Significant Achievements

Headings here are subject to change.

Contents

• When we were just a project
• Pre-setup phase
• Setup phase
• The Future…
• Pre-seed funding phase
• Seed funding phase
• R&D phase
• Product phase
• Profit phase
• Scaleup phase

When we were just a project

A project is personal informal thing; not a company nor a business.

2022:

• Extended Boolean logic.
• Various diagrams regarding the beauty of maths - which may become future blog posts.
• Experiments with circuit simulation in SVG and DOM javascript - which later produce a spinoff enhancing Robin's CV.
• Looking for a challenge, but had already designed various processors, so decided to design an open-source FPGA.
  Because this FPGA was designed to run ontop of an existing FPGA, the project got named FPGA^2.
• Later on, tried running an AI on the FPGA; adapted each half to the other for a few cycles; and ended-up with new types of both.

December 2023:

• Trickier parts of the AI formulated, especially the self-modification thing.
• This involved more maths than code.

January 2024:

• Software simulator created (quite a messy/crude one, looking back), all parts of system added.
• Guestimation of 10^14 power rating formulated, then heavily rounded down, as it sounded "too good to be true", to 10^8.

January-March 2024:

Hardware & software for presenting slides created.

April 2024:

• A talk on the hardware, with php-sw, at Runway East Bristol Bridge.
• Feedback included the observation that as the FPGA's design was so radical, it really needed a different acronym. First try was "universal Complex Logic Array", which had to be dropped after we realised it might be confused with a more famous UCLA.
• The talk turned into annotated transcript on GitHub, and a set of slides.

Summer 2024:

Upgrades to timing equipment, prove the power rating is 10^14, after all.
The system had to be run 100,000 times in a row, so the timers had a chance to get a reading.

Pre-setup phase

The pre-setup phase covers the setting up of the company and its assets.

Thursday 10th October 2024:
Surprise offer of a 50K seed grant - priorities suddenly changed!
One of their conditions, is the setting up of a limited company and a business bank account in its name. Three educating weeks follow. Initial team formed (from relevant friends, initially those who'd earlier expressed an interest).

Wednesday 30th October 2024:
Basic offline placeholder website done.

Thursday 31st October 2024:
Registered domain name, placeholder page added.

Sunday 3rd November 2024:
Placeholder website checks & updates.

Tuesday 5th November 2024:

• Generating website graphics.
• Attempts to set up new CMS, thwarted by WSP.
• Modified an existing CMS so it supports multiple domains; added the new domain to it.
• Small pre-setup loan obtained, from a friend.

Wednesday 6th November 2024:

• Began registration process for Companies House.
• Noticed masthead had the wrong wording; updated.

Thursday 7th November 2024:

• Opened corporate business bank account.
• Intensive funding talks at First Thursday in Bath. We generated a lot of interest, and look like one of two potential unicorns - the other is an amazing medical aid innovation.
• Companies House approved registration.
  Appointed Robin Hodson as a director.
• Bank account approved, and activation in progress - slowed down by essential security measures.
• email and Social media accounts set up.
• Company website goes live, including links back to social media.
• Pre-setup phase concludes.

Setup phase

The setup phase is after all the main stuff has been registered, while things like banks are running standard checks, and before funding comes through.

Thursday 7th November 2024:

• Decisions about upgrading project team members to full directorships etc.
• Website expands.

Sunday 10th November 2024:

• The bank account validated enough to start accepting funds.
  Lots of security there, as you can imagine.
• Appointed Tom Lawton as a director.

Tuesday 12th November 2024:

• email systems integrated and upgraded, to cope with the hammering they're getting:
  Scattered logins and redirects just weren't handling it any more.
• Contacted by Ryan LLC (one of us may have clicked upon a survey), kicking-off an email discussion.

Wednesday 13th November 2024:

• The bank account validates enough to get a (currently pointless) debit card, and the final security check.
• Extended/aligned structure of this history page.
• Posted first proper social media illustrated diagram, across all platforms.
• Repaid pre-setup loan, from unemployment benefits, and frugal living.

Thursday 14th November 2024:

• An excellent meeting at the SETsquared Innovation Centre, with University of Bath, or more precisely, a swap to a new mentor, whom also represents the university. 'Looks like the start of something new. We explored the option of lowering the priority of hardware development, in order to accelerate the AI part. This makes a second time in a week we've had to adjust our priorities. A second, bigger meeting is being arranged, to look at funding and the level of cooperation.
• Participation at the final AI West meeting of the year, in the Engine Shed. Graphcore and Bristol University sent speakers. Met with lots of people, all of whom had useful funding advice, and some ofwhich were potential future B2B customers - although Nigel Toon [Graphcore] escaped before the networking began in earnest.

Friday 15th November 2024:

• GitHub and News links dropped from main page (for the time being), as neither are up to a professional standard.
• Added a slightly-pointless collection of legal policies (privacy, T&Cs, disclaimer), because we don't currently collect any user data - nor have any external users. As part of that, also added a contact page.

Saturday 16th November 2024:

• Joined BlueSky as @asi-uk.bsky.social, which also means it's time to revise our social media buttons.

Monday 18th November 2024:
Intending to refactor the main AI code, instead (surprisingly) rewrote it completely from [wetware] memory as a new major version, adding several new features. This consisted of a burst of inspiration during a meeting of sorts, producing a detailed work plan, rather than code per se. So that'll be the code-writer's block sorted out then. All this was prompted by Bath University's preference for concentrating on the software layer first. At least that won't require any further outlay, until we start integrated research.

Tuesday 19th November 2024:
New staff photos, from a professional photographer, largely for ID upgrades, due to an issue with LinkedIn.

Thursday 21st November 2024:

• TechSpark's Third Thursday networking event:
  • Catch-up with Bath University, on progress. Asked for help contacting mutual contacts, due to an issue with LinkedIn.
  • Discussed pitching practice & support with TechSpark.

Friday 22nd November 2024:

• Zoom meeting with Ryan LLC, following several days of email contact.
  Several funding & support options discussed.

Saturday 23rd November 2024:

• Constructive criticism of the hardware layer, sent by the expert panel at Bath University.
  Reply drafted, but not within reach of a desktop machine for a reply at the time.

Sunday 24th November 2024:

• Old test placeholder pages, taken offline.
• Minor corrections to the metadata of legal pages.
• New downloadable media page added,
  and a version of the masthead logo with a white background generated.

Monday 25th November 2024:

• Company declared as dormant to HMRC. HMRC clarifies that setup activities do not count as trading. This is a tax/declaration thing, not anything which affects progress. As soon as funding comes in, we have to declare we're active.
• New original social media buttons, including the new x-Twitter logo, and BlueSky. Future possible buttons, also uploaded for later use.

Thursday 28th November 2024:
Added a YouTube channel, for later use.

Friday 29th November 2024:

• Still working on slides for a bigger meeting much later today at Bath University's main campus. Swapped priorities with the second half of the AI rewrite, to make up time.
• A very nicely eased-into meeting, and somehow everything was ready on time. Particularly helpful for the design to be "proofread", which revealed an interesting flaw, where the self-modification facility was far too slow. Quadrupled the speed on the spot, but that wasn't nearly enough.
• Noticed & corrected how the favicon was linked incorrectly on the main page.

Tuesday 3rd December 2024 (roughly) :
Following a flash of insight, boosted the self-modification speed up by a quarter of the number of LUTs. (This being a parametric design, if there are a million LUTs, the speed has risen by 250,000 times, from the original accidentally-serial start.

Thursday 12th December 2024:
Prompted by a BOSS meeting, which turned out to be knee-deep in Graphcore staff, got a "Pocket Pitchdeck" and summary of the speed increase ready for potential investors. This sort of deadline rush is becoming a habit.

Tuesday 17th December 2024:

• Added a Threads channel, for later use.
• Actually, a lot of other changes; and a lot of work behind the scenes, to catch up on logging.
• Advent/Christmas versions of the logo are overdue now. (Actually, better done during the summer etc, with less pressure.)

Tuesday 24th December 2024:
Added a corporate LinkedIn page.

January-February 2025:
A 7-week involuntary break, caused by extreme cold weather.
Some therapy and bootstrap funding, indirectly from Dyson of all places. Just after that ended, sunshine arrived and life immediately speeded up.

Tuesday 4th March 2025:
Surprisingly, approached by ChipStart UK, to apply to their incubator, probably as a result of earlier interaction.

Wednesday 5th March 2025:

• Maths symmetry, regarding asynchronous incrementers and decrementers. This may become a blog post, as the results will undoubtedly be useful to others, and don't appear to be published anywhere online. Why? Because counters need flip-flops, and those don't work very well when there's no clock. They're also comparatively slow. Asynchronous logic is useful in all sorts of ways. Very little of this is currently available online. Part of a precursor project [to ASI] became a little textbook-like with things like this; it may get tarted-up and recycled.
• Zoom call with Sean Redmond of ChipStart UK: Initially, we believed they were only for businesses which were ready to go into silicon. They corrected us that they mainly get ideas ready as businesses for investors and guide them through. A lot of prep work went into that application, and continued through the week.

Thursday 6th March 2025:

• First Thursday, and reconnecting with Bath University. Both our mentors turned out to be two thirds of the panel speaking at the presentation.
• Reconnecting with contacts which could provide a venue for Bath Digital Festival, and connecting them to relevant the TechSpark contact. Printing ChipStart forms, for drafting and clarity - a mixture of Bath Central Library and Bristol's cat Pub. Another reason for working on things at that Pub is, it's a short walk to Graphcore.
• A meeting as part of MLops Bristol, at Graphcore. Both fascinating and tiring. Not there formally as ASI, but discussed the company there. Graphcore put some photos online.

Sunday 9th March 2025:
Three hours before the deadline, submitted the ChipStart UK application. It took more work than expected, but then doesn't everything which is worthwhile?

Monday 10th March 2025:

• A few minor updates, including spotting and correcting a minor typo on one of the pages cited to ChipStart. How that slipped past in the first place, is a mystery in itself.
• Added daynames to the dates listed here: So much easier to orientate.
• Minor validation check failure (corrected): Reran HTML &c validation checks on all pages. As part of this, DocType changed from HTML 4.01 Transitional, to HTML 5 across all pages. (Date metadata only updated for significant content changes.)

Monday 17th March 2025:

• First-round screening, which consisted of our presentation, followed by a fascinating Q&A session from a panel of European semiconductor industry experts.
• Our presentation was supposed to take 15mins, but we finished in 10min. Discussion was also supposed to fit into 15mins, but overran to 20mins - so that worked to everyone's advantage.
• As a side-effect of this, we've agreed to a deadline of Saturday 17th May, for comparative results from an improved benchmark.

Tuesday 18th March 2025:
We didn't make it to into ChipStart's incubation programme this year, but they are providing us with invaluable analysis and feedback, so really everyone who gets past the application stage is a winner. We're also attending the announcement ceremony in London on Thursday 27th March, as it provides unrivalled networking opportunities.

Wednesday 19th March 2025:
The University of Bath's Enterprise Day. We were invited as guests.
No real idea what that would be like, which was a good reason to go and find out: It was a something very much related to the Innovation Centre or at least the department it lives within, Research & Innovation Services, so lots of our existing contacts were there, plus a chance to mix with other people; kind of like an extended First Thursday. Unfortunately, we arrived a bit later than ideal, due to a schedule clash.

Friday 21st March 2025:
Anonymised feedback meeting (this summary only skims over itself too), from the ChipStart Cohort #3 application and first-round screening:

• This highlighted that we're not ready enough with our revenue plan nor business plan, and our team structure needs work.
  This is no surprise, because we're a deep tech startup, ie initially research only.
• A reiteration that we don't have a public/verifiable demo/benchmark yet: Don't worry; we have a cunning plan.
• Specific feedback, general review, and chat about our idea without giving anything away.
• Strategic adjustment agreement.
• Thoughts about other help which may be available.

Tuesday 25th March 2025:

• Reached the "Limerick stage", of compiling a compiler without a compiler to compile the compiler we were compiling. (A burst of inspiration.)
• "Future" section added [to this History page].
• Styling added to dates, to delineate them.
• Shifted some items in the Future section, into the preceding one, as time had moved on since that idea was first written.
• Busy getting stationary like contact cards ready, for ChipStart's London event (and beyond).
• Inter-team coordination, so nobody (in theory) gets left behind on Thursday.
• Noticed a spike in searches for our LinkedIn page;
  Posted about this history page - partly to cut down on routine social media updates.

Thursday 27th March 2025:
We attended the ChipStart UK Showcase 2025, at the Royal Institute, London. Please note, we are not in the 3rd cohort.
Being able to attend this taps into networking opportunities not available elsewhere. We also met the until-now invisible other applicants.
Robin and Tom went along to this, where there were already many familiar faces from the first-stage screening, as well members of the first cohort based in the Southwest, and from general networking.
This turned out to be a very tiring day, with some other attendees noting it's best to stay overnight in London for these things, particularly as there was a related event the previous day, in Canary Wharf, called the Future of Computing Conference 2025.
The lead organiser tells us that the keynote speaker at the preceding event outlined an ideal future AI system, which we happen to match: So "We are the best of the best, if we can do the impossible."
This was also one week after the start of Spring, so temperatures and available light were starting to climb, which means a faster pace of work is unlocked.

Wednesday 2nd April:

• Simplifying of the initial system in the v2 simulator, which runs hardware in software. Things need to be as simple as possible so useful tests can be tested (whilst still remaining the same type of system, so Eg a car without seats or a roof works, while a car without an engine or wheels doesn't.) (Another burst of inspiration.)

Friday 4th April:
Decent temperatures hit the UK. This helps work more than some people realise, with dramatic changes: A nosedive in electricity use, freedom to move around and get stuff done. Not confined to typing on a phone in one room. A higher quality of free lighting. Able to go outside, able to air the place, free drying of laundry, spontaneous outdoor meetings, and so on...

Sunday 6th April:

• Another burst of inspiration, relating to a type of parallel processing, and an exploration of faster ways of collating information from independent processes. Parallel processing produces strange bottlenecks not encountered with slower systems. So far, we've hit & solved: Novel memory indexing, asynchronous counters (flip-flops are too slow), and output collation. We've also looked at interlaced routing (don't even ask).
• Selecting & testing emulators for disparate portable systems for collaborative working.
• A staff biographies page added, and linked to from the main page.
• Introductory text on the main page, updated.
• Early schedule review, prompted by the warm weather:
  • We've had three recent bursts of inspiration, moving things along enormously, but progress still needs [more] structuring and sharing.
  • Structured scheduling needs a list of what's been done so far, what the deadline goal precisely is and needs to demonstrate; plus some sensible intermediate steps along the way, and a detailed internal tasklist for coding/hardware tasks. This ties into setting agendas, ofwhich the uni is so fond. (Robin thinks he could do with restructuring & tasklisting the rest of his life too.)
  • Collaboration with the uni still needs more work, with the co-working stuff waiting for more practical portable systems to be ready.
  • Inter-team working is coming along nicely, but the structure of the core team definitely has a lot of room for improvement: It's a bunch of friends who agreed to help each other, not something you would design to share work among - even though we do share some work among us.
  • Introspection: Are all parts of the v2 simulator in place? No. And the v1 simulator is a mess, so that's not suitable for complex demands. But most of the subsystems have identified themselves, as work continues: It's a strange self-defining beast.
  • This all also relates to getting sketches and floating ideas into a central system, ideally some kind of machine(s) which can be worked upon by multiple people, and which can produce verifiable results. This being an AI system, some of it has already been designed by the parts of the system which are operational (as if things weren't recursive enough).
  • Observation: Some of this report could be refumbled as the start of the (re)structuring.
• Added a potted history update of the last two weeks, to the LinkedIn page, to give the growing number of followers something to follow: There's enough happening now, to do this every few weeks/days. Maybe we'll drop some teaser descriptions back in, as originally planned.
• Tarted up the LinkedIn page: Added a cover image, noticed the profile image didn't fit properly and adjusted that, corrected minor errors in descriptive text.
• Added different-colour bullets for nested lists - now we're using nested lists.

Monday 7th April:
The assumed start of a bug which hit the Southwest. A great many people were affected, us included.

Wednesday 23rd April:
Midway-through the day, the bug lifted. And thence, a mad scramble to get out to the networking event.
Bristol AI Brainwave networking event. Not there as ASI.
It was here we discovered just how widespread the bug was. Presumably other people are still being hit by it.
We also discovered it ends by losing your voice for a few hours - not great at a networking event.

Thursday 24th April:
The date of the Bristol SC Expo, and hence a search for bootstrapping funding. Here, we struck gold, being virtually grabbed by employers.
There's a likelihood of working in a different part of the country for 4 days out of 7, leaving one day free for collab at Bath uni.
We made significant progress in coding the simplified joining-together part of the v2 simulator.

Friday 25th April:
A day of rest: Sudden rushing around and intensive activity for two days, following three weeks of illness, really takes it out of you.

Monday 28th April:
Catching up with our email backlog, replying to bootstrapping employment offers, etc. Picking up grant-chasing will have to wait until tomorrow.
Helpfully, warmer weather strikes. Coats no longer required in the daytime. (This could be premature; but we're determined to make the most of this summer, regardless of if it's technically begun or not.)

This basic structure/agenda, and I'm sorry if this is repeated elsewhere:
(This may also become the interim halfway report.)

• Done so far:
  • Come up with the basic concept; realise the need for a properly-structured v2 simulator.
  • Sketch out the overall plan for the v2 simulator, and coding the AI.
  • Fixed the indexing problem.
  • A string of inspirations (which may get listed here).
• Doing and partly-done:
  • Tidying up in filing; summarising the inspirations along the way.
  • Code a simplified v2 simulator, non-parametric etc, including a display module. At first, this will be the simplest thing; an I/O module. Then other parts get added, followed by some test logic. Notation suitable for pen & paper, to be finalised with regard to making clear specifics.
  • Experiments in mobile coding, and specific collab.
• Next to get on with:
  • Work out surrounding structures, as per the five-layer separation of concerns.
  • Add a crude test benchmark; purpose being merely a systemic test.
  • Add back in equivalents from the v1 simulator, eg timing tools, and a crude AI for working out how to encode simple components.
  • Expand the simplified v2 simulator, to full parametric, adding in a better concept of config memory, plus indexing components.
  • Extend the benchmark, to the envisaged video game. Code several games. Try a simple board game, eg Draughts.
  • Hammer the codebase with linting and tests.
  • Mock benchmark test.
  • Probably with collab, create a comparison ANN system, which also uses the same benchmark. This could be running on separate hardware, if that's more practical.
  • Mock benchmark test.
  • Full benchmark test, for reporting purposes, comparing the AI, an ANN, and a random guessing algorithm.
• Beyond the benchmark:
  • FPGA devboard experiments, including separating the display & control system from the simulated hardware.
  • At some point, move the AI over entirely to running on an devboard. Experiment with multiple parallel host FPGAs.
  • Feed the AI a series of problems of increasing complexity, and see how rapidly it adapts, and how well it performs.
  • Maybe a nice game of Chess, or perhaps Go.
  • Experiments with sandboxing hardware, and other security measures.

Thursday 17th April:
(This was previously a planned future date.) Schedule review, halfway to the benchmark deadline.
This got a little delayed, then overtaken by other work, due to a widespread bug (physical illness) which struck down people across the Southwest UK and probably further afield. They say in this game, "Don't fall ill", which is all well and good unless you do. They mean soldiering on through minor illnesses. Sometimes worse things come along, and that's beyond anyone's control.

Thursday 1st May:
(This was previously a planned future date.) Rescheduled halfway point review, three-quarter's-way along.
This didn't happen. We were deep in the maths phase.

Saturday 17th May:
(This was previously a planned future date.) The deadline we came up with, during ChipStart's first-round screening panel discussion, for comparative results from an improved benchmark.
We have no idea if, or by how much, the involuntary three-week stoppage in the middle, will impact meeting this deadline. Deadlines are supposed to be motivational. The warmer weather is likely to speed up progress anyway.
This didn't happen. We were deep in the maths phase. The benchmark stage itself, since got extended from a one-day announcement, into a process.

Sunday 1st June 2025:
Artwork for backdrops and themes for future systems prepared, and fortunately exported for critical systems to use - which means it wasn't clobbered by photoretouching software on a non-critical system going down the following day.

Monday 2nd June 2025:

• A non-critical system dies, SSD interface fried. The only thing related to ASI which was on that system, was a draft update to this history page. Frankly though, it was a mess, and needed a complete rewrite anyway. It also was too early to include any maths indexing. So although it looked like things had stalled, work was as intensive as ever, and no critical systems were affected, because everything assessed as critical is made of tougher stuff anyway.
• An unexpected side-effect of a unit assessed as non-critical failing, was that because all online lookups and fiscal logins had to be switched to geographically-distant facilities, it slowed down everything else. So some sort of substitute was needed while it was being repaired. Several experimental combinations of spare hardware followed - to avoid spending anything on new equipment. This experimentation also interrupted things and burned time.

Thursday 5th June 2025:
Minor alignment meeting at Bath uni. Coincides with an accountability seminar.

Friday 6th June 2025:
Rough ASI Coeus portable system working, but not truly portable yet. This also gets online tasks trialled on it, but it doesn't perform well there; not least because it's designed for fast offline coding only. It is extremely fast for the tasks it was designed for.

Thursday 19th June 2025:

• Installed sunscreens, to stop heat building up from Southfacing windows, as the second early mini-heatwave of 2025 hit.
• Basic Linux install working a RPi4 ripped out of an experimental tablet; online access restored.

Friday 20th June 2025:
Installed aircon, to claw back "unlost" time.

Wednesday 25th June 2025:
Data protection compliance: Registered as exempt [from paying the data protection fee this year] with the ICO - at least until we become a fully-active company.

Friday 27th June 2025:

• Chipstart ARM HQ event (couldn't afford to attend)
• ASI Coeus: Adapted an existing small tablet stand, at a saving of £18 over buying a fancy monitor stand, with hardware ripped out of an experimental tablet.

Sunday 29th June 2025:
Portable solution 3, ie ASI Coeus, becomes fully portable.
Solutions 1 & 2, were based around Android phones, mini-keypads, and just-good-enough text editors and language ports. Although they were much more portable, they weren't practical solutions for serious work.

Thursday 3rd July 2025:

• Minor realignment meeting, Bath uni.
• Busy with neglected filing/indexing for our previous maths breakthroughs, for coding and project management (and history tracking). Also coincided with an pitching seminar.

Tuesday 8th July 2025:

• Full verification for Companies House with the new GOV.UK One Login, except their system is a bit broken, and doesn't link to anything yet. Everything also made ready for the switch to software filing in November - 21 months ahead of the deadline at the start of April 2027.
• Having previously ruled out applying for any grants before the benchmarking process starts, a matching really ultra-stage grant is spotted, and everyone jumps on it.

Wednesday 9th July 2025:
As a side-effect of various declarations being made, enough dates got logged to do an early History page update. [Abbreviated] Maths breakthrough dates will be added when that filing/indexing process is complete.

The Future…

Things booked which haven't happened yet. After they occur, they'll be moved into the relevant past history section, and updated with what happened.

Thursday 10th July 2025:
Deferred deadline, for a basic v2 Simulator, ie the basic framework and display/diagnostics: It always helps to see what I'm doing. Some early progress with preserving the style guide from precursor work, has already been done. The majority of the serious work is still on paper: Nothing else is flexible enough, when charting the unknown.

Thursday 17th July 2025:
Realistic deadline, for a basic v2 Simulator.

Thursday 31st July 2025:
Early deadline, for a full v2 Simulator, ie various layers including FPGA, AI, and monitoring systems. When this done, the benchmarking process can begin.

Thursday 7th August 2025:

• Realistic deadline, for a full v2 Simulator.
• Next Bath uni alignment meeting planned. No seminar this time, because visitors are assumed to all be on holiday.

Pre-seed funding phase

The pre-seed funding phase is when initial funding comes through. During this phase, we move to full employer status, and burn lots of money on accountants, hire a fractional CFO, etc. Alternatively, a drop to corporate dormancy during alternative funding/support. Bootstrapping may play a role.

Saturday 1st November:
A possible minimum start date linked to a grant, formulated by the government, which is probably why it unhelpfully falls on a weekend.
Although that grant can't be spent before this date, prep work can continue. Grant-funded work is for research into applying the system to real-world problems, and for the move into hardware. Spin-off products which may also generate income will live here too, and appropriate engineers and researchers will be hired, along with any resources they may need. At least, that's the theory: Grant applications were being written in parallel with this plan. It's quite likely that other investment will arrive long before this time.

Seed funding phase

The seed funding phase is when more/supporting funding comes through. We won't really know what this will mean until after we're in pre-seed funding: It's partly a standard distinction. I suppose it depends how much of each there is. If pre-seed is under 100K, seed will make it up to that; if not, the seed stage will be much further in the future.

R&D phase

The time when we find ourselves back in the lab.

Product phase

This is when we start to see sellable product(s).

Profit phase

This is when we start selling stuff: An important stage in any business.

Scaleup phase

This will be when we go into silicon, ie start making our own chips. It may possibly arrive before the profit phase.

ASI homepage.