When a utility needs a new high-voltage line, a substation rebuilt, or a data centre wired, it usually does not employ the thousands of linemen and electricians the job needs. It hires Quanta. Quanta owns the trucks and helicopters, trains the workforce, and increasingly makes the scarce hardware itself. As the grid buildout's bottleneck shifts from megawatts to manpower, Quanta is the company that has the people.
Source: Quanta Services Q2 2026 results and 10-Q (30 Jul 2026); FY2026 guidance (30 Jul 2026); 31 Mar 2026 Investor Day; August 2026 senior-note prospectus supplements. Figures marked as guidance are management's.
Why I chose it
1. It owns the bottleneck: skilled crews
The constraint on every grid and data-centre project in 2026 is not equipment, it is labour: the industry needs an estimated 349,000 net new electrical workers this year, and a fifth of existing electricians are over 55. Quanta employs about 85,000 people, self-performs 80% of its work, spends up to $250 million a year training, and adds crews about four times faster than the industry. Capacity that cannot be bought quickly is the moat.
In plain termsEveryone can order the same transformers; almost no one else can field the crews to install them.
2. A record backlog, with the big cycle not yet in it
Backlog reached $53.4 billion, up 49% year on year, and management says roughly 95% of large 765kV and 345/500kV transmission work is not yet in that figure. US utility capital spending is guided to $238.8 billion in 2026, up 17%. The demand behind the numbers is a multi-year grid and data-centre buildout, not a single strong quarter.
In plain termsThe order book is already at a record, and the largest wave of transmission work has not even been counted yet.
3. An investment-grade balance sheet used as a weapon
Quanta owes about $7 billion at a 4.35% average coupon against roughly $4.15 billion of EBITDA this year, rated investment grade by all three agencies. It uses that balance sheet to buy craft-labour businesses at 7 to 9 times earnings, which the market then values inside Quanta at about 25 times. That arbitrage, plus 7 to 10% organic growth, has compounded earnings per share at 21% a year since 2020.
In plain termsIt buys smaller rivals cheaply with cheap debt, and the market instantly re-values those earnings much higher.
4. The capital-structure lens, inverted
My research usually looks for an equity held down by a debt overhang ahead of a refinancing. Quanta is the opposite. The credit is pristine: 1.6x leverage, about 12x interest cover, no convertibles, ample covenant headroom. The fragile leg here is not the balance sheet, it is the equity multiple. That inversion is the differentiated read: the thing to underwrite is the price paid for the earnings, not the durability of the company.
In plain termsThe debt is not the danger here. The danger is paying too high a multiple for a very good business.
What would break it
The business is strong; the risks here sit in the price and in a handful of execution lines. Each is observable in public data.
| Signal | What I would see | Where it shows first |
|---|---|---|
| The multiple compresses | The shares already de-rated from about 50x to 35x forward earnings this year with estimates rising. A further move toward the five-year median (~28x) is roughly 14% with no change in the business. This is the dominant risk. | Forward P/E versus peers; the 10-year Treasury yield |
| A renewables air pocket | About a quarter of revenue is generation and storage. The 2025 tax law requires projects to start construction by 4 July 2026 or enter service by end-2027, which risks a demand gap in wind and solar EPC from 2028. | Renewables backlog trend; bookings mix |
| A fixed-price loss | Performance bonds outstanding rose to $21.5 billion from $14.9 billion in six months, growing about twice as fast as backlog. That points to a rising share of bonded fixed-price work, where a single project loss can hit earnings. | Surety line in each 10-Q; project-loss provisions |
| A data-centre capex pause | Technology and load centres are 10 to 25% of revenue [company's own figures are inconsistent: verify]. Up to ten states were weighing data-centre moratoria or pauses by year-end 2026. | Hyperscaler capex guides; state legislation |
| The arbitrage closes | If the equity de-rated toward a peer's ~15x, acquisitions would stop being accretive, the deal machine would slow, and the 15 to 20% EPS growth would compress toward the 7 to 10% organic rate. | Deal cadence; multiple versus EMCOR and MasTec |
| Acquisition overpayment | The bull case is substantially an M&A case: about $6.4 billion deployed since 2024. Integration missteps or paying up as every rival bids for the same private contractors would erode returns. | Acquired EBITDA versus plan; goodwill; deal multiples |
Known noise, not a thesis breaker: the CEO sold about $120 million of stock at $771 in May under a pre-set plan, roughly 16% above the current price, and the shares fell into late July partly on that read. A founder-era CEO taking money at a 50x multiple is rational, not a signal of trouble. Short interest is only about 2% of the float, so this is not a squeeze name in either direction.
How I value it
A renewables air pocket, a fixed-price loss and a data-centre pause coincide, and the multiple falls to the low end of its five-year range. 2027 EPS near $17.50 at 24x, close to the 52-week low.
Consensus is delivered, the 765kV and NiSource work books through H2, and the multiple holds near today's level. 2027 EPS near $19.80 at ~32.3x, the equivalent of the five-year median on 2028.
Earnings grow ~25%, technology margins converge toward the Electric segment, a further $1 billion of deals close at 8x, and the multiple recovers toward the 2026 high. 2027 EPS near $21.60 at 37x.
Why the street is higher, and why I wait. The analyst mean is about $770 to $803, against my base of $640. The whole gap is the multiple: the street's targets require today's ~33x forward multiple to hold on estimates that are still rising, with the 10-year Treasury at a 19-year high. I value on 2027 earnings at the five-year median multiple instead. I agree the business is the best in its group; I would rather be paid to wait in the price zone the market has already defended than pay 33x for a print the sell-side has fully pre-announced.
Entry discipline and dated checkpoints
The base value sits just below the current price, so this is a quality name to accumulate on a rates- or sentiment-driven pullback, not to chase at spot. The levels and dates below are how I would stage it.
Glossary
- Craft labour / self-perform
- Skilled trades (linemen, electricians) employed directly rather than subcontracted; the scarce input Quanta owns.
- T&D
- Transmission and distribution: the high-voltage lines that move power between regions and the lower-voltage network that delivers it.
- 765kV / EHV
- Extra-high-voltage transmission (500kV and above); 765kV is the highest AC voltage used in the US and the heart of the coming grid cycle.
- Backlog / 12-month backlog
- Estimated future revenue from awarded work; the 12-month figure is the part expected within a year.
- EPC
- Engineering, procurement and construction: one contractor delivers a project end to end, often at a fixed price.
- MSA
- Master service agreement: recurring, usually unit-priced utility work under a multi-year framework rather than a one-off bid.
- Adjusted EBITDA / EPS
- Company-defined profit measures that exclude items such as acquisition amortisation; guidance and consensus are set on this basis.
- Investment grade
- A credit rating of Baa3/BBB- or better; Quanta is Baa2/BBB/BBB, two notches inside.
- Net debt / EBITDA
- Total debt minus cash, divided by annual EBITDA; the standard measure of leverage.
- PEG
- Price/earnings ratio divided by expected earnings growth; a way to compare multiples across companies growing at different rates.
- Surety / performance bond
- An insurer's guarantee that a contractor will finish a project; a rising bonded balance signals more fixed-price risk.
- Forward P/E
- Share price divided by expected next-year earnings; the multiple whose level drives most of this idea's outcome.