Member Info for Chaebol

Lol it's on AIM and its in Nigeria, and I don't think I have an easier hold right now than this !

Im sure this zooms again soon above 1 pound plus, too many good things going on with this.

GLA

Stablecoin ramp over, Lloyd 88 been super quiet and he was hopeful of this.

4 EMI Licenses awarded this year, I am noticing the valuations of these companies and the money they have recently raised to expand with EMI. Incard recent 10 million raise more than TOTAL GST MCAP. I wonder what you say when the regulator asks the group that is backing you is falling like a stone. It’s down 90% since the EMI app started month 15 next week. Thats not volatility, thats a broken company ! Would love to hear their explanation…….’in line with management expectations’ not gonna cut the mustard x;)

1. Ripple (Ripple Markets UK Ltd)

* Estimated Worth: $50 billion.

* Total Raised: Over $890 million.

* Key Detail: Its valuation surged 25% in early 2026 (from $40bn in late 2025) following a $500 million strategic round led by Fortress and Citadel. 

2. Capi Money (Capi Money UK Ltd)

* Estimated Worth: Approximately $91.5 million (£72m).

* Total Raised: $22.6 million (£17.8m).

* Key Detail: The firm secured £15 million in a Series A round in early 2025 led by Creandum to scale its international payments platform for emerging markets. 

3. Crezco (Crezco Limited) 

* Estimated Worth: Estimated between $50 million – $75 million (based on standard Series A multiples).

* Total Raised: $18.1 million (£14.3m).

* Key Detail: Its most recent major raise was a $12 million Series A in late 2023, which allowed it to replace Wise as the primary bill-payment partner for Xero. 

4. Incard

* Estimated Worth: Estimated £40 million – £50 million (based on recent Series A and transaction volume).

* Total Raised: $15.2 million (£12m).

* Key Detail: Just weeks before receiving its EMI licence, Incard raised £10 million (February 2026) to expand its "Financial OS" into the US and European markets.

Could this be the BOD plan....... take it private at a lot less than the claimed assets?

As this SP continues to go lower and the bid disintegrates the usual suspects say isn't this excellent x;) very strange tbh.

When you have muldoons the adams admitting they sold at much lower prices than the 3.5's I would imagine the sellers over the last 4/5 months are people just like Muldoon and Adam.

even the .67 sale from Muldoon looks buffet-esque now.

Was probably 95k once for them at one point.

Must have been a tough sell x;)

I wish them well

And off we go again......

Longbarn you didnt "sell up or own" never have said I sold it! or owned it, you all assumed it. Can you not get into your skull why I was prepared to sacrifice that address and account, I really cant believe you are that low IQ, sincerely.

Baghdad youve had movies and pictures and its still not good enough ! lol thats bitterly twisted to facts and your own eyes deceiving you.

Its beautiful to see you this twisted up and it has nothing to do with your investment, must be driving you mad that I keep calling this right. Go and touch some grass you buffalo x

And youre all in on Chaebol derangement syndrome.

I wish you well.

Im off every day, are you not on holiday more?

Chaebol, you do realise you can use your online broker to get buy and sell quotes and if your quoted it means you can make that transaction if you want.

Aww thanks for that leew, I was always unsure..............

Asterix!!!!! I am taking the pish like you are

Green box delirium won you all the prizes and awards massi.

Dont worry Massi x Im going to Lebanon and Damascus next month, just like my recent visit tt Saudi and Baghdad ill post holiday pics as you all seem so interested. x

Lol and on and on the Massi goes as it goes lower........... tells me hes a winner all with words......

Trading stmnts fake, the land they sit on fake lol, you guys are absurd with it, just couldnt handle that truth! Im 7 shots ahead of you lot every time, its amusing.......you an Aerry massi what a double act! you stumbling over push the envelope with him lol , you love to see it x;)

Sorry I've been successful bud x get over yourself. That I have to prove land sales etc is just another riddle in the rampers mayhem and I should just expect to beleive you, that 33 million shares could be sold because your broker told you. Lol

I have seen these orders on level 2 for months and as it goes lower they never get filled. but like I say this one must be real because you claim your broker told you x;)

All we can do is wait for .35's they're coming. x;)

Lol

Lol so what happened to all those large orders on level 2 from .6 down because theres been some large ones and every time they do not fill but this one is real lol.

Metapay max revs 1.5 mill a month yeh for SPI at 1% thats 15 k a month revenues and thats going to be good news if so..........

What they gonna tell us next they're applying for EMI with Metapay lol

As of February 24, 2026, GSTechnologies (GST) does not have the required PSD2 (Payment Services Directive 2) authorisation for stablecoin-related payment services. While it operates as a VASP (Virtual Asset Service Provider) through its newly acquired Polish entity, this status is insufficient to meet the specific "cliff edge" requirements starting 2 March 2026.

The "March 2nd" Regulatory Cliff

The European Banking Authority (EBA) has mandated that any entity providing "payment-like" services for stablecoins—officially E-Money Tokens (EMTs)—must hold a formal PSD2 license (as a Payment Institution or Electronic Money Institution) by 2 March 2026.

Https://www.semnet.co/post/semnet-and-netformx-announce-strategic-partnership-to-accelerate-ai-ready-data-centre-and-cybersecur

Qn: If power & cooling represent 20–25% of capex yet interact tightly with networking dependencies, how might emerging AI workload shifts (e.g., toward sparse mixtures-of-experts or longer-context inference) accelerate obsolescence in cooling infrastructure, potentially amplifying the stranding risks described?

Our Response: Power/cooling obsolescence from workload shifts: Power/cooling is already 20–25% of capex and tightly coupled to networking (e.g., liquid cooling for dense GPU trays). Shifts to sparse MoE (lower average power but spiky peaks) or long-context inference (higher memory-bound loads) change thermal profiles—more bursty, potentially hotter hotspots. This accelerates obsolescence in fixed air-cooling or early liquid designs that can't adapt without downtime/replumbing. Stranding risk rises if cooling can't keep pace with density ramps, effectively capping usable GPUs even when power is available.

Qn: The proposed Netformx-Vaultrex integration aims to close the Continuous Threat Exposure Management loop—how might this align (or create friction) with evolving regulatory regimes for critical AI infrastructure (e.g., EU AI Act high-risk classifications or U.S. executive orders on AI safety), particularly around auditability and third-party validation?

Our Response: Alignment/friction with regulatory regimes: The Netformx-Vaultrex CTEM loop (continuous discovery → prioritization → remediation → validation) maps well to auditability requirements: EU AI Act high-risk systems demand ongoing risk management and third-party conformity assessments; US executive orders emphasize safety testing and red-teaming. The friction is minimal—continuous logs and attestations actually ease compliance vs. periodic audits. It could even become a competitive advantage for "regulated inference" use cases.

Qn: In a world of proliferating AI data centers resembling industrial assets, what supply-chain concentration risks (e.g., dominance of few GPU/networking vendors) could systematically amplify the lifecycle drift and cyber exposure vulnerabilities outlined, turning individual operator risks into sector-wide systemic threats?

Our Response: Supply-chain concentration amplifying systemic risks: Heavy reliance on 2–3 vendors (NVIDIA GPUs, Cisco/Broadcom/NVIDIA networking, etc.) creates correlated risks: a firmware bug, EoL announcement, or supply disruption hits many operators simultaneously, turning idiosyncratic lifecycle drift into sector-wide stranding or cyber exposure spikes. We've seen this in past GPU shortages; it could manifest as synchronized capex write-downs or higher insurance premiums across the industry.

Again, thanks for the depth—this kind of dialogue is gold.

Alter ego John Tech man loves Melvin, did not expect that tbh

Our Response: Verifiable continuous assurance as an economic moat: In a commoditizing market, operators who can prove "always-current, low-drift, breach-resilient" infrastructure via Netformx-style modeling + CTEM loops can credibly differentiate on SLA terms: higher uptime guarantees, lower risk of training-run invalidation, verifiable compliance. This translates to premium pricing for inference/colocation (5–15% uplift) because customers (hyperscalers, enterprises) pay for reduced variance in performance and lower insurance/financing costs. It's similar to how "green" data centres command premiums today—verifiable resilience becomes a credible commitment that competitors can't easily replicate without the same continuous discovery layer.

Qn: The article highlights irreversible failure modes unique to AI facilities (e.g., a single unsupported switch stranding GPUs or invalidating multi-month training runs)—what specific architectural trade-offs in modern AI fabrics (e.g., RoCE vs. InfiniBand, disaggregated storage) most exacerbate these catastrophic risks compared to hyperscale cloud data centers?

Our Response: Architectural trade-offs exacerbating catastrophic risks: AI fabrics prioritize low-latency, high-bandwidth east-west traffic, but choices amplify fragility:

* RoCE vs. InfiniBand: RoCE (RDMA over Converged Ethernet) is cheaper and more open but relies heavily on lossless Ethernet (PFC/DC-QCN), which can create head-of-line blocking and cascade failures if a single switch firmware drifts or congests. InfiniBand is more closed/self-contained with native adaptive routing and congestion control, potentially more resilient to partial failures, but vendor lock-in accelerates obsolescence when NVIDIA evolves the stack.


* Disaggregated storage: Great for flexibility, but introduces more dependencies (e.g., NVMe-oF paths) that can strand GPUs if a single storage target or fabric link degrades. Hyperscalers mitigate this with massive redundancy and custom silicon; smaller operators are more exposed to single points of failure invalidating runs.

*

* Qn: Beyond the quantified opex leakage of $10–20M/year from reactive maintenance and downtime, what are the harder-to-measure reputational and opportunity costs when a cyber-induced model theft silently erodes a firm's proprietary AI moat over multiple product cycles?

*

* Our Response: Harder-to-measure reputational and opportunity costs from silent model theft: Beyond direct opex, the real damage is erosion of the proprietary moat. A stolen model can shortcut competitors' R&D by 12–24 months, compressing product cycles and forcing price wars or feature parity. Reputational hit compounds if customers perceive the breach (even if undisclosed), leading to churn or higher acquisition costs. Quantitatively, we've modeled it as a 5–15% haircut to lifetime value of IP-protected products—hard to pin down but material in competitive markets like foundation models or enterprise AI.

Our Response: Formalizing maintenance intelligence as risk-adjusted ROI: Treating it as "capital insurance" is spot on. To compete/complement traditional cyber insurance (which covers tail events but not chronic erosion), we can frame a risk-adjusted ROI as: Expected value = (Capex preserved + Opex saved + Utilization uplift) × (1 – residual risk probability) – implementation cost. Using VaR or CVaR for tail risks, you could show that $10–30M preserved over 5 years equates to a 15–30% IRR on the assurance layer, especially when it lowers breach probability enough to reduce premiums or enable better financing terms. The key differentiator vs. standard cyber policies: it addresses both attritional (chronic stranding) and severe (breach) losses, making it more of a "productivity hedge" than pure transfer. We can build a simple decision tree or real-options model to quantify the value of the embedded flexibility (e.g., targeted upgrades avoiding full forklift replacements.

Qn: In an "assume breach" paradigm amplified by east-west traffic dominance and lateral movement risks, how resilient are cryptographic control planes (like Vaultrex) to advanced persistent threats that target key management infrastructure itself, or to near-term quantum grover/shor algorithms weakening encryption primitives?

Our Response: Resilience of cryptographic control planes (Vaultrex) to APTs and quantum threats Vaultrex uses asymmetric multi-key secret sharing (Shamir-inspired but distributed), so no single node holds the full key—compromising one instance doesn't yield usable material. This makes it robust against traditional APT lateral movement, as key material is never reconstituted except ephemerally at the point of use. For quantum: Groshor/Shor attacks target asymmetric primitives, so Vaultrex (like any system) would need post-quantum upgrades to key-gen and exchange layers (e.g., Kyber/ML-KEM hybrids). We're already tracking NIST PQC standards and can layer in hybrid modes today without breaking existing flows. The bigger near-term risk is physical or insider compromise of the key-distribution network, which Vaultrex mitigates via geographic/organizational segmentation and zero-trust access.  Do note that with Vaultrex's assymetric encryption technology, this is already a leapfrog against existing data protection and mainstream encryption methods and philosophy.

Qn: With an impending surge in new AI data centers intensifying competition, how might verifiable continuous assurance (e.g., Netformx-style lifecycle modeling integrated with CTEM loops) create economic moats, enabling operators to command premium pricing for inference/colocation services despite commoditization pressures?

Qn: The article estimates $15–30M annual capex erosion in a $1B AI data center from obsolescence and GPU stranding alone—how might these compounded hidden costs distort traditional DCF valuation models for AI infrastructure investments, particularly when investors over-index on upfront GPU scale rather than lifecycle resilience?

Our Response: DCF valuation distortions from hidden capex erosion: Traditional DCF models for AI infrastructure often front-load the value capture around GPU scale and initial training throughput, applying relatively low discount rates (8–12%) because the assets look long-lived on paper. But when 5–10% of the asset base obsolesces early and GPUs strand at 2–4%, the effective economic life shortens dramatically, and cash flows decay faster than modeled. This creates an upward bias in NPV—investors overpay for scale while underweighting lifecycle drag. A more realistic adjustment would layer in probabilistic obsolescence curves (e.g., Weibull distribution for hardware drift) and run sensitivity analyses where utilization drops 1–2% annually from unaddressed dependencies. The result: implied discount rates could jump 300–500 bps to reflect the compounded risk, or you'd need to haircut terminal value by 20–40%. In practice, we've seen funds quietly add "resilience premiums" (higher hurdle rates or lower multiples) for facilities without continuous lifecycle visibility.

Qn: Given that early obsolescence affects 5–10% of assets within 24–36 months (driven by firmware incompatibilities, EoL/EoS, and evolving traffic patterns), what econometric approaches could quantify the second-order effects on total factor productivity in AI training clusters over a 5–10 year horizon?

Our Response: Econometric quantification of second-order TFP effects: Over 5–10 years, the second-order hits to total factor productivity in AI clusters come from compounding: early firmware drift → unstable fabrics → aborted/abnormal runs → lost gradient steps → slower convergence or higher variance in model quality. Useful approaches include:

* Panel data regressions on cluster-level logs (utilization, MTTR, job completion rates) with fixed effects for vendor/model type.


* Stochastic frontier analysis to separate "technical inefficiency" (drift-induced) from noise.


* Monte Carlo simulations calibrated to obsolescence rates, propagating shocks through training throughput to end-model performance (e.g., perplexity drift). We've toyed with agent-based models where each GPU/node has a hazard function for stranding; aggregate TFP erosion can reach 1–3% annualized in unmanaged clusters, which compounds massively on multi-year training horizons.


Qn: The piece frames maintenance intelligence as "capital insurance" yielding $10–30M capex preservation over 5 years—how could this be formalized into a risk-adjusted ROI framework that competes with (or complements) traditional cyber insurance products, especially for tail risks priced at $50–200M pe