Artemis Gold - Looking Good In The Short To Medium Term

Executive Summary

Artemis Gold Incorporated (“Artemis”) (TSX:ARTG) is a Canadian resources company with a diluted enterprise value (“EV”) of C$1.89 billion. The Company is developing the Blackwater Mine, a gold project located in central British Colombia. 

Artemis recently announced that overall construction at Blackwater was 59% complete, and approximately C$389 million of the guided initial capital expenditure of C$730 to C$750 million had been spent. The project appears to be advancing well. Production is anticipated to start in H2 2024, and then ramp up to a Phase 1 production rate of 321,000 ounces per annum (“ozpa”). So far so good.

Crux Investor values the gold project on an NPV₈ basis at C$2.7 billion. Included in this valuation is a 20% increase in operating cost and a 25% increase in the life-of-mine (“LOM”) capital expenditure, relative to the figures published in the Feasibility Study. Despite the conservative valuation by Crux Investor, Artemis is still trading at a 31% discount, which – at face value at least – represents a good opportunity for investors. 

Underpinning the robustness of the investor case is the Phase 1 open pit. A review of the cross-sections for the 33 million tonne (“Mt”) starter pit shows a high-grade block that is well supported by drilling. Production during the first five years at a strip ratio below 1.8:1 and a feed grade above 1.5 g/t Au should generate strong cash flow, repay capital, and reward shareholders. With a short to medium term investment horizon Crux Investor can see shareholders making money as the project is de-risked and the discount gap to NPV gets closed. Once Blackwater is up and running and producing over 320,000 ounces of gold each year the share price should be considerably higher than it is today. 

Again, so far so good. For those that want independent confirmation that their investment in Artemis is a good one, stop reading here. 

For those that want to look a little deeper, however, read on. There is a complexity to the Artemis story that relies on knowing a bit of the background to both Artemis as a company and also to Blackwater as a project. Here goes:

Artemis was a 2019 spin-out from Atlantic Gold Corporation (“Atlantic”) which owned the Moose River gold project. Atlantic and the Moose River gold project were bought by St Barbara Limited (“St Barbara”) for a fabulous price and was a ‘win’ for Atlantic shareholders. However, the purchase of Atlantic for Moose River by St Barbara has not been a success for St Barbara shareholders. The project has fallen very short of the forecast feasibility study performance. There are many cautionary tales that can be learnt from the work that went into the Moose River project. What is relevant for Artemis investors is that Artemis at Blackwater is running with largely the same technical team – Moose Mountain Technical Services - and the same approach to the data that was applied to Moose River. 

And Moose River has proven to be a technical disaster. In short, the successful commercial exit by Atlantic Gold from Moose River masks the technically inappropriate assumptions that were applied to the Moose River deposit. And Crux Investor observes that a repetition of similar assumptions may result in an equally unsuccessful outcome, in the long run, at Blackwater.  

The three main concerns that Crux Investor has about the longer-term development plans at Blackwater are as follows:

1. The multi-phase development of Blackwater, introducing complexity and risk
2. Aggressive recovery and operating cost assumptions
3. Unheeded resource lessons from the past, with reference to the Moose River project.

Phased Development

When Blackwater was bought by Artemis from New Gold in August 2020 the project was at a Feasibility Study stage with a mineral resource largely established and metallurgical test work largely complete. Soon after the purchase, Artemis announced it had greatly improved the economics of the project by reducing upfront capital expenditure and having a phased production ramp-up. 

On paper a phased approach to developing a mining project is easy. Save capital by building small in Phase 1 and then use free cash flow to fund subsequent development phases. Easy to say, not so easy to do.  In practice such an approach is complicated and rarely works smoothly. The challenge is that the skills and focus required to carry out a build-phase are very different to the skills and focus required to maintain steady-state production. Operating a mine is hard enough, without diverting attention away to a new build phase. Conflicts of strategic interest often arise between the operational and project teams each with their own priorities. 

A phased development was embarked upon by Atlantic for Moose River, and it did not end well. Atlantic had developed and built the Moose River project in Nova Scotia based on the concept of starting the operation on two deposits with an expansion phase to follow a few years later by bringing two other deposits into production. In the end St Barbara never managed to construct an expansion at Moose River.

Bringing it up to date, with the plan for Blackwater, Artemis is taking the concept to an extreme level. The plan is to have three main planned phases of development, requiring investment capital of C$730-750 million, C$347 million, and C$374 million respectively. Crux Investor feels that these plans should come with health warnings.

Aggressive Recovery and Cost Assumptions

The second area of concern relates to cost and recovery assumptions used in the Artemis 2021 Feasibility Study. As noted above a lot of the raw data for Blackwater was generated by New Gold for its 2014 Feasibility Study. 

When Artemis published its own technical report using largely the same data that New Gold used, it reported an approximately 8% improvement in overall gold recoveries. Crux Investor notes that Artemis ignores the importance of the degree of oxidation of the mineralisation on metallurgical performance in the New Gold test work. It is estimated that 3% of the mineralisation is oxide and 11% is transitional. In, addition, although New Gold concluded that “coarse gold is not statistically abundant enough to warrant a dedicated processing strategy like gravity concentration”, Artemis has included gravity concentration in the process flowsheet. Results show recovery improvements of 1-3%. Despite this, the Artemis technical report has much higher overall recoveries in the mid-nineties compared to New Gold’s suggested recovery of in the mid-eighties.  

Crux Investor is concerned by the major difference between the earlier and later test work results and which are not discussed nor substantiated by Artemis. 

When it comes to costs, Crux Investor has a large database of real mining, processing, and G&A operating costs which serve as a useful benchmark. The same applies for capital costs. Crux Investor does not believe that Artemis has a secret sauce that will enable it to operate at costs below industry standards, and yet Artemis is using an overall operating cost which is about 20% below benchmark. From a capital cost perspective, Artemis is using an overall life-of-mine capital expenditure cost which is about 25% below benchmark. Investors would be well advised to expect a late-stage initial capital cost increase announcement. This is in addition to the C$50 million capital increase that was already announced in June 2023.  The Crux Investor’s cash flow model has included 20% higher operating cost and 25% higher life of mine (“LOM”) capital expenditure provisions.

The capital cost overruns are likely to appear very late in the process, just when the balance sheet is at its weakest. Look out for a late-stage financing at a discount to the prevailing share price. 

Unheeded Resource Lessons from the Past

The final area of concern relates to the resource estimate for Blackwater Mine.

By way of a refresher, the Moose River project deposits were characterised by low grade mineralisation with locally some very high grades. The standard way of treating a statistical population with erratic high grade gold samples is to top-cut, or cap grades, or to severely restrict their range of influence. Given the Moose River deposits would not have been economic when cutting these very high grades, the resource estimation consultants of Atlantic chose multiple indicator kriging (“MIK”) as an estimation method. This method does not discount or limit the influence of very high grades, but uses kriging for separate ranges of grades (“bins”). The ranges of influence for each grade bin are established through variography analysis of that particular bin. The work showed that the highest grade bin had an extremely high nugget effect of 80%. Remember that the nugget effect is an expression of the typical or inherent difference between the grade of two samples taken almost adjacent to each other.

In other words, the highest grade bins exhibit extreme short-scale randomness and one cannot responsibly use high nugget effect bins to predict grade over relatively large distances. Unfortunately, the consultant running the Moose River resource estimation did exactly that. Furthermore, the reserve estimation assumes during mining operations a high degree of selectivity in being able to distinguish ore from grade.

Why is this important? Well, production numbers from Moose River showed the grade mined was a lot lower than planned and the actual strip ratio a lot lower than forecast. It implies that much waste was classified as ore and was treated, thereby reducing feed grade and waste stripping. And Crux Investor can see that Moose Mountain Technical Services is also using Multiple Indicator Kriging at Blackwater. 

The grade population for the whole Blackwater deposit shows extreme skewness. Approximately 85% of the gold values are below 0.5 g/t Au and a considerable amount of gold is associated with grades in the upper 10% of the data. Put another way, at Blackwater the average grade of the deposit is highly dependent on how a small number of very high grades are treated, similar to the Moose River deposits. The New Gold resource estimate for Blackwater cut the very high values to reduce the variance in data and then used Ordinary Kriging. Artemis management has again used MIK for grade estimation, the method that proved to forecast production grade poorly for Moose River.  

The fact that the overall mineral resource estimation does not differ much from that of New Gold, which used capping of very high grades and ordinary kriging, should give the market comfort. However, Crux Investor notes that neither New Gold nor Artemis provide the results of variography to motivate the search radii used. In a technical report also dated 2014 New Gold included numerous examples of variograms for Rainy River, its main project. 

Moreover, Crux Investor also records that Artemis does not present the results of an extensive grade control drilling programme over a substantial portion of the deposit to validate the resource estimation. It raises the question why Artemis did not present the grade control drilling results and why did it not use the extensive additional data for resource estimation.

Crux Investor therefore concludes that neither the New Gold nor the Artemis estimation exercises for Blackwater are convincing.

As a standard process, Crux Investor reviews deposits on a section by section basis to better understand grade distribution within a resource estimate. The sections with drillholes results through the Blackwater deposit shows that they generally correlate poorly, section to section and drillhole to drillhole. The exception is for a portion of the deposit along coordinate 5200E where good grades over considerable lengths are consistently present. It is this portion of the deposit that will be developed in Phase 1. It is this portion of the deposit that comprises almost 33 Mt with a low strip ratio of 1.8:1 and an average grade of 1.5 g/t Au. And it is this portion of the deposit which should enable Artemis to have a good run in the short to medium term.

Introduction

Artemis Gold Incorporated (“Artemis”) (TSX:ARTG) is a very young company incorporated in January 2019 and was at the time a wholly owned subsidiary of Atlantic Gold Corporation (“Atlantic”). Management must have been making preparation for the conclusion of the disposal of their Moose River project to an Australian company, St Barbara Limited (“St Barbara”). As part of the purchase of all the holding company’s shares it was agreed to spin out Artemis. As an aside, St Barbara would come to regret its acquisition in the following years with the project falling very short of the forecast feasibility study performance.

On 21 August 2020 Artemis acquired the Blackwater Project from New Gold Incorporated (“New Gold”) for C$140 million cash, 7.4 million Artemis shares, a cash payment of C$50 million deferred for one year and a gold stream in favour of New Gold. The Gold Stream Agreement granted New Gold the right to purchase 8.0% of the refined gold produced from the Blackwater Project up to 279,908 ounces of refined gold after which the gold stream will reduce to 4.0%. New Gold will pay for the gold at 35% of the spot price.

The acquisition was funded through a placement of 64.83 million shares at C$2.70 each.

Figure 1_1 shows the share price performance of Artemis on the Toronto Stock Exchange since 30 September 2019 and how it responded to various events and developments.

he graph shows how the market enthusiastically reacted to the announcement in June 2 of Artemis’s intention to acquire the Blackwater project, which was very advanced and essentially construction ready for which New Gold had published a positive feasibility study in 2014. Somehow New Gold never acted on this, possibly as it has its hands full with the Rainy River project, which was not going well.

‍
A prefeasibility study published almost immediately following the acquisition gave a net present value at a 5% discount rate (“NPV₅”) of C$2.25 billion. Based on this New Gold had sold its project for an extremely low price.

A feasibility study published in September 2021 confirmed the great improvement on the New Gold feasibility study results, yielding a NPV₅ of C$2.15 billion, much higher than the C$0.62 billion arrived at by New Gold using a cut-off grade that was only 10% higher.

‍
According to Artemis it had greatly improved the economics of the project plan by reducing upfront capital expenditure by staging phased production ramp-up. This is similar to the Moose River project disposed to St Barbara and what proved disastrous. Atlantic had developed and constructed the Moose River project in Nova Scotia based on the concept of starting the operation on two deposits with an expansion phase to follow a few years later by bringing two other deposits into production. The Moose River project deposits were characterised by low grade mineralisation with locally some very high grades. As the deposits would not have been economical when cutting these very high grades, the resource estimation consultants of Atlantic had chosen multiple indicator kriging (“MIK”) as estimation method. This method does not discount or limit the influence of very high grades, but use kriging for separate ranges of grades (“bins”). The ranges of influence for each grade bin are established through variography of that bin. Despite having an extremely high nugget effect of 80% for the highest-grade bin, the consultant used these high grades to inform the block grades over relatively large distances. Worse, the reserve estimation assumed a high degree of selectivity in being able to distinguish ore from grade.

The market reacted positively to the feasibility study announcement with the share price rising from C$5.50 to C$7.39 by 19 November 2021.

The conclusion of a Silver Stream Agreement in December for proceeds of US$141 million did not have an impact despite it greatly contributing to funding project development. The construction funding was basically assured in February 2022 with a commitment letter from a consortium of financiers to provide C$360 million plus a C$40 million overrun facility. Typical for a company that embarks on the mine construction phase, the share price thereafter dropped substantially. Since the end of July 2022 the price has however been trending up as construction seem to progress without major setbacks.

In May 2022 the company concluded an EPC (“= Engineering, Procurement and Construction) contract with Sedgman Canada Limited (“Sedgeman”) for a fixed price of C$318 million, assuming that construction mobilisation would occur in Q1 2023 and commissioning activities to start in H1 2024. This contract substantially reduced the risk of overruns for Blackwater as the potential cost adjustments were limited to +/-15%. Also in May 2022, an Equipment Lease agreement with Caterpillar was executed.

With the project almost fully funded and a fixed EPCM contract for plant construction concluded, site works for preparatory activities and facilities started at the end of September 2022. The last funding activity occurred in October 2022 when the company placed shares for total proceeds of C$175 million.

In March 2023 approval of the British Columbian Mines Act Permit was obtained, which is the final step required to allow commencement of major construction activities. A press release in the same month indicated that all was on track with Sedgman having completed approximately 63% of the detailed engineering for the processing plant and that over 90% of the total processing equipment packages have been awarded.

On 1 March 2023 the company announced it had finalised the project financing amounting to C$385 million with a C$40 million standby cost overrun facility announced one year earlier. The company indicated that it is considering bringing Phase 2 expansion forward and the results of the engineering study are expected early in 2024.

By year-end 2023 overall construction was 59% complete and 84% of the lower-end of capital expenditure either spent or entered into contractual commitments. The spending progress is deemed consistent with a typical S-curve.

Valuation of the Blackwater Project

Introduction

The technical information referred to by this report are extracted from two NI. 43-101 compliant reports on the feasibility studies drafted by AMEC Americas Ltd. (“AMEC”) as lead consultant for New Gold, dated 14 January 2014, and another dated 10 September 2021 under the name of Artemis itself, assisted by Ausenco Engineering Canada Incorporated (“Ausenco”) which in turn relied on specialists from other consultancies for such aspects as metallurgical and processing and environmental inputs.

‍
Background to the Blackwater Project

The Blackwater project is located in British Columbia, approximately 112 km southwest of Vanderhoof and 446 km northeast of Vancouver (refer to Figure 2.2_1).

The project is easily accessible by forest and mine roads with a drive from Vanderhoof taking approximately 2 ½ hours.

The tenement covers an area of 148,902 ha and consists of 329 mineral claims (Figure 3.2_2). The Blackwater claim block itself comprises 76 mineral claims totalling 30,791 ha.

The only NSR royalty that affects the proposed open pit operations is called the Dave Option at a rate of 1.5%.

The surface rights belong to the government and construction of any mine infrastructure does not require any additional land tenures other than mining rights.

The company has consulted the various stakeholders including First Nations and has secured their support for the project. There are agreements in place with First Nations groups, but the technical report does not give details on terms and conditions.

Geology and Mineralisation

There is no to little outcrop over Blackwater deposit with most of the area covered by thick glacial deposits of 2 m or more. Geological interpretation has been based primarily on drill information plotted on section and plan views. The lack of data is evident from the geological discussion, which is far from clear.

There is only one geological map available, which is reproduced in Figure 2.3_1.

The red dashed line shown on the map delineates the outer limits of the pyrite probability shell, which was deemed relevant for the estimation of the mineral resources.  It shows that the deposit is wholly present within volcanoclastic rocks. These rocks have been downthrown by two northeast trending faults: Natalkuz in the north and Blackwater in the south. An example of the confusing geological discussion is that the map shows younger rocks between the faults, whereas the discussion in the report states that “these faults juxtapose older Mesozoic and Tertiary rocks in the central part of the uplift against younger Cretaceous and Tertiary volcanic rocks to the north and south”.

The host rocks within the Blackwater deposit area are described as being pervasively fractured, with introduction of pyrite (FeS₂), and silica (up to 25% of total volume) and sericite, which is a mica mineral.  Brittle style deformation affects all rock units but faults are difficult to recognise and correlate because of the intense fracturing and multiple fault sets.  The fracturing grades laterally into unbroken rock with no obvious bounding fault surfaces.  The fault sets recognised have a dominantly northwest-southeast trend, a northeast trend and an east-northeast trend.

A major north–south-trending fault dissects the orebody and east–northeast-trending faults along the 375,600E coordinate. This fault represents a well-defined disruption in lithology, alteration, and mineralisation patterns and was used to subdivide the resource block model into two structural domains, one to the east of it and one to the west.

Drilling has defined the Blackwater deposit as a zone of continuous disseminated gold-silver mineralisation extending at least 1,300 m east-west and at least 950 m north-south. The vertical thickness of the zone averages 350 m but ranges up to 600 m, remaining open at depth to the southwest, north and northwest.

Gold-silver mineralisation is associated with a variable assemblage of dominantly pyrite – sphalerite (ZnS) – marcasite (a different crystallisation form of pyrite) and pyrrhotite (approximately FeS, but with slightly less Fe) and minor other sulphide minerals. Statistical analysis indicates a strong correlation between pyrite and “dendritic black sulphide” (“DBS”).

Mineral Resources Estimation

The database used for the estimation contains results from 1,041 core holes totalling 317,718 m, which is one more hole than used by New Gold for its latest resource estimation. The holes have been drilled at a nominal drill spacing of 25 m and 50 m. Figure 2.4_1 shows the drill collar location map.

Before discussing the methodology used to estimate the resources it is useful to look at some sections through the deposit with drillhole intersections. Refer to Figure 2.4_2 with the results for the N-S sections 400 m apart with the coordinate traces on Figure 2.4_2.

Apart from a block of mineralisation in the south of cross section 5200 E with good grades that correlate well between a number of adjacent drillholes, gold intercepts in the drillholes correlate poorly to adjacent sections.

Figure 2.4_3 shows the gold intersections along coordinate 2800 N. It is noticeable that the block of high grade in section 5200E is not evident in this section along section 5200E being slightly to the east. It points to the high-grade block having dimensions that have a strong north-south component and limited east-west. Again the author of the technical report is sloppy showing the section as being N-S instead of E-W.

There is a block of ground in the west with good gold grades in two adjacent drillholes, but elsewhere this is not the case.

The above points to difficulty to identify sizeable blocks of ground with consistently high grade. This one should expect to be reflected in the block model used for resource estimation.

For the geological model Artemis referred to the New Gold model to draw up mineralisation domains, but selected only three domains from five domains. Figure 2.4_4 shows the domains selected by Artemis determined by the main N-S fault structure and the norther portion of deposit west of the fault due to a change in the mineralisation orientation.

Because of the close relationship between pyrite and “dendritic black sulphide” (“DBS”) with gold content a “pyrite shell” wireframe was created using a combined content of 0.5 vol% as threshold which served as an outer constraint for gold estimation.

Cumulative probability plot (“CPP) graphs for the three domains show the extreme skewed nature of the grade curves with approximately 85% of the values below 0.5 g/t Au and high coefficient of variation (“CV”), which is calculated by dividing the mean by the standard deviation. As the Artemis technical report states: “the distribution for gold contains inflection points above about the 90% of the data, high grades that are not lognormal, and contains a significant amount of gold metal”.

Generally a CV of 2 or less is recommended for ordinary kriging, but for Blackwater it is between 3.1 and 5.6 after composting of the grades to 2 m intervals.

Within the pyrite shell a grade shell domain was generated by New Gold to better delineate areas of very low-grade material. This grade domain uses a threshold of 0.1 g/t Au for 5 m composites (= the increased length reduces the variability compared to using 2 m composites). New Gold further reduced the variability by capping high values to 45 g/t or 5 g/t depending whether or not the value was within or outside a gold grade shell of 0.1 g/t Au.

Artemis approached the very skew grade curve problem differently by employing Multiple Indicator Kriging (“MIK”) for grade estimation. This method splits the grade population in “bins” of grade intervals and carrying out kriging separately for each bin. This method is used when a company does not want to cap very high grades and reduce their influence. Crux Investor records that this method was also used for Moose River and resulted in a large overestimation of plant feed grade when mining was carried out (as was the case for Brucejack and Sukhari). In the case of Moose River the reason was either an overestimation of the range of influence of high grades, or the overestimation of the selectivity that can practically be achieved when assigning blocks as ore and waste during operations. At Moose River much more waste was mined that forecast, resulting in much lower feed grade and very poor financial performance compared to forecast.

One of the main concerns of Crux Investor relating to the resource estimations of New Gold and Artemis is the complete absence of variograms presented in the report to substantiate that these are robust enough to be used for grade estimation.  It is noticeable that New Gold presents such variograms in a technical report for Rainy River (which are very robust) published in the same year as the Blackwater feasibility study.  It raises the question why these were left out of the Blackwater feasibility study.

Similarly Artemis does not present variograms or the various grade bins.  Even stranger, in tables with the search parameters for grade bins, almost all show a nugget effect of nil.  Elsewhere the report it is recorded that the relative difference of grade values in duplicate pairs and field duplicate pairs is high and partially blamed on the nugget effect.

Table 2.4_1 shows the estimated Measured and Indicated Resources by Artemis for a set of cut-off grades (with 0.2 g/t Au considered the Base Case) and New Gold for a cut-off grade of 0.4 g/t Au.

A comparison of the results at a cut-off grade of 0.4 g/t shows a great similarity and should give much comfort.  However Crux Investor has two major concerns:

• The total absence of variography information supporting the validity of the chosen search parameters.
• Artemis not providing any details on the results of a grade control drilling programme covering a substantial surface area of the deposit and involving 561 reverse circulation (“RV”) holes during 2020/21 and without comparing these to the block model results.  It raises the question whether or not Artemis chose to omit this because the results showed poor correlation.

The Artemis technical report gives cross sections through the deposit, but unfortunately not at consistent coordinates with the sections presented in Figure 2.4_2 and Figure 2.4_3 of this report making direct comparison not possible.

Figure 2.4_5 shows two cross sections extracted from the Mineral Resource estimation section (top) and Mineral Reserve estimation section (bottom).  Whereas these sections are 50 m apart and do not allow for direct comparison, the following is apparent:

• The Mineral Resource section shows considerable smearing of the high-grade blocks (e.g. >5 g/t Au).  Should the range of influence of the high-grade values prove shorter than assumed, this will have a major impact on overall grade.
• The Mineral Reserve section shows a different colour coding for which no legend is provided.  However, the block of ground in red in the deepest portion of the pit is indicated in the Mineral Resource section as exceeding 0.5 g/t but less than 1.0 g/t.  This means that red indicates +0.5 g/t grade.
• There are many instances where the red blocks are shown as being rather isolated and in contact with waste.  
  

Figure 2.4_6 shows another comparison, again suffering from not being on exactly the same coordinate, but being 50 m apart.

The block model cross section shows a large block of ground at coordinate 5200E with relatively good grade (i.e. in red) that is relatively poorly supported by drill data in the section above.

In conclusion, Crux Investor finds that Artemis makes a poor case of substantiating the validity of their grade block model.

Mineral Reserves Estimation

The mineral reserve estimation is based on assuming mining in a conventional open pit drill-blast-haul operation.  The cut-off grade used was based on a net smelter return value of C$13.0/t.  The reason why this was used instead of a gold cut-off grade, was to give due consideration to credits from producing silver.  However, silver constitutes a negligible part and Crux Investor has calculated a gold cut-off grade to allow comparison to the cut-off grade used by New Gold in 2013 arriving at 0.28 g/t Au at an assumed long term gold price of US$1,400/oz.

Artemis assumes “edge dilution of 9.5% at ore to waste contacts.  It is not clear what this means for overall assume dilution.

Pit optimisation did not use the Whittle or Lerchs-Grossmann algorithm, but the Pseudoflow algorithm.  It seems to use undiscounted cash flows for pit shell selection, which would indicate a bias to larger overall pit shells.  The result is that nine phases are considered of which the first phase is a burrow pit to project construction purposes.  Thereafter the higher-grade block at approximately 5200 E was logically chosen (see again Figure 2.4_2 bottom cross section), followed by push backs dominantly in an easterly direction.

The above table shows the great difference in input parameters used to calculate cut-off grade with New Gold including sustaining capital expenditures and a profit margin.  Both studies exclude the cost of mining a block designated ore, which is correct when that block is anyway mined and needs to be allocated as plant feed.  However, in the decision on whether or not to mine that block, the cost of doing so should be included.  Both cut-off grades have therefore an optimistic bias, also because these ignore dilution.

Table 2.5_1 contains the mineral reserve statements for Artemis, effective 10 September 2021 and New Gold, effective

Cells that are highlighted brown have numbers that do not make sense.  For Artemis the higher grade of Proven reserves compared to Measured resources could possibly be explained by using a higher cut-off grade (0.26 g/t Au versus 0.20 g/t Au), but for Probable Resources the upgrade is excessive as the grade of Indicated Resources using a cut-off grade is 0.61 g/t Au.  At the same time a conversion rate of 5% for the tonnage of Indicated Resources is exceedingly low.  The Proven Reserves for the New Gold study contains more gold than the Measured Resources, which does not make sense.

Overall the gold content in both reserve statements is very close to each other, which should give some comfort IF the resource estimations are valid.

Mining

Mining is planned with 10 m high benches, but it may possibly involve 5 m split benches to improve grade control.  To better delineate the resource provision is made to drill with RC rigs (no spacing provided) and assay the chips on 3 m intervals.

Ore mined will either be delivered to the crusher, the run-of-mine (ROM) stockpile located next to the crusher, or the low grade ore stockpile when its grade has a NSR value of between C$13/t and C$27/t (less than approximately 0.6 g/t Au).  This material is planned to be processed with the lowest grade treated last.  This strategy will require much grade control, exacerbated by the need to differentiate between various waste types in terms of potential pollution.  For example assaying of zinc is required to separate material that exceeds a grade of 0.1% Zn.

As the low-grade ore will contain a certain proportion of potentially acid generating (“PAG”) mineralisation, it needs to be placed on a low-permeability base with a drainage collection system.  The drainage will be collected, neutralised with lime at the process plant, and discharged to the TSF to which is added the requirement to distinguish potentially acid generated (“PAG) material and non-acid generated material (“NAG”).

The equipment choice is for relatively large size machinery with excavators used for loading ore using 22 m³ buckets and shovels for waste using 34 m³ buckets.  Haulage will be by rigid body trucks for ore with a 190 t payload capacity and for waste with 230 t payload.  At such sized selectivity is limited and dilution can be expected in particular when extracting isolated blocks of ore.

Metallurgy and Processing

Metallurgical Testwork Conclusions

Artemis relied on the conclusions of metallurgical testwork carried out by New Gold between 2008 and 2013, which discounted flotation and heap leaching as options, to selected whole ore leaching as the optimal process route.

Artemis completed additional testwork in 2019/20 which confirmed the generally hard nature of the mineralisation within a wide range of bond index values of 11.8 to 24.6 kWh/t and that grinding finer than 150 μm has no distinct effect on recovery.

There are however a few major differences between the New Gold and Artemis metallurgical results which the discussion of the 2019/20 testwork of Artemis not clearly substantiates. These are:

• Artemis completely ignoring the importance of the degree of oxidation of the mineralisation on metallurgical performance in the New Gold testwork.  It is estimated that 3% of the mineralisation is oxide and 11% is transitional.
• Whereas New Gold concluded that “coarse gold is not statistically abundant enough to warrant a dedicated processing strategy like gravity concentration”, Artemis has included a gravity concentration step as, according to their results, it improves overall recovery by 2.9 percentage points on one composite, 1.7% percentage points on another composite and 0.9% percentage point on another.  It is not clear what the composite characteristics are, but Table 13-6 in the technical report gives grades for three composites between 1.05 g/t Au and 1.41 g/t Au, which is well above the reserve grade and places their representativeness in doubt.
• The most important difference between the two technical report is the much higher overall recovery suggested by Artemis with recoveries in the mid-nineties compared to New Gold’s suggested recovery of in the mid-eighties.  Note that the inclusion of gravity concentration cannot explain this large difference as according to Artemis this improves recovery by only 1-3 percentage points.
• There is no discussion on the large improvement, but Crux Investor records that New Gold’s work was focused on material grading between 0.2 g/t and 1.2 g/t Au.  However, when referring to data of Artemis variability tests, the sub 0.70 g/t Au samples do not have an obviously lower overall recovery.

In conclusion, Crux Investor has a concern about the major difference in conclusions between the earlier and later testwork results and which are not discussed and substantiated by Artemis.

Chosen Process Route

The run of mine ore is first reduced to 80% passing (“P80“) passing 8 mm in a three-stage crushing plant consisting of a primary gyratory crusher followed by secondary cone and tertiary cone crushers.  The crushed product will be conveyed to a covered crushed ore stockpile with 32,100 t total capacity and 8,190 t live capacity.  Two reclaim belt feeders will feed the mill.  The crushed material will be processed through a dual pinion ball mill in closed circuit with cyclones producing a final product with a P80 of 150 μm.  The mill will be in closed circuits with a cyclone cluster that separated the P80 of 150 μm overflow from the coarser underflow that is returned to the mill.

A portion of the ball mill discharge will feed a scalping screen with the oversize going to two parallel gravity concentrators.  Tailings from the concentrators will be transferred back to the ball mill circuit and the concentrate will gravitate to the intensive cyanidation circuit at ground level for treatment in batches.  The undersize of the scalping screen will be pumped to a leach-adsorption circuit using carbon-in-leach (“CIL”) with the carbon advancing counter current to the slurry flow.  Gold in the loaded carbon will be recovered through first elution by adding NaCN and NaOH and the solution then subjected to electrowinning and smelting to produce doré bars.

The above description shows that the process flow is relatively straightforward and conventional.

Economic Evaluation

Economic Assumptions

The spot price on 30 January 2024 of US$2,039/oz Au and US$22.75/oz Ag was used as the Base Case gold price.

Production Schedule

The feasibility study assumed two years of pre-production with mill commissioning and ramp up is during this time and hitting full plant capacity rate of production of 6.0 million tonnes (“Mt”) in production Year 1.  This rate is maintained until the end of production Year 4, completion of the Phase 2 expansion in Year 5 treating 9 Mt, treatment of 12 Mt until the end of production Year 9, completion of Phase 3 expansion in Year 10 with 15 Mt being treated after which production plateaus at 20 Mt per annum (“Mtpa”).

Early mine production is concentrated on areas with relatively high grade and low waste stripping requirements.  Figure 2.4_2 demonstrated that along coordinate 5200E there is a sizeable block of ground with relatively high grade making this strategy plausible.

Figure 2.8.2_1 graphically shows the production schedule with the graph along the top.

Total material mined shows a rising trend until year 10 just when maximum plant capacity is reached.  In order to high-grade plant feed grade there is a net addition of material to the low-grade stockpile which stabilises at a level of around 115 Mt during the period Year 10 – Year 16 after which is the main and from Year 18 only source of plant feed.

The average waste strip ratio over the life of mine (“LOM”) is 2.0, but averages 1.74 during the first five production years.  This in combination with a feed grade that ranges between 1.56 g/t Au and 1.66 g/t Au should give excellent cash flow.

Capital Expenditure

Table 2.8.3_1 shows the overall estimates for initial construction, the two expansion phases, sustaining capital expenditure and deferred (= not defined what this entails) capital expenditure.

The discussion on capital expenditure is full of gaps and difficult to audit.  For example, there is a breakdown of the total initial expenditure on Mining (58% pre-stripping) showing negligible outlay on mining equipment, and this is assumed to be financed through a lease, but no such breakdown for the subsequent expansions.

None of the other tables with breakdown of certain main categories and with expenditure in different categories can be reconciled with the main headings in Table 2.8.3_1.  The capital expenditure estimation is a complete black box.  The sum of the provisions for Crushing and Reclaim and Process Plant is C$228 million, clearly lower than the C$318 million EPC contract price with Sedgman discussed in Section 1.  This amount however still looks very low at US$474/monthly tonne capacity.  In this respect it is noticeable that Artemis had to issue a press release in June 2023 declaring that amendments to the scope of work for the plant would increase cost by C$50 million.  As the total is still well below capital cost for comparable similar sized plants, investors should expect further overruns.

The lack of detail is worse in the cash flow model (which is of a very poor standard) with 12 row items of which two are reserved for “Capital” and “Sustaining Capital”, which total over the LOM respectively C$1,418 million and C$963 million, again not reconciling with the table above.  As C$140 million is forecast being spent post production cessation, this amount is supposedly associated with closure and rehabilitation expenses.  How the stock exchange authorities let companies get away with such poor reporting is puzzling to Crux Investor.

The overall impression is that the provisions are too low.  The feasibility study of New Gold provided in 2014 for C$1.96 billion initial capital expenditure to get to a treatment rate of 21.9 Mtpa.  At the time the Canadian Dollar was much stronger at C$0.95 per US$ than in 2021 when it was C$0.79 per US Dollar.  Whereas the estimated project capital investment was estimated by New Gold to be C$90/t annual treated, Artemis provision is C$71/t annual treatment for an operation that processes 20 Mtpa.

Moreover, Detour Gold with a processing rate of 22 Mtpa had average sustaining capital expenditure of US$230 million per annum in the period 2018-2022 (there no numbers available for 2021 due to the takeover by Agnico Eagle), which amounts to more than US$10/t treated.  In comparison Artemis assumed annual sustaining capital expenditure at C$2.45/t when reaching treatment rates of 20 Mtpa.

The technical report gives no detail about the estimation of C$175 million for closure and rehabilitation cost.  As related expenses are supposedly captured by Bonding costs, the item is not included in the capital expenditure table.

According to the latest construction update published on 30 January 2024 overall construction is 59% complete and C$389 million of the “guided initial” capital expenditure of C$730 to C$750 million (note this is 15 -16% higher than in Table 2.8.3_1) had been spent and expenditure tracked “a typical S-curve”.  The update indicates that key mechanical equipment packages are on-site.  The risk for overall capital expenditure is therefore mostly related to indirect costs.

Based on the considerations discussed above, Crux Investor has amended the capital expenditure stream by adding 25% to annual provisions, which should still be considered very optimistic.

Operating Cost Estimate

Table 2.8.4_1 shows the forecast operating expenses.

The unit cost for mining of C$2.60/t is low for Canadian operations as a benchmarking exercise by Crux Investor has shown such cost to be typically above this level.  Figure 2.8.4_1 has been extracted from an Analyst’s Note dated May 2021 showing unit cost in US Dollar.

The above diagrams imply that the unit mining cost is probably approximately US$2.5/t, which converts to C$3.35/t.  There are no obvious reasons for a very low unit mining cost for Blackwater as the operation requires much grade control activities and the mining equipment is being leased. The capital expenditure in the feasibility study is not clear on where the leasing cost is accounted for, but the cash flow model does not reflect it anywhere else.

Similarly, there is no reason for the unit processing cost to be low.  The material is hard requiring considerable energy to be broken down to the target particle size, the reagent consumption is not particularly low and includes pre-aeration with oxygen and addition of sulphur dioxide and copper sulphate.  Based on the above diagram the unit processing cost for initial production is indicated at C$10.7/t, for Phase 2 C$9.15/t and for Phase 3 C$7.80/t in 2020 money terms.

The G&A provision of between C$27 million per annum and C$38 million per annum for the period during which mining activities occur.  With reference to a technical report by Detour dated 22 March 2017 the annual G&A cost for an operation treating between 22 Mtpa and 23 Mtpa was C$64 million dropping to C$55 million per annum, the drop being very aspirational.

The above rates are based on forecast rates in various technical reports.  As the Analyst’s Notes pointed out the forecasts proved almost invariable low when these could be checked against actual costs.  Figure 2.8.4_2 illustrates this.

The relationship shows actual operating cost being on average slightly higher (almost 9%), but with (Opex + Capex) per tonne milled 28% higher than forecast.

The implication of the above discussion is that the cost rates are low compared to forecast for other open pit projects, which themselves proved to be too low.  Crux Investor is of the opinion that overall operating cost is probably at least 20% higher.

Working Capital

The cash flow model (with 12 rows) includes an item “Working Capital and Bonding” but with the report giving no information whatsoever on how the amounts are arrived at.  In total the cash outflow over the LOM is C$111 million, which must be the bonding expenses as working capital is highly likely assumed to be recovered in full at the end of LOM.

Referring to the feasibility study cash flow model the investments in working capital in Year-1 and Year 1 is approximately C$20 million.  For comparison reference was again made to Detour Gold which had total net current assets at the end of 2018 of C$54 million, ignoring cash.  When accounting for the difference in operational capacity the provisions of Artemis are in the same ball park and Crux Investor has therefore adopted the annual provisions in the feasibility cash flow model.

Royalties and Taxes

With reference to Section 2.2 there is a royalty of 1.5% affecting the surface area over the proposed open pit operations.

As usual the feasibility study report gives almost no information on estimation of taxation apart from mentioning the statutory tax rates.  This valuation has performed an estimation of taxes based on information in a note by PWC entitled Canadian Mining Taxation – 2016.

Applicable taxes for mining companies in British Columbia are:

• British Mining Tax – two tier system based on “net revenue”
• Federal Corporate Income Tax at 15.0%
• British Columbia General Corporate Tax another 12.0%.

The British Columbian Mining Tax is levied in two stages: a 2% tax on “net current proceeds” and 13% tax on “net revenue”.  The net current proceeds tax is a form of minimum tax, which is deducted in full, with an interest component, against the 13% net revenue tax.

The net current proceeds are defined as gross revenue (including forward sales, but ignoring hedging gains and losses) minus operating expenses and post-production development cost.  With the exclusion of deductions for financing cost, capital expenditure and pre-production expenses, the effect is that taxes will be payable as from start of production.

The net proceeds tax is creditable against the 13% net revenue tax with a notional interest of 125% of the prevailing federal bank rate (currently 3.75%).

The 13% net revenue tax base is derived by deduction from net proceeds sustaining capital expenditure, exploration costs and pre-production development costs.

With respect to Corporate Income Taxes, the tax base for Federal and British Columbian Provincial taxes should be calculated on the same basis: Earnings before Tax and Depreciation minus British Columbian Mining Taxes.

Interest is allowed as a deduction as well as amortization/depreciation which has the following components:

• Canadian Capital Allowance (“CCA”) – covering capital assets purchased, distinguished in assets purchased before start of commercial operation. These are allowed at 25% per annum on a declining balance basis.  In recognition to the fact that assets are acquired throughout the year, only half the additions are eligible in the base to which the CCA is applied.
• Canadian Development Expense (“CDE”) - the acquisition cost of a Canadian mining property and associated mine development expenses, both initial and sustaining capital expenditure (exploration, studies, shaft, underground development) – 30% pa.

The feasibility study document does not give information of available balances for CCA and CDE.  Crux Investor has used the cumulative annual totals for Exploration and Evaluation Expenses of C$381 million for CDE deduction and the carrying value of Capital Assets of C$57 million for CCA deduction and the Deferred Tax Asset of C$33.2 million, all effective at 31 December 2022.

Financing

As part of financing the Blackwater project Artemis early on entered into metals stream arrangements.  The first was a Gold Stream Arrangement with New Gold as part of the acquisition of its project in August 2020.  New Gold is entitled to purchase 8.0% of the refined gold produced for 35% of the spot gold price until 279,908 ounces of refined gold have been delivered to New Gold, after which the gold stream will reduce to 4.0%.  In December 2021 concurrent with the conclusion of the silver stream agreement, discussed in the next paragraph, Wheaton Precious Metals Corp. (“Wheaton”) purchased the stream from New Gold.  In June 2023 the agreement was amended to add another 92,000 ounces deliverable after 2034 in return for C$50 million required to cover the additional costs for the processing plant.

In December 2021 Artemis entered into a Silver Stream Agreement with Wheaton to deliver 50% of the produced silver until 18 million ounces have been delivered, after which the delivery rate drops to 33% of production.  For this Artemis received US$141 million.  Artemis will receive 18% of the spot price until the up-front payment has been reduced to zero after which payment increases to 22% of spot.

In February 2022 reached in principle an agreement on the provision of construction project financing with a group of financiers.  This was finalised one year later on 1 March 2023 for C$360 million plus C$25 million capitalised interest plus a C$40 million Standby cost overrun facility.

To secure a minimum revenue stream Artemis has entered into gold hedging programmes for 190,000 ounces starting March 2025 and until December 2027 with a weighted average sales price of C$2,836/oz.

This valuation has modelled the impact of the streams and hedging agreements on the cash flow, but has, as a simplification, accounted for the loan financing in the derivation of the Enterprise Value.

Results

Table 2.8.8_1 shows the forecast financial performance over the LOM, both as per feasibility study at feasibility metal prices and at spot prices, and as per this valuation.  The feasibility study case at feasibility study metal prices has been modelled to check the tax model against what the feasibility study gives for total taxes.  Crux Investor taxes are 1.3% higher, but this is partially explained by Crux Investor modelling C$301 million higher net revenue than the technical report.  When accounting for this at the effective tax rate, the Crux Investor number is 2% lower than as per Feasibility Study.  Crux Investor therefore considers its tax model to be valid and accurate.

Apart from net revenue, Crux investor’s model has the same numbers as per the 12 row feasibility study cash flow “model”, which results in cash flow attributable to shareholders being C$204 million higher for Crux Investor and the NPV5 at C$2,190 million is higher by C$49 million.

When using the feasibility study assumptions, but at current metal spot prices, the net free cash flow increases to C$7.7 billion with a NPV₈ (using an appropriate discount rate for projects) of US$2.79 billion.

Using the input parameters of Crux Investor set out above, the revenue drops due to the impact of the Silver Stream not accounted for in the feasibility study.  Despite this, there is an impressive increase of more than 30% compared to the feasibility study.  Other change in the Crux Investor cash flow modelling is to ignore 2023 for discounting purposes as development has progressed one year further and positive cash flows are one year less out.  Additionally, the cash outflows associated with pre-production capital expenditure have been ignored as fully covered by available funding.  The servicing of loan and lease finance have not been modelled, but is accounted for in the calculation of the Enterprise Value in Section 3.1.

With the above, despite stressing the model by incorporating amendments to the operating and capital cost of respectively 20% and 25% the net free cash flow is substantially higher at C$6,835 million compared to C$4,442 million for the feasibility study.  The amendment does affect the pre-production capex provision, adding C$161 million.

Table 2.8.8_2 shows the sensitivity of the net present values (“NPV’s”) for various discount rates to the main input parameters for the Crux Investor case.

After substantially adjusting the operating cost and capital expenditure there is much reduced risk that these metrices will fall short.  The major remaining risk associated with the project is the feed grade having been overestimated because of an erroneous resource estimation and the large assumed improved in metallurgical recovery.  The gold price is a rough proxy for these as price, grade and recovery will affect the revenue.  For every percentage point variance in grade the NPV₈ would vary by almost C$49 million.  A 46% shortfall (i.e. a LOM grade of 0.40 g/t  Au) would wipe out the NPV₈.

Valuation of Artemis Gold Incorporated

The Enterprise Value of Artemis Gold at 26 January 2024

At the share price of C$6.30 on 30 January 2024 the market capitalisation of Artemis is C$1,251 million, or US$934 million.

At the end of September 2023 the company had 11.6 million options outstanding, with an average exercise price of C$4.76.  An estimated 6.4 million options are in the money at the current share price.  At the end of September 2023, 26.2 million warrants were outstanding as well, which based on the Annual Financial Statements for the year ending 31 December 2022 have an exercise price of C$1.08 and therefore well in the money.

The cash balance and other net current assets have been ignored as fully committed to the project.

According to the construction update press release dated 30 January 2024, remaining capital expenditure was at most C$361 million, covered by C$392 million funding without overrun facility, being cash on hand of C$157 million, and remaining drawdowns of C$235 million the total loan facility of C$385 million.  Crux Investor has therefore included for the purposes of calculating the Enterprise Value as debt: C$361 – C$157 = C$204 million + C$150 million already drawn from the debt facility = C$354 million.

Based on the above, the diluted Enterprise Value for Artemis is C$1,840 million, equivalent to US$1,370 million as derived in Table 3.1_1.

The diluted Enterprise value of C$1,888 million is equal to 69% of the NPV₈ indicating that Artemis is good value at the current share price.

Corporate Cash Flow

Figure 3.2_1 shows the forecast cash flow generated over the LOM.

The above picture is actually flattering as Crux Investor did not model debt servicing and repayment.

The graph shows distinct peaks and valley caused by the large capital expenditures for Phase 1 (year 2029) and Phase 2 (year 2034) expansions. The negative cash flows during 2024 is the result of amending capital outlays by 25%, not covered by funding before using the overrun facility.

Noticeable is the excellent cash generation in Year 1- Year 3 at approximately C$370 million per annum.  Given that during this period a high-grade block of ground is mined, which is well supported by drill intersections, the financial performance during this period is considered of lower risk than later during the LOM.