Welcome to Semi-Literate, a guide to the chip industry through the lens of public policy. Feel free to forward to others you think may be interested.
BLUF: In-Q-Tel, the Central Intelligence Agency’s venture capital (VC) arm, recently published a paper on the US “national microelectronics challenge” arguing for greater public and private sector investment in the US chip ecosystem. However, a review of In-Q-Tel’s current portfolio companies reveals less than 5% are chip startups, making this call for greater chip investment ring a bit hollow. In-Q-Tel should increase investment in US chip firms and coordinate support for those firms with DOD acquisition leaders. A quick analysis finds 25+ US chip startups that align with this report’s recommendations, but lack USG funding. It also finds Chinese investors maintain an active, though reduced, presence in US microelectronics VC.
Background:
Founded in 1999, In-Q-Tel (IQT) “scouts the global market for cutting-edge technologies that have both high national security impact and the potential for commercial success.” Unlike most of the intelligence community, IQT operates in public as an independent, non-profit strategic investor for the CIA. It hires from government, venture capital, and industry. It has a blog, a podcast, and it talks publicly about its investments. It also publishes “insight and thought leadership” reports. Their Winter 2021 issue was on “The National Microelectronics Challenge.”
Summary of Paper:
The paper is short (8 pages) and worth reading in full. The first half is an overview of the industry, technology trends, security considerations, and private sector investment gaps related to microelectronics production and supply chain security.
The second half of the paper is much more interesting and novel. The authors begin by using (January 2018) PitchBook data to establish:
of the 215 U.S. microelectronics companies that received private funding during 2015-2017, only five were design software firms, eight sold novel tooling equipment, and nine were in semiconductor IP.
They go on to argue that, while proposals to boost research funding and build fabrication facilities are important for economic and national security reasons, there are important sub-sectors of the industry that are being neglected by private investors. The punchline isn’t surprising: “there is a dearth of trusted private investment in the microelectronics startups of the greatest potential national security import.”
Key Recommendations
The thing this report does better than any other USG report is emphasize the role of commercialization. Simply put, the burn rate in the semiconductor industry is so high that every chip startup needs to have a commercialization strategy. As a partial solution to reduce this industry-wide burn rate, the authors propose the USG support under-invested areas of the chip industry’s startup landscape with:
Commercialization Infrastructure: a public-private partnership that provides industry and academia access to advanced commercial equipment to replicate industry requirements, and enable feedback loops from commercialization back to research. This “microelectronics commons,” “national semiconductor center,” “Lab-to-Fab facility,” and/or “hardware sandbox” would focus on:
Semiconductor manufacturing equipment (the report calls this “tooling”)
Advanced packaging
Microelectronics Investment Fund: this fund would bridge the gap between the innovation pipeline and scaling of a commercial enterprise. The authors suggest a fund of $250 million over five years, targeting key areas such as toolsets, advanced packaging, and origin-agnostic security technologies. They highlight four firms in need of investment developing chip-related technologies:
“Direct write” of circuits (a non-photolithography approach to patterning)
Automated optical inspection systems for defect analysis of wafers
"Lego blocks" approach to chiplet assembly, accelerating custom designs
Hardware-level security processor for embedded systems
Reviewing These Recommendations: A Mixed Bag
These proposals are good. The way the recommendations are structured indicates the author’s know this industry. But these recommendations are strange for two reasons:
The recommendations are redundant: the CHIPS Act, passed as part of the January 2021 NDAA, calls for a semiconductor investment fund (section 9905) and tasks the Secretary of Commerce with establishing a PPP, the “National Semiconductor Technology Center,” to promote R&D in to semiconductor manufacturing equipment and establish a National Advanced Packaging Manufacturing Program (section 9906). No mention of this in the paper.
In-Q-Tel is designed to address of the very problem it has diagnosed: This report talks about a funding gap in a US advanced technology industry that has national security implications. Surely IQT, the US government entity tasked with commercial funding of innovative advanced technologies that have national security implications, is on the case? A look at information on IQT’s current portfolio companies using a VC investment database suggests otherwise:1
In-Q-Tel’s Current Portfolio Companies
A quick look at IQT’s list of its portfolio companies and a cross-walk of those companies with information from the investment database CrunchBase finds:
By the Numbers:
216 Portfolio Companies listed on In-Q-Tel’s website.
215 IQT portfolio companies have a CrunchBase profile.
72 IQT portfolio companies have been acquired.
69 IQT portfolio companies have one or more DOD contract/purchase order.2
IQT breaks down the companies by their own industry categories:
Analytics 44 companies
IT Platforms 30 companies
Industry 4.0 29 companies
Intelligent Connectivity23 companies
AI & Machine Learning23 companies
Trusted Infrastructure22 companies
Autonomous Systems 18 companies
Biotechnology 13 companies
Digital Intelligence 13 companies
In-Q-Tel most frequently co-invests with:
Google Ventures 15 deals
Intel Capital 12 deals
Sequoia Capital 10 deals
Lux Capital 9 deals
Goldman Sachs 9 deals
In-Q-Tel Investment in Microelectronics
The industry tags on the IQT website don’t include a “microelectronics” breakout, so CrunchBase industry tags were used to establish their companies’ specific scope of work. There are 9 companies in IQT’s portfolio doing microelectronics-related work:
Name // Sub-sector // Status
D-Wave Systems // Quantum // Privately held
GainSpan // Wi-fi // acquired by Telit (UK)
Lime Microsystems // Wireless broadband // Privately held
Mythic // Artificial intelligence // Privately held
Newlans // Radio frequency // Privately held
Pixim // Video semiconductors // acquired by Sony (Japan)
Redlen Technologies // Radiation detection // Privately held
SiOnyx // Photonics // Privately held
WiSpry // Radio frequency // acquired by AAC Technologies (China)
What Does This All Tell Us?
IQT’s recommendations and this follow-on analysis reveal (1) subject matter expertise, (2) a lack of awareness of current US government policy and initiatives concerning microelectronics, and (3) under-investment in the very industry they are claiming needs more investment on national security grounds.
The authors know the industry. In addition to this deep subject matter expertise, IQT is investing with some of the best venture capital firms. This is an uncommonly high level of expertise for a USG-affiliated entity. They have the skills and ability (if not the capital) to solve the problem they’ve identified.
Analytical gaps: (A) the Commercialization Infrastructure and Investment Fund ideas have already been passed as part of the CHIPS Act, making the recommendations redundant. (B) The DOD Trusted Capital program is designed to address the very problem the authors’ diagnose: facilitating trusted private investment into critical and emerging technologies. No mention of the program in the paper. (C) The report doesn’t engage with many ideas in the 25+ substantive reports published about this industry in the last 5 years. For example, arecent study advocates for the same kind of chip PPP the authors propose.
Hardware is hard. Obviously the microelectronics problems facing the United States are not IQT’s alone to solve, but surely they could devote more than 5% of their portfolio to the industry. 50%+ of IQT’s portfolio is non-hardware focused. Microelectronics hardware is notoriously difficult and costly. One recent back-of-the-envelope analysis found the most boot-strapped chip startup would still burn through $1.5 million in 9 months to get a minimum viable product. To get to first revenue would cost $20 million and then scaling that is likely to take another $20 million. The “valley of death” between innovation and commercialization is real and patient capital (like IQT’s) is hard to come by for a hardware startup.
How IQT Can Support US Microelectronics Innovation
IQT has correctly diagnosed the problem, now they need to act on their prescription. (1) IQT already does emerging technology analysis for DHS, they need to start sharing their market research analysis with DOD Service Acquisition Executives who can direct funds towards portfolio companies. (2) The IQT portfolio should be over-weight in microelectronics startups given how fundamental chips are to each of the industries they invest in (Trusted Infrastructure, Artificial Intelligence, Biotech etc) and how under-invested microelectronics is by private VC. To expand on both points:
DOD should buy more from IQT portfolio companies: Only 69 of IQT’s 216 portfolio companies have a recent DOD contract/purchase order. Add to that the 72 IQT portfolio companies that have been acquired (at that point they’re no longer “startups”) and remove the sum from the pool. That leaves 74 startups in which IQT has invested that lack any commercial relationship with the largest source of national security funding. The IQT stamp of approval seems like it should be good enough for AFWERX, DIU, AFC, NavalX, and SOFWERX to evaluate each of these 74 companies in light of their requirements.
IQT should create and share an open-source list of microelectronics startups: IQT should share its market research with USG entities who have funds to spend, who have microelectronics equities, and who don’t have access to investment data/startup expertise. Why not conduct a simple open-source exercise to expand IQT portfolio lead generation and share the results with DHS/DOD?: look up chip startups that have closed a funding round since 2015, see which ones have existing USG relationships, and vet for a scope of work that fits within the report’s recommendations?
Proposed Analysis of US Chip Startup Ecosystem and Eligible Firms3
The findings of this proposed analysis are striking:
300+ US chip startups have taken funding since 2015 (this is a very conservative number of deals but its derived from free data, so you get what you pay for).
85 startups took a loan from the Small Business Administration/Paycheck Protection Program in 2020 (indicative of some level of financial distress).
51 startups have at least 1 DOD contract/purchase order in the last 10 years.
30 startups have taken funding from the National Science Foundation, Department of Energy, and/or NASA.
~20 startups have had a funding round that included one or more Chinese investor. 10% of US microelectronics VC in 2020 had one Chinese investor.
~15 startups are working on advanced materials, ~5 on Electronic Design Automation, and ~5 working on advanced packaging solutions. Most don’t have USG contracts. Seems like a target-rich environment for lead generation.
2 companies in this exercise have taken investment from IQT. Theres lots of room for IQT to expand in the microelectronics space.
The analysis could easily be extended depending on ones priorities:
Look at when companies closed their last funding round (find out which companies are going to need funds soon);
Filter for companies that have already taken investment from Intel Capital, Applied Ventures, Qualcomm Ventures etc. and other leading industry investment funds. Industry knows better than anyone which high-risk/high-reward ideas are worth investing in. Follow their lead when making investment or procurement decisions.
Conclusion
There are plenty of gaps in this proposed analysis but the whole thing took a few hours and used (messy) free data. The subject matter expertise demonstrated by the authors of this paper indicates they could do much better analysis and their findings could inform other parts of government while helping US chip startups. IQT is deliberately not a policy-making organization, but it can at least inform policy decisions that respond to this report’s recommendations with its clear expertise.
Note: The views here are my own and drawn solely from the documents I cite.
1
CrunchBase data has its problems and these numbers are probably off. But PitchBook doesn’t offer free 7-day trials, and CB data is better than nothing. The analysis here was: scrape IQT website for company names, upload list to CB, pull relevant CB data, analyze.
2
This was established via a search on USASpending.
3
The analysis here was: scrape CrunchBase for US semiconductor VC 2015-present (though CB probably has only 2/3rds of all chip VC since 2015), cross-walk company names on USASpending to determine their funding from USG entities, look at each US startups’ top 5 investors and cross-check investor HQ location for Chinese HQs. There are no doubt some US-based but Chinese-funded VCs (ex. Oriza Ventures) whose investments are missing too.
