Sunday, February 4, 2018

Atomera and Moore's Law

Tiny, unheard of Atomera (ATOM) could be Act II of a really big high tech show. What was Act I? It was erbium doping, which revolutionized fiber optic high tech in the '90s. The growth of the network was hampered by lack of a clean way to amplify optical signals. You had to convert light speed down to a crawl at electron speed to amplify, then convert back to light. Then the erbium doped fiber amplifier or EDFA was born where glass fiber was doped with erbium allowing pure optical amplification. The rest is history.

Now high tech is up against another constraint where a similar invention is needed. When Intel put forth Moore's Law, it seemed it would go on forever with transistors on a chip doubling every 18 months bringing down cost while improving performance. It's debated, but 50 years later, it may be coming up against physical boundaries that will change the industry.

As was noted recently in Barron's, Moore’s Law is “no longer a law in terms of the time frame” of improvement of chips, according to Scott Bibaud. “In 2012, the cost per transistor for the first time did not go down,” he observes, an ominous sign for the industry. Bibaud has served as Senior Vice President and General Manager of Altera’s Communications and Broadcast Division (later bought by Intel) and was chief of Mobile Platforms Group at Broadcom. In 2015, he took over as CEO at Atomera.

With Moore's Law in control, chip makers eagerly retooled the next generation smaller node, currently going from 20 nanometer to 14 nanometer. But with the Internet of Things (IOT) the wave of the future, they are not so eager. According to Bibaud:
“People are saying 28-nanometer will be useful for many years ... And 40-nanometer chips are big in the world of automotive chips and for the Internet of Things. We’ve even heard that the 130-nanometer node has the most new design starts of any node today for things like analog and sensor and power applications.”
So the new IOT chips place less demand on Moore's Law continuance as does Nvidia's approach of parallel processing in their GPU. When video and gaming took over a lot of our computer viewing, Nvidia invented, in 1999, a totally different processor (Graphics Processing Unit) that didn't need high transistor density because it dumped its workload into many hundreds of parallel paths that together put all the color blobs and motion on our screens. Since then, however, they have discovered that these processors are very good at concurrent number crunching, and they are teaching them to do a lot of what the old school, Moore's Law dependent chips were doing. For more on this, see my recent Forbes article "Why Moore's Law Now Favors Nvidia Over Intel."

But GPUs and IOT will never replace the ever advancing need for higher semi efficiency. The chip industry desperately needs some kind of big efficiency invention right now, like the fiber networks received in the '90s with the EDFA, giving ramped up performance on existing node designs, calming the costly panic into the next node. Atomera is poised to hand them just that. And they're doing it with some more doping of silica - oxygen this time instead of erbium. And they're doing it with the very same PhD that invented the erbium doping - Dr. Robert Mears, the founder of Atomera. In 2001, it was private and known as Mears Technologies.

In the 2016 annual report (when they went public) they open with:
"There is no manufacturing process as complicated as making a semiconductor chip ... In January we changed the name of our company to more accurately represent the advanced material science we are providing to the industry in an era when semiconductor advances are increasingly happening at the atomic level."
Mears is now the Chief Technical Officer of Atomera. He has an array of patents on his invention called MST (Mears Silicon Technology) that dopes a transistor with oxygen atoms in a way that dramatically improves electron flow. This allows several benefits like smartphones with 50% better battery use, but more importantly it provides "more than Moore" tooling and cost benefits to chip makers, per their website.

The strong link between Robert Mears' Act I and II is evident in Wikipedia's box summary for him. It simply reads "Born: England    Occupation: Physicist and engineer   Known for: Invented EDFA, founded Atomera   Notable work: EDFA, Mears Silicon Technology." If the oxygen doping does what the erbium doping did, ATOM is in for some big things.

I called Atomera "unheard of" but that's not entirely true. There is no analyst coverage yet, but there was the article in Barron's August 17, 2017 "Atomera Hopes To Make Money Solving The Breakdown Of Moore's Law" and a nice article at Seeking Alpha February 1, 2018 focusing on the Internet Of Things, the web connected device revolution now engulfing us.

Then there is the January 30, 2018 press release "Atomera Continues To Grow Customer Pipeline" that seems to have brought some attention and activated the stagnant stock, sending it up about 50% the following week in the face of a sharp market selloff. I have been long the stock since December at $4 and feel it is a growth item you hang onto until the thesis is altered.

This is a risky, pre-revenue, one trick pony with heavily patented oxygen doping being the only product offered. But the market may now be grappling with the possibility of a $67 million market cap company engaging some $3.5 billion of addressable market.

In the Barron's article of late 2017, Bibaud said Atomera is “very well capitalized,” with “almost three years of burn left” in terms of cash on hand. If you look at cash and equivalents at end of the first public year, 2016, it was $26.7 million. By end of 2017, it was $17.4 million - a burn rate of $9.3 million per annum. So that suggests something more like two years of burn left before a dilution may be called on. But they seem poised to either start signing lucrative licence agreements (or start getting rejected) well before two years have elapsed. And they are not in any kind of debt trouble reporting debt and lease level at zero. In the latest quarter reported 5/3/18 they report:


  • Grew the number of customers in Phase III Integration by 50% to nine
  • Initiated first customer multi-process evaluations
  • 17 engagements underway with 14 customers
  • Completed Atomera's first installation of MST technology at a customer fab


From the Barron's article, quoting Scott Bibaud, Atomera CEO:
“I don’t know if I can quantify the degree of confidence I have in their signing,” he says, “but they are spending a lot of money in evaluation, it’s quite expensive,” suggesting that they wouldn’t be doing so only to end up walking away.

Atomera, with just under 20 employees at the moment, is currently staffing up because “we’re bringing in a lot of new customers” in the evaluation stage.
In August, when this piece was done, Atomera had four customers in the evaluation stage. Now they have 14 per the January 30 news release, representing about 50% of the biggest semiconductor companies in the world.

My bullish thesis would change, of course, if the first customer decision was a "no" as they seem likely to accept or reject the technology as a group.

No comments:

Post a Comment