BDC Weekly: Income Held. Portfolio Growth Did Not.

Last updated: August 8, 2026. Originally prepared as the August 4 BDC Weekly issue; the earnings-wave section below has been refreshed with results released since publication.

The BDC income engine is still working. The portfolio-replacement engine is not working equally well everywhere.

That distinction matters more than the headline yield.

A BDC can report a strong quarter because floating-rate loans are still earning attractive coupons, repayments produce fees, and prior underwriting continues to generate cash. It can also finish that same quarter with fewer earning assets, thinner dividend coverage, a softer NAV, or a larger tail of non-accruals.

That is what this earnings season is beginning to show. The income statement still looks resilient. The balance sheet beneath it is moving.

Ares Capital generated $0.47 of core earnings per share against a $0.48 dividend while exits exceeded new commitments by $323 million. Hercules Capital covered its total distribution, but record repayments helped shrink the debt portfolio by $170.2 million at cost. Golub Capital BDC earned exactly enough NII to cover its $0.33 dividend while its portfolio declined by $120.9 million and new commitments fell to just $12.6 million.

Capital Southwest was the counterexample. It originated $222.3 million and received only $19.5 million from prepayments and exits. Its portfolio grew. But pre-tax NII of $0.57 per share did not fully cover the $0.58 regular dividend, NAV slipped, and non-accruals remained material.

The sector is therefore negotiating three pressures at once. Strong borrowers are repaying loans. Weaker borrowers are consuming more attention and sometimes more capital. New opportunities must be large enough to replace the assets leaving the book, but good enough not to become the next restructuring.

That is why the next phase of earnings is not simply about who reports the highest portfolio yield.

It is about who can reproduce the income.

The Quarter in One Table

The table does not identify a single winner. It shows four different ways the same rate and credit environment can reach a BDC balance sheet.

ARCC has scale and liquidity, but activity was net negative and non-accruals moved higher. HTGC produced strong distribution coverage, but part of the quarter’s yield strength came from repayments that removed earning assets. GBDC protected the dividend while shrinking the portfolio and buying back discounted shares. CSWC grew aggressively, but its recurring dividend math tightened.

The important comparison is not income versus income.

It is income versus the capital required to reproduce it next quarter.

For the full framework behind that comparison, start with The BDC Investing Guide and BDCs: The Public Door Into Private Credit.

Interest Rates Are Still Supporting Income—and Extending the Credit Test

The Federal Reserve held the federal-funds target range at 3.5% to 3.75% on July 29. The decision passed 9–3, with three officials preferring a quarter-point increase.

That vote is more consequential for BDCs than a generic “higher for longer” headline suggests.

Most BDC assets are floating-rate loans. When short-term rates stay elevated, asset income remains supported. Many BDC liabilities, however, are fixed-rate or reset more slowly. That spread between floating assets and slower-moving funding costs has been one of the sector’s most important earnings tailwinds.

The same rate structure works against the borrower.

A portfolio company does not experience a 12% loan as a dividend-support mechanism. It experiences it as a cash obligation. The longer rates remain elevated, the more likely it becomes that a weak borrower asks for an amendment, shifts interest into PIK income, sells an asset, raises expensive equity, or enters a restructuring.

This creates a delayed transmission mechanism. BDC income can remain strong before credit losses become visible. The asset yield moves immediately. The borrower’s deterioration takes time to appear in marks, non-accruals, and realized losses.

That was the central argument in BDC Weekly: When Rates Stop Being a Tailwind. Rate cuts are not automatically bullish for BDC income, because floating-rate assets can reprice downward before funding costs fully adjust. But leaving rates high is not free either. It preserves current NII while keeping pressure on the refinancing calendar described in The Private-Credit Refinancing Wall.

The sector is caught between two forms of compression.

Lower rates can compress asset yields. Higher rates can compress borrower cash flow.

The strongest BDCs will not be the ones that guess the next Fed move. They will be the ones whose portfolios can survive both sides of the path.

AI Infrastructure Is the New Financing Outlet—and the New Discipline Test

The portfolio-replacement problem would be easier if private credit lacked demand.

It does not.

The AI infrastructure buildout is creating one of the largest capital-formation opportunities in the market. Goldman Sachs Research estimates that leading technology companies could spend $5.3 trillion on AI and data centers from 2025 through 2030. The International Energy Agency projects global data-center electricity consumption could more than double to roughly 945 terawatt-hours by 2030. Brookfield and Bloom Energy recently expanded a framework to finance AI-related power projects from $5 billion to $25 billion.

Those numbers describe more than a technology cycle. They describe a financing system.

Data centers require land, power, cooling, fiber, equipment, construction, leases, long-duration debt, project equity, insurance capital, asset-backed structures, and refinancing capacity. The market is beginning to understand that AI is not weightless. Compute sits inside buildings. Buildings sit on grids. Grids require capital.

That creates a potential replacement channel for private lenders at the same moment ordinary middle-market portfolio growth has become uneven.

But the opportunity does not flow automatically into public BDCs.

Large alternative-asset managers can route AI infrastructure transactions through infrastructure funds, real-estate vehicles, insurance accounts, private-credit funds, asset-backed strategies, banks, project companies, and separately managed accounts. A manager-level announcement is not evidence that its affiliated public BDC owns the loan.

That distinction is essential for Blue Owl Capital Corporation, Blackstone Secured Lending, Ares Capital, and FS KKR Capital. The surrounding platforms may participate in enormous AI financings. The public BDC owns only what appears in its filings.

The nearer-term BDC opportunity may sit one layer below the headline project: power-service companies, cooling systems, fiber providers, equipment suppliers, data-center contractors, software infrastructure, security providers, and venture-backed companies serving the buildout. HTGC’s innovation-economy orientation makes it a natural credit sensor, but even there, AI exposure should be measured borrower by borrower rather than inferred from the theme.

This is where the portfolio-replacement problem meets the underwriting problem.

AI financing can supply volume. It can also tempt lenders to accept long duration, uncertain residual values, construction risk, technology obsolescence, customer concentration, and aggressive refinancing assumptions because the end market appears inevitable.

The end market can be real and the loan can still be bad.

For the full capital map, read How AI Infrastructure Gets Financed, Who Finances AI Data Centers?, and Asset-Backed Finance and AI Infrastructure. The earlier weekly, How Meta’s AI Data Centers Are Pulling Private Credit Into Infrastructure, explains why the manager-versus-BDC distinction matters.

The AI boom may help replenish private-credit portfolios.

It should not be allowed to lower the standard for what belongs inside them.

ARCC: Scale Absorbed a Slower Quarter, but the Credit Signal Moved

ARCC’s second quarter looked stable at the income line and less comfortable underneath it.

Core EPS was $0.47, one cent below the $0.48 quarterly dividend. GAAP NII was $0.50 per share. The difference does not make the payout immediately vulnerable, but it narrows the recurring earnings cushion investors usually associate with the sector’s largest BDC.

Investment activity also moved backward on a gross basis. ARCC made $2.592 billion of new commitments and exited $2.915 billion. The $323 million gap helps explain why portfolio investments at fair value ended at $29.349 billion, slightly below the year-end level.

The more important change was credit.

Loans on non-accrual represented 2.4% of investments at amortized cost and 1.4% at fair value, up from 1.8% and 1.2%, respectively, at year-end. The portfolio’s weighted-average grade held at 3.1, so the reported picture is not broad deterioration. It is a reminder that a stable average can coexist with a larger tail of problem assets.

ARCC still ended with deep liquidity and a large investment backlog. Its scale gives it room to wait for better transactions rather than chase volume. The broader Ares platform may see opportunities in infrastructure, asset-backed finance, and AI-adjacent lending, but ARCC investors should keep the analysis at the vehicle level.

Selectivity has a cost. When exits outrun commitments, dividend coverage increasingly depends on the income productivity of the assets that remain.

Scale buys patience. It does not eliminate the replacement requirement.

HTGC: Repayments Helped the Quarter and Complicated the Next One

Hercules Capital’s full Q2 analysis showed the cleanest version of the repayment paradox.

NII reached $0.50 per share. That covered the $0.40 base distribution by 1.25x and the $0.47 total distribution by about 1.06x. NAV increased to $12.15 from $11.90.

Those are strong outcomes.

But early repayments reached $572.1 million. The debt portfolio contracted by $170.2 million at cost even after $647.5 million of total fundings.

The yield bridge tells investors why this matters. Effective yield rose to 13.4% from 12.8%, while core yield declined to 12.0% from 12.2%. Repayment-related fees and accelerations improved reported yield even as the recurring portfolio yield softened.

This does not make the income low quality. Repayment economics are part of venture lending.

It does mean the quarter converted some future interest income into current income. HTGC now has to redeploy that capital without giving away credit quality or structure.

AI and data-center demand could expand the venture and growth-credit opportunity set around infrastructure software, cybersecurity, chips, power technology, and specialized equipment. That is promising. It is not the same as saying every AI-linked borrower deserves venture debt.

HTGC reported $927.3 million of new commitments. The test is how much of that pipeline becomes retained, funded portfolio growth after another wave of exits—and whether the new credits produce durable core yield rather than merely a compelling theme.

GBDC: The Dividend Held While the Balance Sheet Chose Defense

GBDC’s quarter was an exercise in preservation.

NII was $0.33 per share. The dividend was also $0.33. Adjusted NII was $0.34. Coverage was intact, but there was almost no excess earnings cushion.

Meanwhile, portfolio fair value declined to $8.196 billion from $8.317 billion. New investment commitments were only $12.6 million. NAV slipped to $14.25 from $14.35.

The quarter’s $0.11-per-share realized and unrealized loss came partly from companies already on non-accrual or placed on non-accrual during the quarter, plus realized losses on two restructurings.

GBDC did not respond by forcing growth. It repurchased about 1.1 million shares during the quarter at an average $12.90 and another 0.4 million shares after quarter-end at $12.86—well below the $14.25 NAV.

That trade is economically rational. Buying discounted shares can add more value per dollar than originating a mediocre loan near par. The mechanics are explained in Discounts to NAV.

It also reveals the state of the opportunity set.

When a BDC chooses buybacks over new assets, it may be telling investors that its own balance sheet is the most attractive credit allocation available.

That can protect NAV per share. It cannot by itself replace the income from a shrinking portfolio.

GBDC’s discipline is therefore both a strength and a challenge. Refusing weak loans protects future credit quality. Refusing too many loans for too long eventually reduces the income base.

CSWC: Growth Was Available, but the Dividend Cushion Tightened

Capital Southwest showed that the sector’s origination engine has not stopped everywhere.

The company originated $222.3 million of new commitments across 11 new and 16 existing portfolio companies. Prepayments and exits were only $19.5 million. The credit portfolio reached roughly $2.0 billion at fair value, with 99% in first-lien senior secured debt.

That growth came with a tighter payout equation.

Pre-tax NII was $0.57 per share. The regular quarterly dividend was $0.58, producing coverage of about 0.98x. The $0.64 total payout, including the supplemental dividend, was covered about 0.89x by pre-tax NII.

Capital Southwest had an estimated $0.87 per share of undistributed taxable income, so one quarter of sub-1.0x coverage does not make the dividend unsupported. The reserve exists for precisely this kind of gap.

Still, the balance-sheet signals were mixed. NAV declined to $16.61 from $16.69. Net realized and unrealized depreciation totaled $10.9 million. Non-accruals represented 2.9% of the portfolio at cost and 1.1% at fair value.

CSWC is proving it can find assets.

The next question is whether those assets can produce enough recurring income and stable marks to keep the regular dividend from leaning on prior-period earnings.

Growth solves the replacement problem only when the new portfolio earns its way onto the balance sheet.

There Is No Single Private-Credit Default Rate

Fitch reported that its U.S. private-credit default rate reached 6.0% for the trailing 12 months ended in the second quarter, up from 5.7% in the first quarter.

Proskauer’s separate Private Credit Default Index reported a 2.51% default rate for the second quarter, down from 2.73% in the first quarter. Its index covered 716 senior-secured and unitranche loans representing $195.6 billion of original principal.

Both numbers can be true because they measure different universes under different definitions.

That is not a technical footnote. It is an investor warning.

Private credit has no single consolidated tape. Default rates depend on which managers report, which loans enter the sample, whether restructurings count as defaults, how repeat defaults are handled, and whether the measure is quarterly or trailing 12 months.

The correct conclusion is not that one index is right and the other is wrong.

It is that credit stress must be tested against the actual BDC portfolio.

For public BDC investors, the most useful evidence remains company-level: non-accrual migrations, realized losses, unrealized marks, amendment activity, PIK income, and the difference between cost and fair value on problem credits. Private Credit’s Discipline Cycle Has Started explains why these signals often appear gradually rather than all at once.

The sector-level default rate sets the weather.

The portfolio tells you whether the roof is leaking.

The Earnings Wave Is Here

The reports that were still ahead when this issue was prepared have now arrived.

Blue Owl Capital Corporation improved the income line in Q2 without fully repairing the balance sheet. Adjusted NII rose to $0.34 per share, GAAP NII reached $0.36, and total declared dividends were $0.33. But commitments of $319 million remained below $747 million of sales and repayments, NAV fell to $14.26, and non-accruals increased to 2.8% at cost. The full read is now live: Blue Owl Capital Q2 2026: Earnings Recovered, but the Portfolio Is Still Shrinking.

FS KKR Capital also moved from anticipation to evidence. FSK reported Q2 results on August 6 and declared a $0.44 per-share third-quarter distribution. The important follow-up is whether lower leverage and improving non-accruals are enough to offset continued portfolio contraction and the role of adviser fee support in reported earnings.

Blackstone Secured Lending has now released its second-quarter 2026 earnings materials as well. The next analysis should focus on whether its predominantly first-lien portfolio is containing the software and restructuring risks that mattered entering the quarter—and, as with Blue Owl, on separating Blackstone platform-level AI and infrastructure activity from assets actually held inside BXSL.

Main Street Capital converted its preliminary range into final results. Q2 NII was $0.97 per share, distributable NII was $1.04, and NAV reached $33.92 per share. The company also reported a $65.0 million net fair-value increase in the quarter. That makes MAIN the clearest case in this wave where income strength and NAV appreciation arrived together rather than fighting each other.

The earnings wave did not invalidate the original thesis of this issue.

It sharpened it.

Across the new reports, the central question remains:

Did the BDC preserve enough earning assets—and enough NAV—to reproduce the dividend next quarter?

Investor Quick Answers

Are BDC dividends still covered?

Mostly, but the cushions differ. HTGC covered its total distribution by about 1.06x. GBDC covered its dividend exactly at 1.00x using GAAP NII. ARCC’s core EPS covered about 0.98x of its dividend, while GAAP NII covered about 1.04x. CSWC’s pre-tax NII covered about 0.98x of its regular dividend and 0.89x of its total payout. OBDC’s Q2 adjusted NII covered its $0.33 total declared dividend by about 1.03x.

Why does portfolio contraction matter if current NII is strong?

A repayment can create fees and accelerated discount income, lifting current NII. It also removes an earning asset. If the BDC cannot replace that loan at a similar risk-adjusted yield, future income can weaken even after a strong reported quarter.

How do interest rates affect BDCs now?

Elevated short-term rates support income on floating-rate assets, but they also increase borrower debt-service burdens. Rate cuts could reduce asset yields before all funding costs adjust, while unchanged rates extend the credit test for weaker companies.

Can AI infrastructure solve the BDC portfolio-growth problem?

It can expand the private-credit opportunity set, especially through power, equipment, software, services, fiber, cooling, and data-center supply chains. But large AI project financings often sit in infrastructure funds, insurance accounts, banks, or private vehicles rather than public BDCs. Investors need portfolio evidence, not platform association.

Is a shrinking portfolio always bad?

No. Allowing unattractive loans to repay can be better than forcing new volume. Discounted share repurchases can also create more value than weak originations. The problem begins when contraction persists long enough to reduce recurring earnings and dividend capacity.

What does the gap between non-accrual cost and fair value show?

It shows how deeply problem loans have already been marked down. A large gap can mean the BDC has recognized substantial expected loss, but it can also signal weak recovery prospects. Investors should watch both percentages and the direction of change.

Why do Fitch and Proskauer report different private-credit default rates?

They track different loan universes and apply different methodologies. Private credit has no single comprehensive market tape, so sector default statistics are not directly interchangeable.

What should investors watch next?

Focus on recurring NII rather than fee-inflated yield, portfolio growth after repayments, new non-accruals, realized restructuring losses, NAV movement, discounted repurchases, and whether regular dividends are supported by current-period earnings rather than reserves.

The Bottom Line

This earnings cycle has not broken the BDC income story.

It has made the replacement problem impossible to ignore.

ARCC’s scale absorbed a quarter in which exits exceeded commitments, but non-accruals increased and core earnings slipped just below the dividend.

HTGC covered the payout and grew NAV, but record repayments converted part of tomorrow’s interest income into today’s earnings.

GBDC protected its dividend and bought discounted shares, but the portfolio contracted and credit losses remained visible.

CSWC found abundant new assets, yet its recurring dividend coverage tightened and NAV moved lower.

OBDC then added another version of the same tension: better coverage, lower leverage, and a smaller portfolio with softer NAV. MAIN showed the opposite combination, with solid NII and stronger NAV arriving together.

At the same time, the AI infrastructure boom is creating enormous demand for private capital. That is the opportunity sitting beside the problem. Data centers, power systems, equipment, leases, and service companies need financing at a scale that will pull banks, insurers, infrastructure funds, asset-backed lenders, private-credit funds, and some BDCs deeper into the physical economy.

But capital demand does not guarantee credit quality.

High base rates still support asset yields. They also keep pressure on borrowers, refinancing paths, construction economics, and BDC funding costs. Strong borrowers repay. Weak borrowers restructure. New thematic opportunities invite capital before their full risks are visible.

The lender must replace the first without inheriting more of the second.

That is the real contest now.

Not who reports the highest quarterly yield.

Who can rebuild the earning portfolio without spending the balance sheet—or the underwriting standard—to do it.

Read Next

Start with The BDC Investing Guide, the permanent BDC sector map, and Are BDCs a Good Investment?.

For the company-level mechanics, read Ares Capital, Hercules Capital, the HTGC Q2 deep dive, Blue Owl Capital Corporation, the OBDC Q2 deep dive, Golub Capital BDC, and Capital Southwest.

For the AI financing system, read How AI Infrastructure Gets Financed, Who Finances AI Data Centers?, Asset-Backed Finance and AI Infrastructure, and BDC Weekly: How Meta’s AI Data Centers Are Pulling Private Credit Into Infrastructure.

For the rate and credit framework, read Floating-Rate Loans Explained, The Private-Credit Refinancing Wall, NII Coverage Ratio Explained, What Is NAV?, What Are Non-Accruals?, and PIK Income Explained.

For the next company deep dives, follow FS KKR Capital, Blackstone Secured Lending, and Main Street Capital.

Follow the BDC Weekly hub for the next earnings and private-credit update.

Source Notes

This issue uses The Drift’s August 4, 2026 BDC Weekly research export, covering SEC filings, company and news events, and macro signals from the prior seven days. Material company-specific figures were checked against issuer earnings releases and SEC filings. The August 8 refresh incorporates official OBDC, FSK, BXSL, and MAIN earnings materials released after the issue was prepared.

Major external sources include Ares Capital’s second-quarter release, Hercules Capital’s second-quarter release and filings, Golub Capital BDC’s fiscal third-quarter release, Capital Southwest’s fiscal first-quarter release, Blue Owl Capital Corporation’s Q2 2026 results, FS KKR Capital’s Q2 2026 results, Blackstone Secured Lending’s Q2 2026 earnings materials, Main Street Capital’s Q2 2026 results, the Federal Reserve’s July 29 policy statement, Goldman Sachs Research on private-market data-center financing, International Energy Agency analysis of data-center electricity demand, Brookfield and Bloom Energy’s AI power-financing partnership, Fitch Ratings’ second-quarter private-credit monitor, and Proskauer’s Q2 2026 Private Credit Default Index.

Source links:

• Ares Capital second-quarter 2026 results
• Hercules Capital second-quarter 2026 results
• Blue Owl Capital Corporation second-quarter 2026 results
• FS KKR Capital second-quarter 2026 results
• Blackstone Secured Lending second-quarter 2026 earnings
• Main Street Capital second-quarter 2026 results
• Golub Capital BDC fiscal third-quarter 2026 results
• Capital Southwest fiscal first-quarter 2027 results
• Federal Reserve July 29, 2026 policy statement
• Goldman Sachs Research: Private markets and data-center financing
• International Energy Agency: Energy demand from AI
• Brookfield and Bloom Energy expand AI infrastructure financing partnership
• Fitch: U.S. Private Credit Default Rate Reaches New High in 2Q26
• Proskauer Q2 2026 Private Credit Default Index

This article is market education and analysis, not individualized investment advice.