the box’ for accessible innovation tools High school hackathons and competitions Leverage venture equity for hardware acceleration Modernize hardware curricula, across the board Launch fabternships with TSMC, Intel etc. Courses linking market insights with hardware innovation and ventures Expert-led fund, using public-private money to generate flow Experimental lab classes, like e-waste teardowns
Validating our solution Academia Industry
Investors YES But we need interesting classes to get hooked by hardware Students US Govt YES We need balanced curricula that target a diverse array of students YES We need a greater role for experts in the VC process; K-12 education will bring young talent YES We need co- location of industry and education to unlock EoS PENDING However, the principal author of the CHIPS Act supports our discovery
Next steps ● Exploring the
opportunity to work with the Defense Investor Network and MITRE ● Considering continuation of the project at H4D during the spring quarter ● In continuing talks with DARPA, NVIDIA, EU Commission, Synopsys, CSET at Georgetown
Recommended Policy Component BEAR CASE
(We overestimate US capabilities and underestimate adversaries; worst-case outcomes) Educational overhaul package (redesigned EE curricula, table-top fab, institutional funding etc.) ● Fail to spark sufficient interest in hardware/EE ● Expansion of resources available to undergrads at a selection of universities following our top-down strategy Investor overhaul package (establish expert-led fund, underwrite risk, tax breaks) ● Mid-term realignment toward PRC capital in non-DefAdjSecs ● Political backlash against PE tax breaks in the future ● Lack of deal flow/failures disincentivizes future growth Industrial overhaul package (priming the pump via supply-chain certification, fabrication investment, H-1B/EB-1 visa expansion) ● Supply chain certification creates cost overruns for govt, industry fails to match existing efficiency/productivity standards ● We’re still importing technical expertise by 2031 The Bottom Line: Doing Something Badly Might Be Better Than Nothing At All (at least in education and R&D) Appendix - Likely challenges to our solution
● Premise: Select a pool
of experts from academia and industry, selected on the basis of pure expertise on fundamentals of semiconductor research. ● Capitalize them with joint-public and private investment, with an open mandate to invest in ideas they are party to, either through their students or research. ● Condition: this funding must go toward making the product of that research into either a company or a deployable solution for a current problem implemented by another company. ● Recommended initial fund size: US$3bn (2.5bn govt., 0.5bn private equity) Core Problem Hypothesis: Investment professionals in the venture capital space have poor knowledge of fundamental science and consequently utilize ill-suited valuation techniques, vastly undervaluing the potential of modern hardware solutions in the semiconductor space. Another Big Problem: Experts in this space are confined largely to academia or consulting roles for large corporations, and do not have an incentive or role to play in the venture space. We want to connect our expert minds to capital, to create a model for leveraging existing expertise to generate returns in what we strongly believe is an under-invested in industry. Appendix - Whiz Kid Capital, Inc.
Appendix - list of interviewees
Academia 1 John Hennessy Professor of Electrical Engineering and Computer Science, former Stanford President, currently Chairman of Alphabet, Inc. 2 Subhasish Mitra Professor of Electrical Engineering and of Computer Science at Stanford University and former Principal Engineer, Intel Corporation 3 Jennifer Dionne Senior Associate Vice Provost for Research Platforms/Shared Facilities, Professor of MatSci 4 Priyanka Raina Professor of EE at Stanford 5 Jim Plummer Professor of EE and MatSci, Former Dean of Engineering, Boards of Intel and Cadence 6 Byung Hoon Lee Professor of EE at POSTECH 7 Young Ja Bae Professor of Diplomacy at Konkuk University, Korea 8 Tom Lee Professor of EE at Stanford, Former Microelectronics Director at DARPA 9 Tadahiro Kuroda Professor, Director, Graduate School of Engineering, The University of Tokyo 10 Kyle Squires Dean, ASU Fulton Schools of Engineering
Appendix - list of interviewees
11 John Shen Professor of EE at Carnegie Mellon University Industry 1 Antun Domic Former CTO of Synopsys 2 Jeff Rittener Chief Trade Officer at Intel, former Chief Government Affairs Officer 3 Peter Cleveland Vice President of Global Policy, TSMC 4 Rupert Hammond-Chambers President, U.S.-Taiwan Business Council 5 Edlyn Levine Chief Technologist at MITRE Accelerator 6 Jimmy Goodrich Vice President of Global Policy at SIA 7 Richard Lin Head of Semiconductor Group, Egon Zehnder
Appendix - list of interviewees
8 Nerissa Draeger Director of Global University Engagements at Lam Research 9 Bob Colwell Former Chief Architect at Intel, Former Director of Microsystems at DARPA 10 Gene Irisari Vice President for Public Affairs and Head of Semiconductor Policy at Samsung 11 Matt Johnson CEO GQ Quantum 12 Tim Donoghue Senior Program Manager at BMNT 13 Claire Sanderson Senior Director, Global Government Affairs at TSMC Investors 1 Heather Richman Founder, Defense Investor Network 2 Philip Clark Chief of Staff at 8VC 3 Anita Biddappa Operating Partner at BVA
Appendix - list of interviewees
4 Eileen Tanghal Managing Director at In-Q-Tel 5 John Hurley Partner at Cavalry AM, GSB lecturer, former US Army Government 1 Sahar Hafeez Senior Advisor, Office of Under Secretary, Bureau of Industry and Security, Department of Commerce 2 Felipe Palacios Sureda Technology, Security, FDI Screening at EU Commission Think tanks / Other 1 Matt Turpin Visiting fellow Hoover Institution 2 Glenn Tiffert Research fellow at Hoover Institution
Appendix - list of interviewees
3 Eric Sayers Nonresident Fellow at AEI 4 Will Hunt Researcher at CSET 5 Oriana Skylar Mastro Center Fellow at Stanford FSI
And so what we realized was that it’s not just about a chip shortage now. Its about a whole lot more. There’s different chips for different applications each with their own issues. And moreover there’s dozens of players in the game each with their own list of priorities. It’s a very messy picture.
And so our trajectory looked a bit like this. We began thinking, we have a chip shortage and are worried about kinetic disruptions, clearly manufacturing should be our focus. And so as the weeks went on we moved first from manufacturing to realizing maybe manufacturing isn’t the issue. Then R&D. But then realizing, hey maybe R&D isn’t the root, and eventually, hey, its education stupid. And so I’ll pass it over to Youngjun to talk this journey in a little more detail.
One of the key elements of arriving in the education direction was juxtaposing the risks and priorities we heard throughout our interviews that initially seemed to be in contrast with each other.
Students identified less accessible coursework and employment opportunities as compared to software as the key issue. Academia emphasised the difficulties of securing funding and the decreasing number of good researchers. Investors, are primarily worried about low RoI and a much higher risk than software. The industry desires more tax incentives and less strict immigration policies to secure adequate workforce. Finally, the government’s key concern is chip shortage due to manufacturing dependence, amid China’s continuing advances across the whole semiconductor value chain.
However, after discussions, we found that these competing narratives had shared elements. Although the risks are multifaceted, they are shared by many stakeholders and the challenges often overlap.
Fundamentally, when analyzing the identified priorities of each stakeholder, we found that education is the missing piece that may be best suited to address all the different stakeholders and their priorities.
Based on our interviews, we identified further evidence that education requires more attention. We heard the concerns of eroding US strengths from the industry, as well as that investors are not interested in hardware from students. At the same time, think tanks and academia suggested that we are at the verge of a new paradigm shift, which requires maintaining our advantage over competitors, and assumes continuous progress in research and education.
Dave and Abeer will explain how our recommended solution addresses these concerns.
So with this new focus on education and innovation, we began to discover a long list of novel and exciting ideas from our stakeholders that targeted students in the future semiconductor workforce.
To synthesize these ideas, we researched 13 years of Space Race era data from NASA and the Aerospace Research Center that quantified best-practice R&D apportionments among academia, industry, not-for-profits, and other capital sources.
The aggregate apportionment in these data is what you see represented here – all of these actors have a stake in education of the future workforce.
Indirectly, industry subsidies can incentivize hands-on learning in actual fabs in what we’ll explore later as the “fab-ternship” model – stimulating education through practical learning.
Neutral non-for-profit organizations can provide performance feedback and pioneer K-12 learning initiatives targeting American youth.
Academic and industry expert venture capital initiatives can stimulate deployable solutions across the many inputs of the semiconductor production cycle.
A significant takeaway from this historical case study analysis, however, was that direct investment in education is crucial: a quarter of any apportionment should go to the educational domain directly – in ways that my colleague Abeer will explore further.
As you can see, we’ve developed a four pronged approach to attack what we believe are the most under-invested in, and consequently most impactful vectors in this problem set:
K-12 education in electronics, something as simple as bringing in an iPhone into class to open it up and see what’s inside
Improving undergraduate access to industry is particularly important for a sector that’s seen as inaccessible without a PhD
Modernizing hardware curricula, which ironically some of the biggest advocates were Stanford EE professors themselves,
and private equity investment, particularly in hardware acceleration, which we’ve identified as a major growth area that is easily accessible to a new-entry player given intellectual property restrictions in the space.
Thanks INCENTIVIZE HARDWARE EDUCATION AND INNOVATION
Given the complexity in the stakeholder matrices, we’re quite happy with the responses we’ve received so far- they show us we’re hitting the right notes.
We do want to emphasize the importance of sharing information- which you all are part of now. Thank you for hearing us throughout our journey, and for bringing America’s chips back on the table.