Alfidi Capital Blog: 2018

I could not close out my 2017 adventures without attending the REUSE 2017 Semiconductor IP conference in Santa Clara. I am a regular down at this convention center and nothing big in tech escapes my notice. I also rarely escape the tech sector's notice, which is why techies invite me to events. Gaze upon my badge selfie below so we can get to work.

Alfidi Capital at REUSE 2017.

The first keynote on IP theft should have opened the eyes of anyone in Silicon Valley who still thinks that authoritarian countries can still be friends of the United States. Just look at the People's Republic of China, for crying out loud, and see how their Communist government causes us problems. The US Commission on the Theft of American Intellectual Property (aka the IP Commission) reported that China has stolen up to $600B worth of IP from the US economy, including information critical to our defense industrial base. I started to see the potential for this kind of disaster years ago when I first started following the rare earth element sector. China's stated strategic desire to attract high value manufacturing to its territory is a long-term threat to democracy.

Our REUSE keynote speaker had solid recommendations for defenses against foreign IP theft, with the caveat that litigation after the fact is far more expensive than erecting cyberdefenses (like encryption) before it happens. I would add that any US company considering a high-tech joint venture with a Chinese company needs to abandon such a project immediately, before the inevitable ripoff happens.

Automobile IP will be a hot topic as automakers deploy autonomous vehicles. Get to know the Single Edge Nibble Transmission (SENT) protocol for sensors. The CAN in Automation folks hold their plugfests to see what kind of automotive tech will work as we move into self-driving cars. The proliferation of sensor tech in IoT and cars means there will be more pure-play IT possibilities for startup business models.

The Arm folks made their presence known once again. Their Arm Compute Library for developers now includes ML and AI enablement. The next step in architecture evolution is integrated encryption capability. Expect to see ambient energy harvesting for IoT devices via radio frequency (RF) or solar attachments. Arm's DesignStart program offers no-cost licensing up front for development based on Arm's core systems. I look forward to watching startups build hardware-infused encryption and energy processors onto Arm platforms. The bottom line for developers is that partner IP built on platform IP like Arm enables faster hardware product time-to-market.

I will offer a warning to US developers based on what we already know about Chinese IP theft. Many partners in some foundry support ecosystems have Chinese-sounding names, so they deserve a thorough review for any US partner concerned about IP protection or economic dependence. In case of a trade war, US companies would be cut off from Chinese supply chains, and I doubt they will have access to the WTO or other avenues for recourse in the event of IP theft.

I learned a few things from a keynote on how IP ecosystems evolve. Springer's "Security Assurance Guidance for Third-Party IP" is a start for developers who must verify their vendors' trustworthiness. Vendors must use a formalized ticketing system to track IP knowledge sharing. I get that there will be new opportunities for developers in subsystems, chiplets, and services. I also foresee new threats as inexperienced tech developers launch those startups without IP tracking or vendor vetting on day one.

I never pass up a chance to hear about anything from open sources, so of course I went to the panel on open innovation. I was very impressed with one business model empowering fabless product generation, and it may be very useful for one concept I have in development. Any FPGA circuit, programmed input/output (PIO) device, or electronic design automation (EDA) tool can now be developed from open sources. Microsoft's Project Catapult created a white paper "A Cloud-Scale Acceleration Architecture" that provides a decent road map for FPGA development. Google search results for "open source prototyping" show that open sourcing is harder and costlier for hardware than for software.

The closing keynote on semiconductor industry trends gave me some new food for thought on AI impacts. If you have not heard of those things yet, they are coming to a startup pitchfest near you after the wheels fall off this crypto ICO craze. Cheap open source development will become more prevalent now that only the largest semiconductor companies can afford to survive the long lead times between major nodes in tech shifts. Smaller design rules are part of the push against Moore's Law, and they increase the cost of conventional semicon development. Extreme ultraviolet lithography (EUV) is the next step in small measurement process development. The sector's drive for Design-Technology Co-Optimization (DTCO) pushes the simultaneous development of every step in the semicon value chain, including the foundry.

REUSE is an informal showcase for smaller companies. I got my trade show passport stamped at every expo vendor booth because I'm just that thorough. I will now offer some wrap-up wisdom for any techies considering a hardware startup. Secure your foundation IP, encrypt your data, investigate your vendors, and track your workflows if they connect outside your enterprise. Deny any potential Chinese partners access to any and all American tech, including napkin sketches, and erect impenetrable digital barriers to Chinese cyber intrusions. I would be happy to share these observations and more if REUSE offers me a speaking slot net year.

Wednesday, February 28, 2018

The Haiku of Finance for 02/28/18

Build blockchain network
Rising energy cost curve
Nonviable point

Alfidi Capital Visits IoT - AI - Blockchain Expo North America 2017

I ventured down to Silicon Valley towards the end of 2017 for something I have never experienced before. I sought the combined excitement of IoT Expo, AI Expo, and Blockchain Expo North America 2017 because everyone is talking about all the money they can make from this stuff. I have always been willing to buy the arguments for IoT and AI, but I have been skeptical about the blockchain's utility. I am more skeptical than ever after this combined conference.

It's the Alfidi Capital badge on a program booklet for AI, blockchain, and IoT in 2017.

The first speaker I encountered compared ICO tokens to 19th Century railroad investment certificates, when America was developing its frontier and banks issued multiple competing currencies. Yeah, it's the Wild West all over again in ICO land. There were plenty of stagecoach robberies back then as adventurers moved out west to seek their fortunes, and there are new stagecoach robberies happening today with each ICO. A rundown of all the buzzwords needed to understand this space, like "smart contracts" and "off-chain computing," deserves a separate blog article.

Blockchain probably cannot be valued as an asset using traditional financial metrics like DCF valuation, unless we draw an analogy with monetized API calls. It also probably cannot be valued as a currency; it has no comparable factors like GDP or current accounts in national trade. The best valuation method will have to adapt data valuation (remember, it's the new sun, not the new oil) by measuring a blockchain network's ability to process such data. I am certain that such processing ability reaches an inflection point when the energy cost of adding new hashes to the chain exceeds the marginal revenue collected from each additional data input. You heard it here first at Alfidi Capital.

I have a hard time accepting the blockchain concept of "gas," the amount of resources needed to execute a smart contract. If it has a dollar value attached, it can form a basis for valuation. I believe a network that calculates its remaining gas amount will eventually see it turn negative once they compute the energy cost needed to build smart contracts past some critical point.

I initially thought the Technology Services Industry Association (TSIA) is emerging as the governing body for the more integrated parts of the IoT / AI / blockchain sector. Their "about" page describes a research and advisory firm, so now I have no idea whether they represent a sector or themselves. They also track the cloud sector (of course). The confusion is typical of a growth sector pushing brand new networked business models, so in fairness I don't think the microcomputer revolution of the 1960s-70s was any more organized.

The venture capitalists' perspective at this conference offered further confusion. They claimed to invest in quality tech, but never specified any criteria. How would they know if a blockchain has quality if they have never compared its processing logs to those of a conventional cloud PaaS provider? The VCs' claim that ICOs will push more venture investment into early-stage blockchain startups must be a symptom of FOMO (fear of missing out), because experienced VCs should know that the SEC is taking a hard look at ICOs. Risking legal trouble just to jump into an ICO shows desperation. I still cannot understand how ICO tokens can be distributed to both investors and early tech adopters without the startup's cap table turning into a mess. A messed-up cap table is a recipe for securities fraud litigation.

I had to LOL at VC investors who thought tokens enabled network effects that give a startup some competitive advantage. One investor finally made sense by admitting that tokens are not as advantageous for investors as equity (LOL, no kidding). I do not understand how an ICO evolves into a publicly traded stock so early investors have a liquidity exit. None of the investors appearing at this conference ever made that clear. All of these people are trying to change the definition of success to "funding the network" where tech adopters participate, whatever that means. It is totally unclear how token buyers will ever see an ROI. Maybe that's the whole point; maybe ICO promoters just want to abscond with money from excited people. It's time for scam alerts!

One presenter from a data company made a very interesting point about equity in data itself rather than the distributed network. If we owned the data we generate, and companies had to pay us fees or licenses to use it, it would create universal basic income (UBI). He did not mention the fact that not all data is equally valuable. Data for an ultra-high net worth (UHNW) investor would presumably be worth more than data for a low-income person, if data value is influenced by a target's projected consumer spending.

I perused the expo floor for some ray of hope that startups can make a honest business case for a blockchain. A brochure for some online VC portal claimed to tokenize startup investing, whatever that means. There's a scam alert right there. I picked up other brochures touting tokens as loyalty rewards that doubled as investments. That concept makes no sense, but scams are never intended to make sense.

The panel on AI investing made a lot more sense. Venture investors now use AI to assess startup teams based on their public domain background info, along with the tech's strength. The traditional metrics of team, tech, and traction (that means sales!) still matter, and some investor wanted to see two of those out of the three at the seed stage. One VC admitted to subjectively evaluating teams, which I have long suspected is the norm. Hey folks, just so we're clear, these VC firms that hire analysts to do busy work while they default to favoritism in their investment choices are fooling their institutional investors. They are not fooling me. Bringing AI into the process removes the subjectivity and leaves an audit trail for fund investors to inspect. My caveat is that AI used to identify winning startups may be susceptible to survivorship bias.

The "VC in AI" panel dragged on and I was disappointed to discover zero discipline or consistency to anything those so-called VCs said. The best VCs (like Vinod Khosla) have well-developed pattern recognition abilities. Their brains are biochemically different from those of most VCs. The final observation that bad product/market fit and running out of cash were still the leading causes of startup failure shows how little things have changed in startup land after all these years.

It was good to see ARM showing its Mbed cloud for IoT device connectivity, a.k.a. the open source Mbed OS. I think I've seen the LoRA Alliance before at these types of shows, bringing the wisdom of low-power wireless to one and all. I want the vendors pushing low-power devices to be more well-versed in their use cases. One presentation did not thrill me and I cannot imagine the vendor's PaaS partners would be thrilled that a sales rep doing a new product demo cannot demonstrate the economic value of a use case.

I sat on the notes I took and handouts I collected at this conference for some time, wondering what to make of the bizarre business models I encountered. The IoT and AI potential is there and people are certainly using that tech to transform enterprise computing. That's the good news, and now here's the bad news. I am now resigned to the likelihood that much of the blockchain ecosystem is a waste of computing power and investment capital. Many of the brochures I picked up touting blockchain businesses had weak use cases but were nonetheless raising money through ICOs. I stay far away from ICOs and their nonsensical tokens. The coming blowups in blockchain will be painful. Watch the headlines in 2018 for fraudsters getting sued and prosecuted for ICOs. I will keep attending these conferences, with a watchful eye out for bad ideas to expose.

Wednesday, January 31, 2018

The Haiku of Finance for 01/31/18

Spending on research
Lifeblood of innovation
Driving future growth