Stock Analysis · Bulls vs Bears

Boost Run Inc. WLAC

🐂 Bulls vs Bears 🐻Report #110 Oct 2026
★ Willem owns this stock · 0.4% of the portfolio
67
high risk
🐂 6vs8 🐻
Bears lead
31%
▲ Chance of the biggest tailwind the forecasts miss
62%
▼ Chance of the biggest headwind the forecasts miss
The verdict

Boost Run is a real but very young GPU-rental business. The demand side has been proven by two blue-chip AI-lab contracts; the supply side (power, hardware delivery and above all financing) has not. Owning it is a bet that a sub-scale, founder-controlled de-SPAC can finance and energise roughly $2.6B of contracts on time in a sellers' market. Today the evidence is stronger on demand than on funding and execution.

The Bulls vs Bears podcast is being made
Two AI hosts debate this report, NotebookLM-style. It appears here automatically within a few hours.

Written by Claude (Anthropic’s AI) from public sources · Willem owns WLAC · AI can make mistakes — numbers in brackets point to the sources · Not financial advice

🐂 Bull arguments 6

  1. 1Real take-or-pay contracts with frontier AI labsStrong
  2. 2A sellers' market for GPU capacityMedium
  3. 3Customers fund a large part of the buildMedium
  4. 4Hardware access through NVIDIA and DellMedium
  5. 5The revenue ramp is real and already EBITDA-positiveMedium
  6. 6Founder alignment and a reset share priceWeak

🐻 Bear arguments 8

  1. 1The financing gapHigh
  2. 2Execution and delivery risk, with a poor forecasting recordHigh
  3. 3No durable competitive edgeMedium
  4. 4Customer concentration and customer creditMedium
  5. 5Dilution and share overhangMedium
  6. 6Governance, controls and soft metricsMedium
  7. 7GPU obsolescence and residual valueLow
  8. 8Rates and AI-capex sentimentLow

Bull arguments show how well they survive the devil’s advocate (strong / medium / weak); bear arguments show the damage they could do after the counter-argument (high / medium / low).

Bull case — all positive points (most important first)

🐂 Bull #1Real take-or-pay contracts with frontier AI labs

  • The point: Boost Run has turned a small on-demand business into a backlog of multi-year, non-cancellable contracts with well-funded customers.
  • Evidence:
    • Thinking Machines: $471.7M over 36 months, non-cancellable, fees due regardless of usage [5].
    • Cohere: $525.6M over about 5 years [6].
    • Two unnamed orders: $207.6M and $222.5M [3].
    • Claimed contract value (TCV) rose from $120M (Dec 2025) to $440M (2026-06-30) and $1.9B (2026-07-31) [13]; about $2.6B including Cohere per press [7].
  • Why it matters: Signed contracts with prepayments are what lenders lend against. They turn a speculative build-out into a financeable one, and they give revenue visibility into 2029–31.
  • 😈 Devil's advocate:
    • TCV is a company-defined metric that includes executed LOIs (letters of intent) and customer renewal options, not only legally binding obligations [13].
    • Most contracts start in late 2026–2027, so none of it is revenue yet.
    • The Cohere deal has a delivery condition and a refund right [6].
    • Both labs are private and loss-making. Their ability to pay depends on their own fundraising.
  • Verdict: Strong The two 8-K-filed contracts are hard evidence. The headline TCV is softer than it looks.

🐂 Bull #2A sellers' market for GPU capacity

  • The point: In late 2026 demand for AI compute exceeds supply, so even a sub-scale provider can fill capacity at attractive rates.
  • Evidence:
    • The SemiAnalysis H100 1-year contract index rose about 40% from October 2025 to March 2026; on-demand was "sold out across all GPU types" [20].
    • CoreWeave raised list prices about 25% in July 2026 and called near-term capacity "effectively sold out" [16].
    • A Nebius capacity auction cleared about 15% above prior Blackwell highs [21].
    • Boost Run: "Demand is not our constraint" [4].
  • Why it matters: Utilisation stays high, the on-demand slice earns spot-like prices, and new contracts can be signed with prepayments.
  • 😈 Devil's advocate:
    • The H100 spot index (Ornn) fell 5.7% in the 30 days to 2026-10-09 [22].
    • Boost Run's long contracts are fixed-price, so it captures little upside but would feel any downturn at renewal.
    • Shortages in compute have historically flipped to gluts.
  • Verdict: Medium A true tailwind today, but contract structure limits how much of it Boost Run keeps.

🐂 Bull #3Customers fund a large part of the build

  • The point: Prepayments plus leased (not owned) data centres mean less equity is needed per dollar of revenue than in a classic data-centre build.
  • Evidence:
    • Customer deposits rose from $15.4M to $128.4M in H1 2026; H1 operating cash flow was $114.2M, of which $112.9M was deposit growth [3].
    • Average prepayment is about 22% of TCV [13]. The July order was 25% prepaid [3].
    • Management targets TCV of at least 1.4x the capex it triggers [23].
  • Why it matters: If 20–25% of hardware cost arrives upfront, the remaining 75–80% can be financed with equipment leases secured on the servers and contracts.
  • 😈 Devil's advocate:
    • Deposits are an obligation to deliver capacity, not free cash. The company itself says the cash flow "should not be run-rated" [13].
    • Management conceded on the call that non-investment-grade customers create a "financing gap" [23].
    • Cohere's prepayment must be refunded if delivery slips past 2027-07-15 [6].
  • Verdict: Medium A smart structure, but it covers a minority of the capex.

🐂 Bull #4Hardware access through NVIDIA and Dell

  • The point: In a shortage, getting chips is the moat. Boost Run has preferred NVIDIA status and a large Dell commitment.
  • Evidence:
    • NVIDIA "Preferred Cloud Partner" with "Exemplar Cloud" status on Blackwell [1,4].
    • Dell agreement with $1.44B of minimum purchases over 5 years, "fully committed and allocated", with Dell Financial Services financing "aligned to customer contract timelines" [3,13,14].
    • A further $4–5B of hardware procurement across OEMs is "in process" [13].
  • Why it matters: It can promise customers delivery dates, which is what labs pay for.
  • 😈 Devil's advocate:
    • Preferred-partner status is held by dozens of clouds.
    • There is no long-term allocation guarantee [10].
    • The $4–5B has "no definitive agreements" [13].
    • The Dell agreement is a purchase obligation: $578M in year one is a commitment Boost Run must fund.
  • Verdict: Medium Access is real but not exclusive, and it creates obligations as well as options.

🐂 Bull #5The revenue ramp is real and already EBITDA-positive

  • The point: This is not a pre-revenue story.
  • Evidence:
    • Q2 2026 revenue $31.1M, +270% year on year [3].
    • ARR $30M → $145M in six months [13].
    • Company-defined adjusted EBITDA was about 40% of H1 revenue [13].
    • Operating loss was $12.9M in Q2, after $18.1M of depreciation and $10.6M of colocation rent [3].
  • Why it matters: At contracted rates, each new energised megawatt appears to be cash-generative before financing costs.
  • 😈 Devil's advocate:
    • The 94.5% "gross margin" excludes depreciation and colocation [3]. On a full-cost basis the company is loss-making.
    • Q2 included stock comp of $7.3M and a $7.0M consultant share award to a director [3,12].
    • 4-year depreciation on GPUs whose contract life is 3–5 years leaves residual-value risk.
  • Verdict: Medium The growth is real; profitability is not yet demonstrated.

🐂 Bull #6Founder alignment and a reset share price

  • The point:
    • The CEO takes $1 salary, holds 29.5M Class B shares, about 37% of the roughly 79.5M shares outstanding, and has not sold [9,10,12].
    • The stock has fallen about 69% from its $42 high to $13.18 (2026-10-09), a market value of roughly $1.0B against about $2.6B of claimed contracts [12,24].
    • Sell-side targets are $36–45: BTIG Buy $45, D.A. Davidson Buy $45, Northland Outperform $36 [25].
  • Evidence: Form 3/4 filings show no Karos sales [12]. Price data [24].
  • Why it matters: A large owner-operator has every incentive to make the contracts pay off. Some of the dilution fear may already be in the price.
  • 😈 Devil's advocate:
    • Control cuts both ways (see Bear #6).
    • BTIG advised the SPAC and is not independent [26].
    • Analyst rationales are paywalled.
    • The share price is not the business: a lower price makes the coming equity raise more dilutive.
  • Verdict: Weak Alignment is genuine, but it does not offset the funding problem.

Bear case — all negative points (most important first)

🐻 Bear #1The financing gap

  • The point: Boost Run has signed far more contracts than it has money to build.
  • Evidence:
    • FY2026 capex guidance is $1.0–1.4B (call summary; secondary) [18], against H1 capex of about $37M [13].
    • A further $4–5B of OEM procurement is planned [13].
    • Balance sheet at 2026-06-30: unrestricted cash $120.2M, equity $100.0M, total liabilities $642.2M, working capital −$23.6M [3]. Warrant exercises added about $130M gross in July–August [27].
    • Committed facility size, lessor names and rates: not disclosed [3].
    • Filings mention plans for "asset-backed securitizations and rated parent-level debt" [12].
    • Management said a non-investment-grade customer creates a "financing gap" [23].
  • Why it matters: Boost Run must either:
    1. issue equity, which is dilutive at $13;
    2. borrow at high rates, since lab credit is weaker than hyperscaler credit and the Fed hiked on 2026-09-16 [28]; or
    3. slip deliveries and risk contract refunds.
  • 🛡️ Counter-argument:
    • Prepayments (about 22%) plus Dell Financial Services and NVIDIA-linked programmes (AICP was mentioned on the call [23]) could fund most of it.
    • GPU-backed debt markets were open in 2026: Lambda placed a Baa2-rated term loan at SOFR+300 bp [29].
    • The 10-Q states liquidity is sufficient for 12 months [3], though that refers to existing operations, not the full build.
  • Severity: High This single issue decides whether the backlog becomes revenue or dilution.

🐻 Bear #2Execution and delivery risk, with a poor forecasting record

  • The point: Almost all value depends on energising new sites and racks on time over the next 9 months. Past forecasts have been missed.
  • Evidence:
    • H1 2026 revenue was $42.1M versus $180M forecast for FY2026 in the S-4 [3,11].
    • The company's own schedule puts TCV "in production" at $0.4B in Q2'26, rising to $1.9B by Q1'27 [13].
    • The 253 MW of power is "access", and not all of it is energised [13].
    • Cohere may terminate and recover all prepayments if a minimum amount is not accepted by 2027-07-15; the parent guarantees the refund [6].
    • Q1 and Q2 10-Qs were both filed late [12].
  • Why it matters: A slip of one or two quarters pushes revenue out, raises financing costs and could cost the Cohere contract.
  • 🛡️ Counter-argument:
    • Colocation in existing energised buildings (for example 10X's first 20 MW site) is faster than greenfield [19].
    • The team delivered a 4x ARR increase in H1.
  • Severity: High The timeline is tight, and the track record on forecasts is weak.

🐻 Bear #3No durable competitive edge

  • The point: Bare-metal GPU rental in rented data centres is close to a commodity. Boost Run is sub-scale against rivals with cheaper capital and their own power.
  • Evidence:
    • CoreWeave: 1.5 GW active, 4.2 GW contracted, $104B backlog [16].
    • Nebius: $3B run-rate and 5 GW power target [21].
    • Boost Run's 424B3 names CoreWeave, Nebius and WhiteFiber as competitors, plus the hyperscalers [10].
    • No proprietary technology was found. Customers can move to another provider when a contract ends.
  • Why it matters: When the shortage ends, pricing reverts toward the lowest-cost provider, and Boost Run is unlikely to be that.
  • 🛡️ Counter-argument:
    • Large providers prioritise hyperscaler-sized deals. Mid-sized labs and regulated buyers need nimble suppliers.
    • Compliance certifications and the Carahsoft government channel are niche moats [10].
  • Severity: Medium High. This matters at contract renewal (2029–31), not tomorrow.

🐻 Bear #4Customer concentration and customer credit

  • The point: A few private AI labs make up most of the future revenue.
  • Evidence:
    • Thinking Machines plus Cohere are about $1B of roughly $2.6B TCV (my arithmetic) [5,6,7].
    • History: the top customer was 71.6% of 2024 revenue and 44.5% of 2025 [10].
    • 2026 concentration is not disclosed [3].
  • Why it matters: A lab that fails to raise its next round, or that is acquired, could default, renegotiate or slow acceptances. That would also impair the financing secured on that contract.
  • 🛡️ Counter-argument: Both labs are among the best-funded private AI companies. Take-or-pay and prepayments give some protection.
  • Severity: Medium High.

🐻 Bear #5Dilution and share overhang

  • The point: Large blocks of cheap stock can be sold, and more equity is coming.
  • Evidence:
    • 10,968,750 earnout shares were issued once the stock passed $17.50 [3].
    • The resale registration covers 58.7M shares, about 77% of fully diluted shares [12]. Selling holders paid far less than today's price.
    • Lock-ups ended at the earlier of 2026-11-08 or a $12-for-20-of-30-days price test. The price test was very likely met by June (interpretation, unconfirmed) [12].
    • RSU grants on 2026-09-29: 4.59M shares worth $76.4M at $16.65 to the CFO (2.89M), CTO and COO [8].
    • Director Goodrich sold 698,200 shares in August–September at $20.15 and $14.00 [12].
    • No open-market insider buys [12].
  • Why it matters: Persistent supply caps the share price, and every equity raise lowers each share's claim on the backlog.
  • 🛡️ Counter-argument:
    • The warrant overhang is gone (99.8% exercised for about $130M of cash) [27].
    • The CEO has not sold.
  • Severity: Medium

🐻 Bear #6Governance, controls and soft metrics

  • The point: Several signs suggest the accounting and governance are still those of a private company.
  • Evidence:
    • Controlled company with about 87–90% founder voting power [9,12].
    • Material weaknesses and two late 10-Qs [3,12].
    • Headline metrics are company-defined:
      • TCV includes LOIs and renewal options [13].
      • The 94.5% "gross margin" excludes depreciation and colocation [3].
      • Adjusted EBITDA was redefined before closing [11].
      • The free-cash-flow target excludes lease-financed capex [13].
    • A $100M software licence deal with an unnamed vendor, 2.4x H1 revenue [3].
    • A 336,000-share consulting award to a director [12].
    • The CFO's RSU grant is about $48M at grant [8].
  • Why it matters: Investors cannot independently check the figures that matter most: real backlog, fleet size, utilisation and financing terms.
  • 🛡️ Counter-argument:
    • Typical of a company six months after a de-SPAC. Remediation hires (CAO, CIO) were made [10].
    • The 10-Q dropped the going-concern language [3].
  • Severity: Medium

🐻 Bear #7GPU obsolescence and residual value

  • The point: GPUs lose value fast as NVIDIA moves to new generations.
  • Evidence:
    • Vera Rubin started shipping in Q3 2026 [30].
    • Boost Run depreciates hardware over 4 years [3]. Cohere's term is about 5 years per rack [6].
    • The S-4 model assumed rental rates fall 2.5% per quarter [11].
  • Why it matters: After the first contract, older GPUs re-lease at lower rates, and some lease debt may outlast the cash flow.
  • 🛡️ Counter-argument:
    • Contracts are roughly matched to depreciation (3–5 years versus 4).
    • CoreWeave reports H100s re-leasing at about 95% of original price [31].
    • 4 years is conservative versus the 5–6 years used by peers.
  • Severity: Low Medium.

🐻 Bear #8Rates and AI-capex sentiment

  • The point: The business relies on cheap capital and enthusiasm for AI infrastructure.
  • Evidence:
    • The Fed hiked to 3.75–4.00% on 2026-09-16, and most members expect another hike [28].
    • Neocloud stocks sold off in August–September on financing worries [32] (opinion pieces; causality unconfirmed).
  • Why it matters: Higher spreads make each new lease more expensive and narrow the gap between rental income and financing cost.
  • 🛡️ Counter-argument: Contract prepayments and short lease terms (30–36 months) limit rate exposure on existing deals.
  • Severity: Low Medium.

Tailwinds NOT in the forecasts

  • A large contract-backed financing facility at a reasonable cost. 31% chance within 36 months · Impact if it happens: High

    • What it is: A committed facility of $1B or more (Dell Financial Services, NVIDIA-supported lending, or a GPU ABS) that funds most of the $1.9B+ TCV build without big equity issuance.
    • Why it's not priced in: The market is pricing in heavy dilution (stock −69% from peak). No facility has been disclosed.
    • What would trigger it / the sign to watch: An 8-K announcing a credit agreement, ABS or vendor-finance programme, with the size and spread disclosed.
    • Why this probability: Peers (Lambda, IREN) have raised GPU-backed debt. But Boost Run's sub-scale balance sheet and lab customers make a large, cheap facility less likely than not in the near term.

  • Sustained GPU shortage lifts on-demand pricing and renewals. 38% chance · Impact: Medium

    • What it is: Blackwell and Rubin demand keeps capacity tight through 2027, so the over-12% on-demand slice and early renewals reprice up.
    • Why it's not priced in: The S-4 model and consensus assume prices fall about 2.5% a quarter [11].
    • What would trigger it / the sign to watch: The SemiAnalysis/Ornn indices rising again, or CoreWeave and Nebius announcing more price increases.
    • Why this probability: The signal is strong today, but the H100 index has softened over the last 30 days and new supply (Rubin) is coming.

  • Another named lab or sovereign contract of $300M or more. 55% chance · Impact: Medium

    • What it is: Converting part of the $4–5B procurement pipeline into new multi-year contracts.
    • Why it's not priced in: Consensus revenue (about $551M in 2027, secondary [33]) appears to cover existing TCV only.
    • What would trigger it / the sign to watch: 8-K Item 1.01 filings like those for Thinking Machines and Cohere.
    • Why this probability: Management has signed roughly one major deal every 1–2 months since May. The binding constraint is financing, not demand.

  • Acquisition by a larger neocloud or infrastructure fund. 12% chance · Impact: High

    • What it is: A buyer acquires the contracts and the power pipeline.
    • Why it's not priced in: It is not a consensus topic, and Karos' control means it only happens on his terms.
    • What would trigger it / the sign to watch: A strategic-review announcement, or 13D filings.
    • Why this probability: Industry consolidation is plausible, but controlled companies rarely sell early.

Headwinds NOT in the forecasts

  • Equity or convertible raise below today's price within 12 months. 62% chance · Impact: High

    • What it is: An at-the-market programme, follow-on offering or convertible note to fund the capex gap.
    • Why it's not priced in: Consensus share counts and the $36–45 price targets appear to assume debt-funded growth (analyst notes paywalled; inferred).
    • What would trigger it / the sign to watch: An S-3 shelf filing (eligible about 12 months after listing, unless done sooner via S-1), an ATM announcement, or convertible pricing.
    • Why this probability: Equity of $100M against more than $1B of planned capex makes some equity component likely. Debt alone at this size is unusual for a 2-year-old company.

  • Delivery slip that triggers Cohere refund or renegotiation. 18% chance · Impact: High

    • What it is: The minimum Cohere racks are not accepted by 2027-07-15, so Cohere can terminate and get its prepayment back [6].
    • Why it's not priced in: The deal was only announced on 2026-10-06, and consensus counts it as revenue from mid-2027.
    • What would trigger it / the sign to watch: Delays at the 10X/Pryor sites, GB300 supply delays, or language about "acceptance timing" in the Q3/Q4 reports.
    • Why this probability: There are about 9 months of buffer after the planned Q2 2027 start. But GB300 NVL72 racks need high-density liquid-cooled power, and the company has no record with them.

  • A major customer defaults or renegotiates. 13% chance within 36 months · Impact: High

    • What it is: A private lab customer fails to raise funding, or merges and cuts its compute commitments.
    • Why it's not priced in: Contracts are treated as firm backlog.
    • What would trigger it / the sign to watch: Lab funding news, layoffs, acquisition.
    • Why this probability: Both named labs are well capitalised, but AI-lab funding is cyclical and 3–5-year terms are long.

  • Restatement or adverse audit finding. 15% chance · Impact: Medium

    • What it is: The material weaknesses lead to a restatement, delayed 10-K or auditor change.
    • Why it's not priced in: Material weaknesses are treated as routine for a de-SPAC.
    • What would trigger it / the sign to watch: Another NT 10-Q/10-K, an 8-K Item 4.01 (auditor change) or Item 4.02 (non-reliance).
    • Why this probability: Two late filings in two quarters, rapid growth in complex lease accounting, and the first public-company 10-K audit is due early 2027.

Investment Risk Score: 67/100

Risk area Weight Score (0–100) Why
Business & competition 25% 68 Commodity GPU rental with no owned power or sites, sub-scale versus CoreWeave and Nebius; helped for now by an industry-wide shortage.
Financial health & dilution 20% 78 $100M equity against $642M of liabilities, more than $1B of planned capex with no disclosed facility, 58.7M shares registered for resale, $76M of new executive RSUs.
Execution (delivering the plan) 20% 74 H1 revenue was 23% of the S-4 full-year forecast; $1.9B to be deployed by Q1'27 across sites not yet energised; Cohere delivery deadline 2027-07-15.
Dependency & concentration 15% 72 Two private labs about $1B of contracts; reliant on NVIDIA allocation, Dell, colocation landlords and one controlling founder.
Regulation, policy & geopolitics 10% 40 Section 232 chip tariffs (US data-centre carve-outs), export controls on GPU supply, and power-interconnection rules; mostly indirect.
Sentiment & expectations 10% 52 The stock is down 69% from its peak, but price targets of $36–45 and 2027 consensus revenue (about $551M) assume on-time delivery.

Overall = 67/100 (high). The weighted calculation: 0.25×68 + 0.20×78 + 0.20×74 + 0.15×72 + 0.10×40 + 0.10×52 = 67.4. Boost Run has proven that it can sign customers. It has not proven that it can fund and deliver what it signed, and its accounting and governance are still immature. First deep dive, so there is no prior score to compare.

Conclusion

  • What kind of company this is: a two-year-old, founder-controlled GPU landlord that went public via SPAC at the peak of the AI-compute shortage. Owning it is a bet that it can turn about $2.6B of signed and claimed contracts into energised racks, on time and without excessive dilution. In effect it is a leveraged project-finance bet on a handful of AI-lab customers.
  • The 2–3 things that will decide the outcome:
    1. The size, cost and structure of the financing for the 2026–27 build.
    2. On-time energising of the 10X, Pryor and Rock Island sites, and Cohere acceptance by mid-2027.
    3. Whether the AI-lab customers stay solvent and keep signing.
  • Signposts that would confirm the bull case:
    • A committed debt or vendor-finance facility of $1B or more at a sensible spread.
    • Q3/Q4 revenue tracking toward about $400M ARR by year-end.
    • A third named customer contract.
    • Material weaknesses remediated and the 10-K filed on time.
  • Signposts that would break it:
    • A large equity raise or convertible at or below the current price.
    • Missed handover dates or language about "acceptance timing".
    • Disclosed 2026 customer concentration above 50% in one name.
    • Another late filing or an auditor change.
  • Balanced verdict:
    • The strongest bull case is that Boost Run got into the AI-compute boom early and signed blue-chip labs on take-or-pay terms with prepayments. In a sold-out market, that is the hardest part, and the $1B-ish market value is well below claimed contract value.
    • The strongest bear case is that signing was never the hard part in 2026. Funding and energising are, and Boost Run has neither the balance sheet nor the record. Its metrics are self-defined, its forecasts have been missed, and a large share overhang meets a likely equity raise.
    • Today the evidence leans toward the bear side on financing and execution, and toward the bull side on demand. The financing announcement, whenever it comes, is the single most important piece of missing information.

Open questions & data gaps

  • Financing terms. The lessors and rates on the $238M of GPU finance leases are not found, and neither are the Dell Financial Services terms. This matters most.
  • Fleet size and utilisation. GPU count and utilisation are not disclosed. Without them, unit economics cannot be checked.
  • What makes up TCV. The split of the $1.9B between binding contracts, LOIs and renewal options is not disclosed. The bridge from $440M (6/30) to $1.9B (7/31) to "$2.6B" is unreconciled.
  • 2026 customer concentration. Not disclosed, and management did not answer the question on the call (summary only).
  • The $100M software arrangement (2026-04-17). Counterparty and purpose not disclosed. Interpretation: possibly NVIDIA AI Enterprise or similar software licences, unconfirmed.
  • Guidance and transcript. Whether FY2026 revenue guidance of $170–180M still stands is unknown, and no verbatim call transcript was available (AI summaries only).
  • Price moves. No source was found for the causes of the large drops on 2026-08-28 (−17%), 2026-10-05 (−13%) and 2026-10-08 (−12%).
  • Lock-up and results date. Lock-up release status and the Q3 results date are both unconfirmed. The 10-Q is likely due by around 2026-11-16.
Sources
  1. Boost Run Completes Business Combination with Willow Lane (Ex 99.4) — SEC/Boost Run — 2026-05-08 — https://www.sec.gov/Archives/edgar/data/0002090646/000149315226023208/ex99-4.htm
  2. Boost Run Inc. completes business combination with Willow Lane — Winston & Strawn — 2026-05 — https://www.winston.com/en/experience/boost-run-inc-completes-business-combination-with-willow-lane-acquisition-corp-71269100
  3. Form 10-Q, quarter ended 2026-06-30 — SEC/Boost Run — 2026-08-18 — https://www.sec.gov/Archives/edgar/data/0002090646/000149315226039031/form10-q.htm
  4. Q2 2026 results press release (8-K Ex 99.1) — SEC/Boost Run — 2026-08-14 — https://www.sec.gov/Archives/edgar/data/2090646/000149315226038014/ex99-1.htm
  5. 8-K, Thinking Machines Lab master services agreement — SEC/Boost Run — 2026-05-28 — https://www.sec.gov/Archives/edgar/data/2090646/000149315226025672/form8-k.htm
  6. 8-K, Cohere service agreement — SEC/Boost Run — 2026-10-06 — https://www.sec.gov/Archives/edgar/data/0002090646/000149315226045861/form8-k.htm
  7. Boost Run signs $525.6M cloud deal with Cohere — Crypto Briefing / DatacenterDynamics (secondary) — 2026-10-06 — https://cryptobriefing.com/boost-run-526m-cohere-cloud-deal/
  8. 8-K Item 5.02, executive RSU and cash awards — SEC/Boost Run — 2026-10-05 — https://www.sec.gov/Archives/edgar/data/0002090646/000149315226045765/form8-k.htm
  9. 8-K, closing report (controlled company, consideration) — SEC/Boost Run — 2026-05-14 — https://www.sec.gov/Archives/edgar/data/2090646/000149315226023208/form8-k.htm
  10. Proxy statement/prospectus 424B3 (business, risk factors, bios, projections) — SEC — 2026-04-08 — https://www.sec.gov/Archives/edgar/data/2090646/000149315226015740/form424b3.htm
  11. Form S-4 (original projections) — SEC — 2026-01-13 — https://www.sec.gov/Archives/edgar/data/2090646/000149315226002095/forms-4.htm
  12. Resale prospectus 424B3 (customer concentration, lock-ups, 58.7M shares), POS AM (2026-10-08) and Form 3/4 filings — SEC — 2026-07-10 / 2026-10-08 / 2026-05-15 to 10-06 — https://www.sec.gov/Archives/edgar/data/2090646/000149315226032874/form424b3.htm ; https://www.sec.gov/Archives/edgar/data/0002090646/000149315226046345/formposam.htm ; https://www.sec.gov/Archives/edgar/data/2028811/000149315226042924/ownership.xml
  13. Q2 2026 investor presentation (8-K Ex 99.2) — SEC/Boost Run — 2026-08-14 — https://www.sec.gov/Archives/edgar/data/2090646/000149315226038014/ex99-2.htm
  14. Boost Run announces $1.44B agreement with Dell Technologies — PR Newswire via Finviz — 2026-04-22 — https://finviz.com/news/346553/boost-run-announces-144-billion-agreement-with-dell-technologies-to-further-facilitate-enterprise-demand-for-ai-compute-and-storage-infrastructure
  15. GPU cloud Boost Run plans to go public via de-SPAC merger — DatacenterDynamics — 2025-09 — https://www.datacenterdynamics.com/en/news/gpu-cloud-boost-run-plans-to-go-public-via-de-spac-merger/
  16. CoreWeave Q2 earnings call highlights — MarketBeat (secondary) — 2026-08-11 — https://www.marketbeat.com/instant-alerts/coreweave-q2-earnings-call-highlights-2026-08-11/
  17. How the GPU became collateral — Data Gravity (secondary, analysis) — 2026 — https://www.datagravity.dev/p/how-the-gpu-became-collateral
  18. Boost Run Q2 2026 earnings summary — Quartr (AI summary, secondary) — 2026-08-18 — https://quartr.com/events/boost-run-inc-brun-q2-2026_F1oYt6MR
  19. Boost Run secures long-term lease with 10X Capital to expand GPU compute capacity — Business Wire via FinancialContent / Citybiz — 2026-08-14 — https://www.financialcontent.com/article/bizwire-2026-8-14-boost-run-secures-long-term-lease-with-10x-capital-to-expand-gpu-compute-capacity
  20. The Great GPU Shortage: H100 1-year rental price index — SemiAnalysis (via search summary) — 2026-03 — https://newsletter.semianalysis.com/p/the-great-gpu-shortage-rental-capacity
  21. Nebius Q2 2026 earnings summary — Quartr (secondary) — 2026-08 — https://quartr.com/events/nebius-group-n-v-nbis-q2-2026_ozKEzQzr
  22. OCPI H100 SXM index — Ornn — 2026-10-09 — https://data.ornn.com/markets/h100-sxm
  23. Earnings call transcript: Boost Run Q2 2026 (AI-assisted summary) — Investing.com — 2026-08-14 — https://www.investing.com/news/transcripts/earnings-call-transcript-boost-run-q2-2026-revenue-surges-270-as-stock-jumps-93CH-4860745
  24. BRUN price history — StockAnalysis.com — 2026-10-09 — https://stockanalysis.com/stocks/brun/history/
  25. Boost Run analyst ratings (BTIG, Northland, D.A. Davidson, Weiss) — MarketBeat syndication via dailypolitical.com — 2026-09 — https://www.dailypolitical.com/?p=10515120
  26. Boost Run Q2 2026 slides summary — Investing.com — 2026-08 — https://ca.investing.com/news/company-news/boost-run-q2-2026-slides-arr-triples-19b-backlog-fuels-growth-93CH-4801459
  27. Boost Run announces completion of warrant redemption — Business Wire via Finviz — 2026-08-26 — https://finviz.com/news/385677/boost-run-announces-completion-of-warrant-redemption
  28. FOMC statement — Federal Reserve — 2026-09-16 — https://federalreserve.gov/newsevents/pressreleases/monetary20260916a.htm
  29. Lambda $3B pre-IPO round / term loan — TheNextWeb (secondary) — 2026 — https://thenextweb.com/news/lambda-3bn-pre-ipo-round
  30. NVIDIA confirms Vera Rubin launch for Q3 — Digital Citizen (secondary) — 2026-05 — https://www.digitalcitizen.life/nvidia-confirms-vera-rubin-launch-for-q3-as-blackwell-demand-keeps-climbing/
  31. GPU depreciation and durability debate — FourWeekMBA (secondary) — 2026 — https://fourweekmba.com/ai-nvidia-gpu-depreciation-durability-debate-huang-2026/
  32. IREN, Nebius, CoreWeave: why these neocloud stocks are falling — Invezz (opinion) — 2026-09-14 — https://invezz.com/news/2026/09/14/iren-nebius-coreweave-heres-why-these-neocloud-stocks-are-falling/
  33. Boost Run soars 11% on $526M sovereign AI cloud deal — TIKR (secondary) — 2026-10-06 — https://www.tikr.com/ja/blog/boost-run-soars-11-on-526-million-sovereign-ai-cloud-deal-an-important-step
Disclaimer. AI-generated research and podcast (Claude by Anthropic, voices by Google Gemini) for educational purposes — not financial advice and not a recommendation to buy or sell. Willem (MR.INVEST) owns this stock. Check the company’s own filings before making any decision.
← All Bulls vs Bears reports