Description

We have reviewed Hindenburg's September 17, 2019 report, Credit Suisses's September 20th report, Morgan Stanley's September 18th report and Bloom Management's comments. Our assertion is that the equity is worthless and that bankruptcy is a very real scenario for several reasons that we detail below.

However, before we proceed any further, an introduction to the Company is in order for those readers not familiair with the Company.

Introduction:

Fuel cell technology has been the next big thing since it was invented in the early 19^th century, however we’re now 180 years on and not a single fuel cell company has ever made a profit. Bloom Energy has been the fuel cell company that has generated the most buzz in Silicon Valley in recent years. The founder and CEO of Bloom, KR Sridhar had originally invented the technology for NASA to power life on Mars but after NASA folded the program Sridhar founded Bloom Energy. By 2010 Bloom Energy was generating considerable buzz in Silicon Valley, having raised hundreds of millions of dollars, but still had not yet released a product.

Bloom Energy was also one of the very first Kleiner investments that was part of Kleiner’s infamous cleantech portfolio that infamously nearly ruined the VC firm and included notable dud investments such as Miasole and Fisker Automotive

Bloom’s big unveiling came in 2010 when 60 Minutes did a puff piece on Bloom Energy, starting the program with “in a world of energy, the holy grail is a power sources that’s inexpensive and clean, with no emissions.” At the unveiling a few days prior to the 60 Minutes piece, Sridhar presented the technology to a star-studded room that included Bloom board member, Gen. Colin Powell, former Secretary of State and Theranos board member George Shultz, California governor Arnold Schwazenegger and CEOs from Google, eBay, Walmart, FedEX & Coca-Cola. Senator Dianne Feinstein stated via video, “this technology is going to fundamentally change the world, Bloom boxes not only can help consumers reduce dependence on the electric grid here at home but can also bring modern energy to the planets most remote villages.”

With the star-studded unveiling Bloom Energy was hyped as the holy grail of clean & distributed energy. Today, nine years after the star-studded unveiling, Bloom has managed to raise billions of dollars in VC funding and Bloom and Bloom’s customers have received over a billion dollars in government clean energy subsidies. The only problem is that today, Bloom’s technology does not work as promised. Bloom fuel cells are NOT clean, Bloom energy is NOT cheap, and Bloom is losing hundreds of millions of dollars a year with signs of weakening demand.

It’s hard not to draw some parallels between Bloom and Theranos. Both were highly secretive Silicon Valley companies that were “changing the world”, backed by a charismatic leader and had a star-studded roster of leaders and customers who supported them.

We believe Bloom has been incredibly misleading about their technology and financial prospects. In 2018, the WSJ reported that two founders of Advanced Equity, a defunct brokerage firm, who in 2012 were charged and barred by the SEC for misstating and exaggerating Bloom’s financials and technology to Bloom investors, claimed that it was in fact Bloom representatives who were the source of the false information.

With the introduction provided, we will review the work by Hindenburg and provide a factual rebuttal to the narrative Bloom management is putting forward.

1.       Fuel cell life

·       Credit Suisse went through the California data and determined that the beginning of life efficiency in California was 63% and the that the efficiency degrades by ~4% per year. This is same as our findings. However, we believe fuel cells are retired at around a 50% efficiency and not a 45% efficiency.

·       Using a 50% efficiency retirement we calculate the average life of Bloom fuel cells to be 38 months.

·       We reviewed the entire California & NY data sets which combine to represent approximately 250 Bloom projects, 96MW or 25% of Blooms installed base. The California data set includes approximately 42% of Bloom’s total acceptances from the 2015 vintage.

·       We are unable to find any evidence of improvement in the useful life of the fuel cells from the data sets. The 2015 vintage appears to have the same useful life as the 2018 vintage.

2)      Service liability

·       We recalculate Bloom’s service liability using our own estimates and a slightly different methodology than Hindenburg. We arrive at a $1.1b to $2.9b estimate for future services liability.

·       Morgan Stanley asserts that Bloom offers their customers immediate cost savings of 10% to 15% and that Bloom could “change the pricing structure of its service agreements going forward to match the assumed product lives in order to avoid the large liability”. We estimate that if Bloom was to fully price the full cost of servicing and replacing fuel sells in their service agreements the delivered cost of power would be $.15c to $.20c per kwh which is significantly more expensive than the grid. 

3)      Environmental claims

·       Bloom is not the “cleanest power source from natural gas” as Sridhar claimed on the Morgan Stanley analyst call.

·       Politicians in California and New Jersey are now realizing that Bloom has misrepresented their clean energy claims and technology.

4)      Morgan Stanley asserts that Bloom customer feedback is positive: A review of Bloom’s largest customer AT&T

After reviewing the details from Bloom’s largest customer, AT&T, we believe that the entirety of this revenue in 2018 was for replacement units. Bloom has never disclosed this.

5)      Appendix A

·       Why Bloom’s marginal emissions argument doesn’t make sense.

1.   Fuel cell life

·       Credit Suisse went through the California data and determined that the beginning of life efficiency in California was 63% and the that the efficiency degrades by ~4% per year. This is similar to our findings.

·       However, Credit Suisse concluded that fuel cells in California last 4.6 years by assuming, based on Bloom’s management comments, that fuel cells are retired at 45% efficiency.

·       We believe Bloom retires fuel cells between a 48% efficiency and a 52% efficiency, not a 45% efficiency like Bloom claimed.

·       Because Bloom retires at ~50% efficiency the useful life of fuel cells is only 38 months.

New York State

·       To be eligible for the NYSERDA program in NY the eligibility[1] clearly states a total system efficiency of 45% or greater on higher heating value. All the Bloom data in California and New York is in lower heating value. To convert[2] the higher heating value to lower heating value you need to do this math 45%*(1030/930) = 50%. Fuel cells in New York need to be maintained at a 50% efficiency.

·       Here is the data from all 19 projects in New York. As you can see most projects are replaced in the 48% to 50% range.

·       The longest line in grey, is a Bloom fuel cell from Morgan Stanley’s office in NY. You can see the efficiency spike back up at around the 24 months mark. Because efficiency declines on a linear basis, any spike in the improvement of the efficiency indicates that Bloom has replaced a portion of the fuel cells for the project in order to get overall project efficiency back above the required rate.

Delaware

·       From the data[3] in Delaware we can see that Bloom is maintaining an efficiency around 49%. We also believe Bloom energy misrepresented their original application in Delaware. The original application stated “a maximum natural gas usage of 1.32 MMBtu/hr” [4]. This is equivalent to Bloom saying the lowest efficiency will be 57.2%.

·       Hindenburg in their report erroneously stated that the fuel cells in Delaware are 7 years old. This is not true. Bloom has been replacing fuel cells in Delaware since 2015.

Efficiency warranty

·       Search Bloom’s financials and prospectuses for the term “efficiency warranty”. The majority of Bloom projects have an efficiency guarantee or an efficiency warranty for a minimum level of efficiency.

·       From the Bloom S1: “Our performance guarantees are negotiated on a case-by-case basis for projects deployed through the traditional lease program, but we typically provide an output guaranty of 95% measured annually and an efficiency guaranty of 52% measured cumulatively from the date the applicable Energy Server(s) are commissioned.”

2.       Looking through the NY and California data we did not find any material improvement to the life of the fuel cell from the 2015 vintage through to the 2018 vintage.

·       We reviewed the entire California & NY data sets which combine to represent approximately 250 Bloom projects, 96MW or 25% of Blooms installed base. The California data set includes approximately 42% of Bloom’s total acceptances from the 2015 vintage.

·       When we study the data from California and NY we do not find any material improvements to the beginning of life efficiency (BOL) from 2014 to 2018. The beginning of life efficiency is the efficiency of the fuel cell project on day 1. Below we present the BOL efficiency data from California & NY.

·       In the NY data set we do not see any improvement in beginning of life efficiency from 2014 to 2018.

·       In the California data set we did see improvement in the beginning of life efficiency from 2012 to 2015 but no material improvement from 2015 to 2018. Average BOL efficiency in the California data set was 60.9%, 62.7%, 62.6% and 62.8% in 2015, 2016, 2017 & 2018. There has not been any improvement in the BOL efficiency in California since before 2016.

·       When we study the data from California and NY we do not find any improvements in the rate of efficiency deterioration in the Bloom Fuel Cells.

·       The charts below show the slope of the lines from NY & California data sets for each individual Bloom project. This is the slope of the line for the first year of operation. 

·       The degradation rate of efficiency (slope of the line) is the same for the 2018 vintage as for the 2012 and 2013 vintages. Said differently, we can’t find any evidence of Bloom extending the useful life of their fuel cells.

·       The beginning of life efficiency for Bloom’s 2018 vintage is the same as Bloom’s 2015 vintage

·       Bloom’s newer vintage of fuel cells do not appear to last longer than older vintages

We can sanity check this with a little back of the envelope math

·       In an August 18th, 2018 press release, Bloom claimed that they had just manufactured their “one-millionth” stack or 1,000th MW. Total installed capacity of Bloom fuel cells as at September 30th 2018 was 348 MW. Therefore, approximately 650 MW of fuel cells had been manufactured that didn’t go towards the total install base metric. It means that Bloom has replaced 650 MW of fuel cells at some point. Given that the 2017 and 2018 vintages had likely not been replaced by September 30th, 2018 most of that 650 MW of replacement cells likely went to replacing vintages from 2016 and earlier. Given that the cumulative installed base of Bloom fuel cells at December 31, 2016 was 235 MW, this implies that the fuel cells from the 2016 vintage and earlier have turned over close to 3 times on average which implies a fuel cell life of 2016 vintages and earlier of much shorter than the 4.6 years management claims.

3.   Service liability

We recalculate Bloom’s service liability using our own estimates and a slightly different methodology than Hindenburg. We arrive at a $1.1b to $2.9b estimate for future service liability.

·       There are two differences between our estimate and Hindenburg.

1)      We don’t include a $1.4b liability for server replacement cost. While Bloom has been replacing full servers, we will give Bloom the benefit of the doubt that going forward Bloom will be able to replace just the fuel cells on a modular basis without having to replace the entire server.

2)      Instead of using a $892 estimate per KW for fuel cell replacement cost we use a range of between $1,400 per kw to $2,500 per kw to estimate the fuel cell replacement cost.

·       We believe Hindenburg’s estimate of $892 inadvertently confused the stack with the fuel cell. From a Bloom Energy presentation, we can see that the fuel cell is 80% of the overall product cost and the stack is only 11% of the product cost. 

·       Last quarter Bloom’s total product cost was $3,045 per kw. 80% of $3,045 is $2,436 per kw. This is the cost of 1kw of fuel cell raw material and components. Bloom will be able to reuse and recycle some components so this cost will be a bit lower for a replacement cell, however offsetting this reduced expense will be cost to repair, refurbish and uninstall and install the replacement fuel cell. At $2,500 per kw replacement cost, we estimate the net future service liability to Bloom would be approximately $3.1b. Credit Suisse in their June 3, 2019 note estimated that replacement cost was $1,400 per kw. We think $1,400 is likely a very ambitious target and we believe the replacement cost for Bloom is closer to $2,500 per kw and therefore the liability is closer to $2.9b. For the sake of transparency, we’ve shown both. 

·       Even if we assume Credit Suisse’s 4.6 year estimate is correct (55 months) and we assume a $2,500 per kw replacement cost we still calculate a $1.6b future service liability (scenario 3).

Morgan Stanley asserts that Bloom offers their customers immediate cost savings of 10% to 15% and that Bloom could “change the pricing structure of its service agreements going forward to match the assumed product lives in order to avoid the large liability”. We estimate that if Bloom was to fully price the full cost of servicing and replacing fuel cells in their service agreements the delivered cost of power would be $.15c to $.20c per kwh which is significantly more expensive than the grid. 

·       It stands to reason that if you must replace the fuel cell every 38 months, which is 80% of the overall cost that it’s going to be very difficult for fuel cells to be cost competitive with the grid.

·       In Bloom’s IPO roadshow, Bloom shows an all-in delivered cost of power of $11.4c per kwh with service cost representing only $2.2c of the $11.4c per kwh. If we assume a 36-month fuel cell life and either a $1,400 or a $2,500 fuel cell replacement cost we get to a service cost of $.06c and $.10c per kwh, which increases Bloom’s all in delivered cost of power to $.15c and $.20c respectively depending on your estimate of cell replacement cost. This $.15c and $.20 is significantly higher than what Bloom showed in their IPO roadshow.

·       So, to Morgan Stanley’s point, Bloom can offer their customers immediate cost savings because Bloom is eating the cost of replacing fuel cells. If Bloom were to fully price the cost of replacing old fuel cells, we do not believe Bloom would be cost competitive with the grid.

·       At $.15c per kwh Bloom is uncompetitive in almost every state in the US.

4.   Environmental claims

Bloom is not the “cleanest power source from natural gas” as Sridhar claimed on the Morgan Stanley analyst call.

·       On the MS call Sridhar claimed that in when it comes to having to the use natural gas for energy, Bloom is the cleanest and lowest carbon footprint power source from natural gas. This is simply not true.

·       Here is a list of modern combined cycle natural gas plants in California with the lbs of CO2 emission per MWH on the far-right column. [5]

·       Bloom at 850 lbs of CO2 per MWH Bloom would only be the 8th cleanest natural gas plant in California.

·       There is a direct relationship between efficiency and CO2 emissions. If we assume that the efficiency of a Bloom fuel cell averages 52% over its life that implies that the Bloom fuel cell emits on average 850 lbs of CO2 per MWH over its life.

·       If Bloom was retiring fuel cells at 45% efficiency which they claim that means Bloom fuel cells would be emitting over 1,000lbs of CO2 into the air per mwh. This would make Bloom one of the dirtiest combined cycle natural gas plants in California

·       Not only does Bloom emit more CO2 than these plants in California but Bloom’s levelized cost of energy is multiples of the cost of energy that these cleaner natural gas plants produce[6]. Furthermore, these natural gas plants don’t receive hundreds of millions in California rate payer subsidies like Bloom does.

Politicians in California and New Jersey are now realizing that Bloom has misrepresented their clean energy claims and technology

New Jersey Clean Energy Program

·       In 2016 Bloom was removed from eligibility for subsidies from New Jersey’s clean energy program. Prior to Bloom being removed from eligibility for the clean energy program in New Jersey we found some of Bloom’s NJCEP applications[7][8].

·       It’s easy to understand why Bloom was removed from the NJCEP program. Look how horrible these payback periods were. Even worse, these payback periods were based off an efficiency and capacity of 56% and 100% which we know Bloom doesn’t achieve. It appears Bloom produced inaccurate information in their NJCEP applications to receive government subsidies.

·       Two commissioners on the NJ Public Utilities Board were aware that Bloom produced inaccurate information on their application about their capacity factor.  These two commissioners voted “no” on giving Bloom the NJCEP financial incentives, however they lost the approval vote 3-2. The two commissioners wrote the following note[9]:

                     “These projects are simply not the best use of CHP/FC program funds…”

“Overall efficiency is about 56%. The efficiency would be substantially higher if the waste was recovered and used. A typical CHP facility would have a system efficiency upward of 75%.”

“Second, the projects have no black start capability. Because they rely on the grid to operate, they would be non-operational during the system blackout. The is contrary to the Energy Master Plan… “

“Another negative aspect to these projects is the lengthy payback period. The estimated equipment life of the FC equipment is 15 years, yet the payback period is… 16 – 21 years without incentives.”

“I believe there are other projects in the program queue that have more favorable economics than those presented here. Good business practice dictates that the rate payers in NJ… get the best value for their dollar and we as a Board, are entrusted with that fiduciary responsibility.”

“Lastly, I was disappointed with the quality of the technical information submitted by the applicant. Other than the project application forms my staff received only one “project snapshot” document and found an error on it, most significantly the estimated annual generation which was about three times more than is possible.”

“The projects’ applicant also paints an overly optimistic picture, to say the least, of the fuel cell’s capacity factor. The projected energy output is calculated at 100 percent capacity factor. While fuel cells generally run full out, a more appropriate capacity factor figure would be a 91 – 95 percent. This adjustment would not result in a significant change in project economics, but in my opinion, would be more accurate.”

California SGIP Program

·       From 2004 to 2015 Bloom received $400m from California’s Self-Generation Incentive Program (SGIP), which provides subsidies and incentives to new and emerging clean distributed energy resources. Bloom was heavily reliant on SGIP program from 2014 to 2017.

[10]

·       In 2014 and 2015 Bloom and Tesla both reached the 40% individual manufacturer cap. Together Bloom and Tesla represented 80% of SGIP rebate dollars. From 2007 to 2015 Bloom projects received $400m out of the total of $1.4b in subsidies, or 27% of the total available[11].

·       In 2015 California Public Utilities Commission issued a report on a proposal to modify the SGIP program. In the report CPUC stated the following[12]:

           “Fuel cells fail, by a small margin to be cleaner than the grid…”

“It is expected that in 2020 the total societal cost of fuel cells and natural gas – and directed biogas-fired microturbines will greatly exceed their benefits – fuel cells because of high cost of capital…”

·       After the CPUC report, the SGIP program was modified to increase the CO2 requirements to be eligible for SGIP credits and fuel cells were required to use biogas. As a result, Bloom has not been approved for a new project with SGIP credits since February 2016[13].

5.   Morgan Stanley asserts that Bloom customer feedback is positive: A review of Bloom’s largest customer AT&T

·       After reviewing the details from Bloom’s largest customer, AT&T, we concluded that the entirety of this revenue in 2018 was for replacement units. Bloom has never disclosed this.

·       We think it makes perfect sense that feedback from Bloom’s customers would be positive. This is because it is Bloom and not Bloom’s customers that are on the hook for service cost.

·       AT&T is Bloom’s largest customer with installed capacity of 47MW. In 2018, 11% ($81m) of Bloom’s revenue came from AT&T[14]. We believe that all $81m of the revenue from AT&T in 2018 was for replacement fuel cells and not new projects.

From the July 2018 prospectus:

“AT&T has over 85 facilities where Bloom systems are contracted or deployed, representing approximately 47 MWs across the U.S.”

·       AT&T’s website corroborates this 46 MW installed as at the end of 2017:[15]

·       AT&T’s website today[16] shows the same installed capacity of Bloom Energy servers at the end of 2018 as they had at the end of 2017, 46.2MW.

·       Therefore, AT&T did not grow their installed MW capacity with Bloom Energy in 2018. Yet, AT&T was 11%, or $81m of Bloom’s revenue in 2018. We believe this revenue was entirely from selling replacement fuel cells back to existing AT&T projects.  Assuming an ASP of $5,000 per kw, $81m implies that approximately 16 MW were replaced, or approximately 35% of AT&T’s installed base.

·       Furthermore, this has the accounting benefit of shifting service cost to COGS. If we assume a TISC of $3,000 per kw this implies servicing cost of approximately $50m that was shifted from servicing cost to COGS. Had this been included in servicing costs, servicing costs in 2018 would have been similar to the level of servicing costs in 2016.

6.       Appendix A

Why Bloom’s marginal emissions argument doesn’t make sense.

·       Here is why Bloom’s marginal emission argument doesn’t work. In California 68.7% of California’s power needs come from combined cycle power plants. These combined cycle power plants are highly efficient, and many are cleaner than Bloom Energy. Despite being extremely clean these combined cycle power plants only had a capacity factor of 50.5%.

 [17]

For example, here is the source of California’s energy on July 8, 2019.

[18]

And here are the CO2 emissions on July 8, 2019.

·       On July 8th, 66% of California’s energy needs came from renewables, hydro and nuclear, and average emissions in California on July 8th was 339lbs of CO2 per MWH. At 14:00 when renewable was at its peak the grid was producing 156lbs of CO2 per MWH.

·       Bloom has said when doing this CO2 analysis that we can’t look at average emissions we need to look at marginal emissions. But what Bloom conveniently leaves out of this analysis is that California has 19,700 MW of capacity of combined-cycle natural gas plants, of which many are cleaner than Bloom. On July 8th, despite having 19,700 MW capacity of combined-cycle natural gas power plants, California was only using a tiny fraction of their capacity.

·       Here is the output from Palomar Energy, a natural gas plant in California that is cleaner than Bloom Energy and emits only about 819 lbs of CO2 into the air per mwh. As you can see from the chart below Palomar is rarely ever operating at full capacity, yet it is cleaner than Bloom Energy.

[19]

·       So on a hot summer day in California, Bloom is displacing power from other baseload power sources like Palomar Energy and from renewables like solar and hydro. On July 8th, Bloom was emitting 850 lbs of CO2 per MWH into California when California already had more than enough power from renewable sources. On this day at the peak, Bloom was 5x dirtier than the grid.

·       The other question politicians in California should be asking themselves is why California has given Bloom Energy hundreds of millions of dollars in state subsidies when natural gas power plants like Palomar are cleaner and cheaper than Bloom.

[1] https://www.nyserda.ny.gov/All-Programs/Programs/Stationary-Fuel-Cell-Program

[2] 1030 is the rate of 1030btu per cubic foot sold by the utility to the customer. 930 is the amount of btu per cubic foot with steam capture.

[3] https://delafile.delaware.gov/CaseManagement/ViewDocket.aspx

[4] http://www.dnrec.delaware.gov/Admin/CZA/Lists/Coastal%20Zone%20Act%20Application%20Status/Attachments/16/CZA%20Permit%20Application%20-%20Nov%2016%20complete.pdf

[5] http://docs.cpuc.ca.gov/PublishedDocs/Efile/G000/M191/K497/191497887.PDF

[6] https://www.lazard.com/perspective/levelized-cost-of-energy-2017/

[7] https://www.bpu.state.nj.us/bpu/pdf/boardorders/2015/20150722/7-22-15-8G.pdf

[8] https://www.state.nj.us/bpu/pdf/boardorders/2016/20160127/1-27-16-8B.pdf

[9] https://www.state.nj.us/bpu/pdf/boardorders/2016/20160127/1-27-16-8B.pdf

[10] Bloom financials

[11] https://www.greentechmedia.com/articles/read/californias-sgip-or-how-not-to-structure-an-energy-incentive#gs.pmppei

[12] http://docs.cpuc.ca.gov/PublishedDocs/Efile/G000/M156/K013/156013203.PDF

[13] https://www.selfgenca.com/documents/reports/statewide_projects

[14] Prospectus & 10K

[15] https://web.archive.org/web/20180703111908/http://about.att.com/content/csr/home/issue-brief-builder/environment/energy-management.html

[16] https://about.att.com/csr/home/issue-brief-builder/environment/energy-management.html

[17] https://ww2.energy.ca.gov/2017publications/CEC-200-2017-003/CEC-200-2017-003.pdf

d[18] http://www.caiso.com/Pages/default.aspx

[19] https://www.eia.gov/electricity/data/browser/#/plant/55985/?freq=M&pin=

Catalyst

Company cannot sustain its flawed business model nor meet future service liabilities.