Alphabet borrowed $25bn. $10bn of it matures after 2045

Somebody at Alphabet has decided the company will still be paying for this in 2066.

On 6 August it priced $25.0 billion of notes in nine tranches. The shortest is floating-rate paper maturing in 2028. The longest is a 6.500% bond due 15 August 2066 — forty years, at a company that had almost no debt three years ago.

The nine tranches

Moody’s rates the notes Aa2; S&P rates them AA+. This is about as good as corporate credit gets, and the pricing shows it at the short end — 33 basis points over Treasuries for two-year money is close to free.

The long end is a different story. By 2066 the spread has widened to 130 basis points and the coupon is 6.500%.

Read the maturities, not the total

$25 billion is a headline. The distribution of it is the disclosure.

$10.0 billion — 40% of the offering — matures after 2045. A company borrowing for a two-year working capital need does not sell forty-year bonds. A company doing this is funding assets it expects to be running, and paying for, deep into the century.

The filings do not say what those assets are. Anyone who has read Alphabet’s cash flow statement lately can make a reasonable guess, and we are not going to make it for you in a sentence the document does not support.

What it costs

2028   $1,250M × 4.500%  =    $56.3M
2029   $2,000M × 4.625%  =    $92.5M
2031   $3,500M × 4.875%  =   $170.6M
2033   $2,500M × 5.200%  =   $130.0M
2036   $4,500M × 5.450%  =   $245.3M
2046   $3,000M × 6.250%  =   $187.5M
2056   $4,500M × 6.375%  =   $286.9M
2066   $2,500M × 6.500%  =   $162.5M
─────────
$1,331.5M a year

The weighted average fixed coupon is 5.606%.

Now put that next to the number we published two days ago. In a panel of 1,519 listed non-financial companies, Alphabet was the single largest contributor to the rise in US corporate interest expense this year — from $295 million in the first half of 2025 to $1,811 million in the first half of 2026. That was already a 37.6-fold increase across four quarters: $34M, $261M, $533M, $1,278M.

Reported, first half 2026        $1,811M
Half a year of new coupons        +$666M
────────
$2,477M   →  about 37% higher

The comparison that should be uncomfortable

Eleven days after Alphabet priced, AMD sold $4.75 billion of notes including a 5.500% tranche due 2036.

Five basis points. That is the difference between the two disclosed ten-year coupons.

We want to be careful here, because the honest version has a caveat. The two deals priced eleven days apart and Treasury yields moved in between, so the coupon gap is not a clean credit spread — Alphabet’s own spread to Treasuries was 85 basis points, and AMD’s 8-K does not disclose its spread. A proper comparison needs both spreads on the same day, and we do not have them.

Alphabet’s forty-year tranche carries a 6.500% coupon. Its maturity and pricing date differ from AMD’s ten-year tranche, so these coupons alone cannot measure the benefit of Alphabet’s credit rating.

Widen the frame past investment grade and the five basis points stop looking like the story. Three days after AMD, CoreWeave signed a secured term loan at Term SOFR plus 5.50% with a floating benchmark. The applicable Term SOFR fixing is not verified here, so no all-in rate is stated. Alphabet, AMD, Intel and CoreWeave announced financing within three weeks of each other, and the five instruments they used land in five different places on the financial statements.

What would make this wrong

• Coupon is not interest expense. The notes priced below par, there are underwriting discounts of 0.125% to 0.600%, and accounting interest will run above the coupon. Our $1,331.5 million is contractual.
• The $2,477 million figure is arithmetic, not guidance. It adds half a year of new coupons to a reported half-year figure and assumes nothing is retired.
• We do not know what the money is for. The prospectus uses customary language. Any link to capital spending is an inference, and the connection is not stated in the document.
• The AMD comparison is rough, for the reason given above. It is included because the raw numbers are striking, with the caveat attached rather than buried.
• A long maturity is not a warning. Locking in forty-year funding at a fixed rate can be prudent, expensive, or both, and which one it turns out to be depends on rates nobody can call.
• Ratings are the agencies’ opinions, quoted from the term sheet, not our assessment.

Check it yourself

Everything in the tranche table is on pages 2 to 5 of the pricing term sheet, linked below — amount, coupon, yield, spread, benchmark Treasury, offering price and underwriting discount for each of the nine tranches. The 8-K of 10 August records the closing. Alphabet’s quarterly interest expense is one API call.

The coupon arithmetic is in the article and can be redone in a minute.

If a figure here does not match a filing, tell us and it will be corrected on the article and on the corrections log, with the date.